improvements in hazard and life cycle impact …...that made this work very exciting. i am also...
TRANSCRIPT
IMPROVEMENTS IN HAZARD AND LIFE CYCLE IMPACT
ASSESSMENT METHOD FOR METALS IN FRESHWATERS -
ADDRESSING ISSUES OF METAL SPECIATION, FATE,
EXPOSURE AND ECOTOXICITY
by
Nilima Gandhi
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy
Department of Chemical Engineering and Applied Chemistry
University of Toronto
© Copyright by Nilima Gandhi, 2011
ii
Improvements in Hazard and Life Cycle Impact Assessment Method for
Metals in Freshwaters - Addressing Issues of Metal Speciation, Fate,
Exposure and Ecotoxicity
Nilima Gandhi
Doctor of Philosophy – 2011
Division of Environmental Studies
Department of Chemical Engineering and Applied Chemistry
University of Toronto
ABSTRACT
Methods of chemical hazard ranking and toxic impact assessment estimate fate and toxicity
assuming the chemical exists in dissolved and particulate phases and, for metals, that all
dissolved species are equally bioavailable. This treatment of metals, similar to organic
chemicals, introduced a significant error in their estimates of hazard ranking since metal
bioavailability and ecotoxicity are related to truly dissolved phase and specifically free metal
ion within it. My thesis addressed this concern by developing a new method that introduced
Bioavailability Factor (BF) to the calculation of Comparative Toxicity Potentials (CTPs) for
hazard ranking of chemicals; also known as Characterization Factors for use in Life Cycle
Impact Assessment (LCIA). First, the metal speciation/complexation was incorporated into
fate calculations by loosely coupling commercial geochemical metal speciation model,
WHAM, with a multimedia fate model, USEtoxTM
, which is originally designed to calculate
CTPs for organic chemicals. Second, Biotic Ligand Model (BLM) was used to calculate the
bioavailability-corrected adverse toxic effects of metals.
iii
This new method was applied to assess the implications of choosing environmental
characteristics, notably freshwater chemistry, by calculating BFs and CTPs of several
cationic metals (e.g., Cd, Cu, Co, Pb, Ni and Zn) using 12 European, 24 Canadian
ecoregions, several distinct freshwater-types selected from large river and lake systems
world-wide. The newly estimated metal CTPs (i.e., ecotoxicity potentials) are up to ~1000
times lower than previous values used in LCIA. Notably the model results showed that the
absolute values of CTPs, and their relative ranking amongst chemicals, are a product of the
characteristics of a receiving environment. Hence it is crucial to select a generic freshwater
archetype on which this analysis should be based. Finally, the new model framework was
extended to apply within the Unit World Model (UWM) framework to estimate critical loads
(CLs) of cationic metals to surface aquatic systems.
iv
ACKNOWLEDGEMENTS
I would like to take this opportunity to thank many individuals who helped me in getting my
PhD. First, I am very thankful to Prof. Miriam Diamond for her inspiration and enthusiasm,
her continuous support and guidance, and her encouragement to explore novel research ideas
that made this work very exciting. I am also grateful to her for enabling, facilitating and
encouraging multi-disciplinary discussions and collaborations within the members of
Diamond lab group from which I have learnt a lot.
I acknowledge funding from UNEP SETAC Life Cycle Initiative, International Council of
Mining and Metals (ICMM), Iron Ore Company of Canada (IOC), and Xstrata, for this
project. I am thankful to Natural Sciences and Engineering Research Council of Canada
(NSERC) for providing me the Alexander Graham Bell Canada Graduate Scholarship (CGS)
and Collaborative Research and Development (CRD) grant to Prof. Miriam Diamond to carry
out this research. This work was also supported in part by the Society of Environmental
Toxicology and Chemistry (SETAC) Chris Lee award and sponsored by the International
Copper Association (ICA) that I received for this research in November 2010.
I would also like to thank Mark Huijbregts and Dik van De Meet (Radboud University,
Nijmegen, Netherlands), Willie J. G. M. Peijnenburg, Martina Vijver, Jeroen Guinée and
Arjan De Koning (Leiden University, Leiden, Netherlands) for providing me technical
guidance and comments on my research progress. I also thank Bill Adams (Rio Tinto), John
Atherton (ICMM), Michael Hauschild (University of Denmark), and Kevin Farley
(Manhattan College, NY) for technically guiding this research project. Special thanks to
Jasim Chowdhury (International Zinc Association), Tom Brock (Cobalt Development
Institute), Bill Stubblefield (Oregon State University), Andy Bush (International Lead
Association), Delphine Haesaerts and Frank Van Assche (International Zinc Association,
Europe) for sharing their metal toxicological data. The United Nations Global Environment
Monitoring System (GEMS) Water Programme provided environmental water quality data
for global freshwaters. Pradeep Goel, Jocelyn Heneberry and Satyendra Bhavsar (Ontario
Ministry of the Environment) provided/assisted in collecting water chemistries for Canadian
systems.
v
I am grateful to Profs. Charles Jia and George Arhonditsis for providing technical guidance
through my reading and defence committee meetings. I like to thank Joan Chen, Gorrette
Silva, Julie Mendonca, Pauline Martini, Leticia Gutierrez, and Arlene Smith for assisting in
administrative aspects of the department. I am also thankful to Mircea Pilaf, Marika Maslej
and Bruce Huang from the department of Geography; Laurane Harding, Pavel Pripa and
Mona El-Haddad from the Centre for the Environment for their assistance and guidance in
various administrative tasks.
All my colleagues in the lab have been very supportive and helpful. They shared their
knowledge and experiences which often resulted in an excellent collaborative research. I
extend my special thanks to many current and past members of my research group, especially
to Jennifer Sawyer, Susan Csiszar, Sarah Gewurtz, and Erin Hodge for helpful
interdisciplinary discussions.
Finally, I offer my special thanks to my husband, family and friends for providing support
and continuous encouragement throughout the completion of my PhD thesis.
vi
TABLE OF CONTENTS
ABSTRACT ............................................................................................................................ II
ACKNOWLEDGEMENTS ................................................................................................. IV
TABLE OF CONTENTS ..................................................................................................... VI
LIST OF TABLES ................................................................................................................. X
LIST OF FIGURES ........................................................................................................... XIII
1. ENVIRONMENTAL ASSESSMENT OF METALS: AN INTRODUCTION .................................. 1
1.1 Metals in the Environment .....................................................................................1
1.2 Metal Emissions and Behaviour in the Environment ............................................1
1.2.1 Speciation and Fate ...................................................................................2
1.2.2 Exposure and Toxicity ..............................................................................3
1.3 Environmental Regulations in Canada...................................................................4
1.4 Environmental Assessments ..................................................................................5
1.4.1 Hazard Assessment ...................................................................................5
1.4.2 Risk Assessment .......................................................................................6
1.4.3 Life Cycle Assessment ..............................................................................7
1.5 Modelling Metal Movement ..................................................................................7
1.5.1 Modelling Challenges ...............................................................................8
1.5.2 Research Developments ..........................................................................11
1.6 Research Goals.....................................................................................................15
1.7 References ............................................................................................................17
2. NEW METHOD FOR CALCULATING COMPARATIVE TOXICITY POTENTIAL OF CATIONIC
METALS IN FRESHWATER: APPLICATION TO COPPER, NICKEL, AND ZINC ............................ 25
2.1 Abstract ................................................................................................................25
2.2 Introduction ..........................................................................................................26
2.3 Methods................................................................................................................28
2.3.1 Current Practice ......................................................................................28
2.3.2 Proposed Framework ..............................................................................29
2.4 Model Selection and Parameterization ................................................................31
2.4.1 Fate Model ..............................................................................................31
2.4.2 Geochemical Speciation-Bioavailability Model .....................................32
vii
2.4.3 Aquatic Ecotoxicity Model .....................................................................33
2.4.4 Overall Model Structure and Parameterization.......................................34
2.5 Results and Discussion ........................................................................................38
2.5.1 Kd Values ................................................................................................38
2.5.2 Fate Factors .............................................................................................40
2.5.3 Bioavailability Factors ............................................................................41
2.5.4 Effect Factors ..........................................................................................43
2.5.5 Comparative Toxicity Potentials.............................................................43
2.6 Practical Implications...........................................................................................45
2.7 References ............................................................................................................46
3. IMPLICATIONS OF GEOGRAPHIC VARIABILITY ON COMPARATIVE TOXICITY POTENTIALS
OF CU, NI AND ZN IN FRESHWATERS OF CANADIAN ECOREGIONS ....................................... 52
3.1 Abstract ................................................................................................................52
3.2 Introduction ..........................................................................................................53
3.3 Methods................................................................................................................55
3.3.1 Modelling Framework ............................................................................55
3.4 Model Selection and Parameterization ................................................................58
3.4.1 Fate ..........................................................................................................58
3.4.2 Speciation/Complexation ........................................................................60
3.4.3 Ecotoxicity ..............................................................................................60
3.4.4 Model Parameters ...................................................................................60
3.5 Results and Discussion ........................................................................................63
3.5.1 Metal Partitioning (Kd) ...........................................................................63
3.5.2 Fate ..........................................................................................................65
3.5.3 Bioavailability .........................................................................................66
3.5.4 Ecotoxicity ..............................................................................................66
3.5.5 Comparative Toxicity Potential ..............................................................68
3.6 Sensitivity Analysis .............................................................................................69
3.6.1 Freshwater Residence Time ....................................................................69
3.6.2 Background Metal Concentrations .........................................................70
3.6.3 Total Suspended Sediment Concentrations ............................................74
viii
3.7 Summary ..............................................................................................................76
3.8 References ............................................................................................................77
4. IMPLICATIONS OF CONSIDERING METAL BIOAVAILABILITY IN ESTIMATES OF
FRESHWATER ECOTOXICITY: EXAMINATION OF TWO CASE STUDIES ..................................... 82
4.1 Abstract ................................................................................................................82
4.2 Introduction ..........................................................................................................84
4.3 Methods................................................................................................................86
4.3.1 Case Studies ............................................................................................86
4.3.2 Model Applications .................................................................................89
4.3.3 Scope and Assumptions ..........................................................................96
4.4 Results and Discussion ........................................................................................97
4.4.1 Comparison of Metal CFs .......................................................................97
4.4.2 Freshwater Ecotoxicity of Case Studies ...............................................103
4.4.3 Comparisons with Previous Case Study Results ..................................110
4.4.4 Improvements in USEtox(new) Approach............................................111
4.5 Conclusions ........................................................................................................112
4.6 Practical Implications.........................................................................................113
4.7 References ..........................................................................................................113
5. CRITICAL LOAD ANALYSIS IN HAZARD ASSESSMENT OF METALS USING A UNIT WORLD
MODEL............................................................................................................................... 118
5.1 Abstract ..............................................................................................................118
5.2 Introduction ........................................................................................................119
5.3 Methods..............................................................................................................122
5.3.1 Modelling Approach .............................................................................122
5.3.2 Model Parameterization ........................................................................126
5.4 Results and Discussion ......................................................................................130
5.4.1 Model Evaluation ..................................................................................130
5.4.2 Speciation/Complexation Results .........................................................130
5.4.3 Fate-Transport Results ..........................................................................134
5.4.4 Aquatic Ecotoxicity ..............................................................................137
5.4.5 Critical Load .........................................................................................139
ix
5.5 Conclusions ........................................................................................................142
5.6 References ..........................................................................................................143
6. CONCLUSIONS AND RECOMMENDATIONS ................................................................... 148
6.1 Scientific Significance of my Research .............................................................148
6.2 Major Findings ...................................................................................................151
6.3 Lessons Learned.................................................................................................154
6.4 Recommendations for Future Work...................................................................160
6.5 References ..........................................................................................................164
APPENDIX – A ................................................................................................................... 170
1. THE CLEARWATER CONSENSUS: THE ESTIMATION OF METAL HAZARD IN FRESHWATER
170
1.1 Abstract ..............................................................................................................170
1.2 Background, Aim and Scope .............................................................................171
1.3 Conclusions and Recommendations ..................................................................173
1.4 References ..........................................................................................................177
x
LIST OF TABLES
Table 1.1: Summary of major differences in behaviour of organic chemicals compared to
metals and inorganic metal compounds in humans. .......................................... 10
Table 2.1: Freshwater chemistry data used in the geochemical model, WHAM 6.0, to
estimate Bioavailability Factors (BFs) for Cu, Ni, and Zn in 12 EU water-types
(background metal concentrations of 1 µg/L for Cu and Ni and 10 µg/L for Zn
were used for all water-types). ........................................................................... 36
Table 2.2: Values of conditional binding constants (LogKBL) of binding metals and other
competing cations with biotic ligand for chronic BLMs used in this model
application. ......................................................................................................... 37
Table 2.3: Estimated Bioavailability Factors (BFs, dimensionless), Fate Factors (FFs, days),
Effect Factors (EFs, m3/kg) and Comparative Ecotoxicity Potentials (CTPs,
day.m3/kg) for Cu, Ni and Zn for the 12 EU water types listed in Table 3.1.
Coefficients of variance (CV) are reported for each modelled parameter among
water-types and metals. Note that FF represents residence time of total metal in
freshwater after its unit (1 kg/day) emission to freshwater compartment. ........ 45
Table 3.1: Summary of model parameters used to calculate Fate Factors (or residence times)
of metals in freshwater compartment of 24 ecoregions of Canada. ................... 59
Table 3.2: Freshwater chemistry data used in the geochemical model, WHAM 6.0, to
estimate Bioavailability Factors (BFs) for Cu, Ni and Zn in 24 Canadian
ecoregions. A background metal concentration of 1 µg/L for Cu and Ni, and 10
µg/L for Zn were used for all water-types. ........................................................ 62
Table 3.3: Measurements of metal background concentrations and total suspend sediment
concentrations in freshwaters across Canada. This information was used to
estimate ranges in parameter values to conduct the sensitivity analysis of model
results for bioavailability, fate and ecotoxicity potentials of metals in
freshwaters of 24 Canadian ecoregions. ............................................................ 71
xi
Table 4.1: LCI data for 1 kg of copper pipe with emission estimates by the processing stage
(source: Gloria et al. 2006). ............................................................................... 87
Table 4.2: LCI data for zinc gutter system reported by Gloria et al. (2006; original source:
Eggels et al. 2000).............................................................................................. 88
Table 4.3: Comparisons of calculation methods and model parameters of four LCIA models
used in the freshwater ecotoxicity assessment of case studies. ......................... 93
Table 4.4: Default model parameter values required for the calculation of metal CFs as
reported in the original USES-LCA 1.0, USES-LCA 2.0 and USEtoxTM
(see
Table 4.3). .......................................................................................................... 94
Table 4.5: Ambient chemistry for freshwater archetypes used to calculate CFs of metals
using the new framework proposed by Gandhi et al. (2010). ............................ 95
Table 4.6: Estimated metal bioavailable fractions (BFs, dimensionless), LogKd (L/kg) and
average chronic toxicity (HC50; mg/L) values corrected for the speciation of
metals in various freshwater archetypes used in the analysis of USEtox(new)
method as discussed in Table 4.3. ...................................................................... 95
Table 4.7: Comparison of previously reported (USES-LCA 1.0; Huijbregts et al. 2000) and a
range of archetype-specific metal CFs (kg eq. 1,4-DCB) calculated using the
method of Gandhi et al. (2010, 2011) for use in metal LCIA. ........................... 98
Table 4.8: Comparison of metal CFs (kg eq. 1,4-DCB) estimated for freshwater ecotoxicity
using four LCIA models, time-scales of infinity and 100 years of environmental
impacts after metal emissions, and for seven freshwater types as mentioned in
the text. ............................................................................................................. 101
Table 4.9: The relative importance of metals (in an increasing order) based on the numerical
ranking of estimated CFs towards the freshwater ecotoxicity in LCIA. .......... 101
Table 4.10: Relative ranking in the order of low to high cotoxicity potential for organic
chemicals and metals based on the CFs calculated in each model. Note that CFs
for the organics used in the relative ranking of USEtox(new) approach are from
USEtoxTM
. ........................................................................................................ 102
xii
Table 4.11: Percentage contribution of metal emissions to air and water towards the total
freshwater ecotoxicity estimated for Cu pipe case study. Note that six metals
for which new CFs are currently available were considered in this analysis. . 105
Table 4.12: Literature derived ranges and geometric averages of measured chronic toxicity
test data, expressed as total dissolved concentration, for metals considered in the
case studies....................................................................................................... 112
Table 5.1: Parameters values selected for the Unit Lake in the metal fate calculations using
TRANSPEC, in comparison to the values measured for the Ross Lake (MB,
Canada), Kelly Lake (ON, Canada), and Lake Tantaré (QC, Canada). Data for
Ross Lake were obtained from HBMS (unpublished data); for Kelly Lake from
field study and Lock (unpublished data); and for Lake Tantaré from Alfaro-De
la Torre and Tessier (2002) and Alfaro-De la Torre (unpublished data). ........ 127
Table 5.2: System-specific chemistry parameters for Lakes Ross, Kelley and Tantaré used in
WHAM for speciation calculations to assess the effects of chemistry on fate and
toxicity using the Unit World Model. Data for Ross Lake (MB, Canada) were
obtained from HBMS (unpublished data); for Kelly Lake (ON, Canada) from
field study and Lock (unpublished data); and for Lake Tantaré (QC, Canada)
from Alfaro-De la Torre and Tessier (2002) and Alfaro-De la Torre
(unpublished data). Background metal concentrations of 0.1 µg/L for Cd, 1
µg/L for Cu, Ni and Pb, and 10 µg/L for Zn were used in WHAM calculations.
.......................................................................................................................... 128
xiii
LIST OF FIGURES
Figure 2.1: Model results for Cu, Ni, and Zn using the chemistry of 12 EU water-types
described in Table 2.1. (a) WHAM estimated metal partition coefficients, LogKd
(L/kg), used in fate model, (b) WHAM estimated BFs (Bioavailability Factors;
dimensionless; calculated as a fraction of total metal in the bioavailable form), (c)
freshwater FFs (Fate Factors, days) for emissions in freshwater compartment
calculated using the default parameter values of USEtoxTM
model and WHAM
estimated values of Kd for each water-type, (d) BLM estimated metal EFs (Effect
Factors; m3/kg) corrected for chemistry of each water-type, and (e) comparison of
metal CTPs (Comparative Toxicity Potentials; day.m3/kg) for water-types and
those calculated using the default parameters for metal assessment in USEtoxTM
(●). Note FFs are for total metal and represent the residence time of metals in
freshwater due to a unit emission. .......................................................................... 39
Figure 2.2: Estimated freshwater Fate Factors (FFs, days) of Cu, Ni, and Zn for their unit
emissions into the freshwater compartment using the default setting of USEtoxTM
model and WHAM estimated values of Kd for the 12 EU water-types (see Table
2.1). Here FFs represent residence times for total metals in freshwater after
emission. ................................................................................................................. 41
Figure 2.3: Values of Bioavailability Factors (BFs; dimensionless) calculated as the fraction
of total metal in the bioavailable form for Cu, Ni, and Zn using the chemistries of
selected EU water-types shown as the function of LogKd. .................................... 42
Figure 2.4: Comparison of metal ranking according to values of Comparative Toxicity
Potentials (CTPs; day.m3/kg) calculated for the 12 EU water-types. The lowest
value of CTP among three metals within one water-type represents the lowest
concern (or relative hazard) and vice versa. The relative ranking is displayed as
hatched for the lowest, brick for medium and dotted pattern for the highest. ........ 44
Figure 3.1: Model results for Cu, Ni, and Zn using the chemistry and landscape
characteristics of 24 Canadian freshwater-types and the overall Canadian water-
xiv
type described in Table 3.2. (a) WHAM estimated metal partition coefficients, Kd
(L/kg), used in fate calculations, and (b) WHAM estimated BF (Bioavailability
Factors; dimensionless) calculated as a fraction of total metal that is bioavailable
and is assumed to be within the truly dissolved fraction of total metal (c) FFs (Fate
Factors; days) for unit emission of each metal in freshwater compartment using the
fate parameter values of Canadian ecoregions and WHAM estimated Kd for each
ecoregion freshwater-type (d) BLM estimated EFs (Effect Factors; m3/kg) that
were corrected for chemistry of freshwater-type in each ecoregion, and (e) CTP
(Comparative Toxicity Potential; day.m3/kg), where the variability in values of
CTP reflects variability in chemistry of freshwater-types and landscape properties
of Canadian ecoregions. ......................................................................................... 64
Figure 3.2: Model estimated ranges in (a) Bioavailability Factors (BF; dimensionless), (b)
Fate Factors (FFs; days) for unit emission of metals in freshwater compartment, (c)
Effect Factors (EFs; m3/kg) that represent average potential ecotoxicity, and (d)
Comparative Toxicity Potentials (CTPs; day.m3/kg) for Cu, Ni, and Zn calculated
to examine variability in chemistry of freshwater-types and landscape properties of
Canadian ecoregions............................................................................................... 67
Figure 3.3: Sensitivity of metal Fate Factors (FFs; days) to landscape properties of
freshwater compartments of Canadian ecoregions. ................................................ 70
Figure 3.4: Sensitivity of modelled metal BFs (bioavailability), FFs (fate) and CTPs
(Comparative Toxicity Potentials) to background concentrations of Cu, Ni and Zn
in freshwaters of Canadian ecoregions. The numbers on x-axis represent Canadian
ecoregions as listed in Table 3.1. The results on y-axis are displayed as percentage
changes from the respective base case for each modelled parameter and ecoregion.
................................................................................................................................ 72
Figure 3.5: Percentage of total Cu in dissolved, colloidal and particulate phases estimated for
Cu background concentrations of (a) 0.1 µg/L, (b) 1 µg/L (base case scenario), and
(c) 10 µg/L. ............................................................................................................. 73
xv
Figure 3.6: Sensitivity of modelled metal BFs (bioavailability), FFs (fate) and CTPs
(Comparative Toxicity Potentials) to total suspended sediment (TSS)
concentrations in freshwaters of Canadian ecoregions. The numbers on x-axis
represent Canadian ecoregions as listed in Table 3.2. The results on y-axis are
displayed as percentage changes from the respective base case for each modelled
parameter and ecoregion. ....................................................................................... 75
Figure 4.1: Relative contribution of metals towards total freshwater ecotoxicity potential
(toxicity impact indicator) based on the model-specific CFs (1,4-DCB eq.) if an
unit emission of each of these metals occurs to the freshwater environment. ..... 103
Figure 4.2: Total estimated metal emission (kg) that will eventually end up in the freshwater
compartment due to the release of metals to air and water during the processing of
Cu pipe considered in the case study.................................................................... 104
Figure 4.3: LCIA results presented as the total freshwater ecotoxicity score of metals
estimated for the Cu case study. Here the ecotoxicity was estimated for total
emission of metals to freshwaters due to release of metals in both air and water
compartments during the processing of Cu pipe (see Table 4.1). ........................ 106
Figure 4.4: Relative contribution of each metal emitted in (a) air and (b) water as listed in
LCI towards the total freshwater ecotoxicity score for the LCIA of Cu case study.
.............................................................................................................................. 107
Figure 4.5: Total estimated metal emission (kg) that will reach the freshwater compartment
due to the release of metals to air and water in the case study of Zn gutter system.
.............................................................................................................................. 108
Figure 4.6: LCIA results presented as the total freshwater ecotoxicity score of metals
estimated for the Zn case study. Here the ecotoxicity was estimated for total
emission of metals to freshwaters due to release of metals in both air and water
compartments as listed in LCI data for Zn gutter system (see Table 4.2). ........... 109
Figure 4.7: Contribution of each metal listed in LCI of Zn gutter system towards the total
freshwater ecotoxicity score in the analysis of its LCIA...................................... 110
xvi
Figure 5.1: WHAM estimated (a) phase distribution, (b) partition coefficients, Kd, between
particulate and total dissolved phases, (c) metal speciation in the total dissolved
phase, and (d) percentage of metal in the free ion form relative to the total metal
(sum of total dissolved and particulate phases) for five metals (Cd, Cu, Ni, Pb, and
Zn) and for the selected chemistries of eutrophic, mesotrophic and oligotrophic
systems in the UWM analysis. ............................................................................. 133
Figure 5.2: Estimated fate and transport of total metals in shallow unit world lakes with
physical properties described in Table 5.1 and chemistries of the eutrophic,
mesotrophic and oligotrophic systems (Table 5.2). Transport rates and
concentrations are in g/day and ng/L, respectively. Unit loadings were considered
with the concentrations being close to background values for each metal. ......... 136
Figure 5.3: Comparison of metal residence time (days) estimated using the Unit Lake set up
in TRANSPEC model. ......................................................................................... 137
Figure 5.4: “Water critical concentration (W-LC50)” and “Inflow critical concentration (I-
LC50)” (mg/L) for (a) Cd, (b) Cu, and (c) Zn estimated for three levels of aquatic
organisms using BLM and the characteristics of eutrophic (Ross Lake),
mesotrophic (Kelly Lake) and oligotrophic (Lake Tantaré) systems. .................. 139
Figure 5.5: Ratios of I-LC50 (mg/L) to W-LC50 (mg/L) for Cd, Cu and Zn that would protect
Fathead minnow in the selected aquatic systems. ................................................ 140
Figure 5.6: The effect of pH on values of in-lake and inflow LC50 (W- and I-LC50) for Zn for
the three receptors considered in BLM, Rainbow trout, Fathead minnow and D.
magna. Here water chemistry parameters are representative of eutrophic system,
Ross Lake. ............................................................................................................ 141
Figure A.1: Fractions of total chemical. For metals, the truly dissolved fraction, which is
assumed to be bioavailable, is within the total dissolved fraction. In turn, the
fraction of free metal ion (e.g., Me+2
) is within the truly dissolved fraction. ...... 172
1
1. Environmental Assessment of Metals: An Introduction
1.1 Metals in the Environment
Metals are naturally present in varying amounts in all environmental media. Several metals,
such as copper, zinc, lead, mercury, nickel, cobalt, and chromium, are common trace
constituents in the earth’s crust. Approximately 8000 years ago in western Asia, almost pure
deposits of gold and copper were first discovered. Shortly thereafter, early societies found
that metals possess both malleability and strength. Since then, humans have extracted these
and other non-ferrous metals from the earth's crust and have used them to serve many
different purposes. A critical consequence of this widespread use, which accelerated many-
fold after the industrial revolution, has been anthropogenic changes to the abundance and
cycling of metals in the environment.
Many metals play an essential role in plant and animal physiology as micronutrients. For
example, the enzymes that synthesize DNA and RNA contain zinc ions, and cobalt is an
integral part of coenzyme B12 and vitamin B12. Metals that are micronutrients behave
hormetically, with adverse effects occurring at very low and high exposures. However,
nonessential metals such as lead and mercury play no beneficial role in plant or animal
physiology and as such these metals can induce toxicity at low doses. Metals exert toxicity
through a multiplicity of physiological pathways. To make matters more complicated, many
metals exist in multiple oxidation states, which control their toxicity.
1.2 Metal Emissions and Behaviour in the Environment
Although natural environmental processes such as the weathering of rocks are responsible for
metal cycling and their redistribution in the environment, anthropogenic activity has
significantly changed their abundance. Emissions from metal mining, smelting and refining,
power generation and solid-waste incinerators, manufacturing, and transportation sectors are
major sources of metals in the environment.
2
It is common practice in the mining and mineral processing industries to discharge treated
waste water with elevated metal concentrations (relative to background) into surface waters.
Over the past decade and more, these industries have significantly reduced metal discharges
to the environment by orders-of-magnitude below historical levels. Older facilities not only
contend with advancing technologies to reduce today’s emissions, but they must also manage
historically discharged materials.
After release, metals distribute among environmental media such as water, soil, and
sediments. Toxicological impacts associated with the excessive release of trace metals into
the environment can arise because metals neither biodegrade nor are they permanently
eliminated from systems. Each metal in each environmental medium has its characteristic
geochemical speciation, depending on chemistry, that controls the metal’s bioavailability,
fate and toxicity. The speciation, bioavailability and fate of metals are integral parts of
assessing the overall risks that metals pose to biota.
1.2.1 Speciation and Fate
In aquatic systems, trace metals can form complexes with ligands from organic matter,
and/or may be sorbed to suspended particulate matter (SPM) that can be transported to
sediments (e.g., Carignan and Tessier 1985, Petersen et al. 1995). Metal speciation, and
hence their bioavailability and mobility, are influenced by factors such as pH, redox
potential, formation of organic complexes, and salinity (Forstner et al. 1986, Achterberg et al.
1997). Kinetically controlled microbial reactions may also affect the metal mobility and
toxicity by changing the minerals to which trace elements are bound (e.g., Petersen et al.
1995, Smith and Jaffe 1998).
Metals deposited in sediments participate in a variety of processes, including microbial
reactions, redox transformations, adsorption-desorption, and the precipitation and dissolution
of minerals. In general, pH and redox status are among the most important factors that affect
the mobility of sediment-bound metals (Patrick and Verloo 1998, Wen and Allen 1999).
Acidic conditions increase metal dissolution (Forstner et al. 1986). Moderately reducing
conditions that occur in the transition from oxic to anoxic conditions, increase the mobility of
3
Fe and Mn due to the dissolution of their oxide forms (Wen and Allen 1999) and metals that
co-precipitate with those oxides. Under strongly reducing conditions, metals such as Zn, Pb,
Cu, and Cd (Carignan and Tessier 1985, Forstner et al. 1986, Smith and Jaffe 1998) are
immobilized due to precipitation as metal sulphides (Hamilton-Taylor et al. 1996, Achterberg
et al. 1997). Conversely, these metals can be released from their metal sulphides as
conditions shift from reducing to oxidizing as oxygen is introduced (e.g., Slotton and Reuter
1995). In weakly buffered sediments, an increase in redox potential may decrease pH, which
may increase the mobility of most metals (Carvalho et al. 1998). Thus, metals deposited in
sediments are not necessarily permanently immobilized, but rather they may be remobilized
through diagenetic processes (Carignan and Tessier 1985, Petersen et al. 1995) involving
biological and chemical agents (Forstner et al. 1986) and by physical movement (Diamond
1995).
In the terrestrial environment, soil properties such as soil organic matter, pH, cation exchange
capacity and mineralogy affect metal solubility and bioavailability (Sauvé et al. 1998).
Aging and weathering processes exchange metals between tightly bound and exchangeable
phases. Metal in the soluble phase of soil can be taken up by plants where it may be stored
and then returned to soil upon plant death. Metals can be transported to nearby surface
aquatic systems through runoff and lost to groundwater through leaching. Surface soils could
be a major source of metals to aquatic systems.
1.2.2 Exposure and Toxicity
Exposure generally describes the potential or actual contact or co-occurrence of a stressor
with a receptor. Exposure analysis includes the study of a stressor’s sources, its distribution
in the environment, and the extent and pattern of contact or co-occurrence to produce
identifiable or measurable effect(s) to the ecological receptor(s) of concern.
As mentioned earlier, metals can cause adverse health effects in humans and ecological
receptors. At least five metals are known carcinogens, and several other effects of metals are
also well documented, including effects on the neurological, cardiovascular, haematological,
gastrointestinal, musculoskeletal, immunological, and epidermal systems. Diversity in
4
observed toxicities of different metals likely reflects the variety of biochemical mechanisms
by which they exert their effects and variability in their toxicokinetic properties.
1.3 Environmental Regulations in Canada
Concerns over the release of metals to the environment from human activities have existed
for decades. For example, emissions of lead, along with particulate matter (PM) and sulphur
dioxide (SO2), from a Canadian smelter were the basis for one of the earliest trans-boundary
disputes (Canada-U.S. International Joint Commissions, 1920's). When the Canadian
Department of the Environment was created in 1971, heavy metals were addressed through
federal environmental legislation. Various metals are target substances in several domestic
and international agreements and plans. There are 1021 inorganic substances on the
Domestic Substances List (DSL) compiled under the Canadian Environmental Protection Act
(CEPA). The following metals have now been assessed for toxicity and added to Schedule 1,
List of Toxic Substances of the Canadian Environmental Protection Act (CEPA 1999): (1)
lead, (2) mercury, (3) inorganic cadmium compounds, (4) chromium (VI) compounds, (5)
PM containing metals that is released in emissions from copper smelters and/or refineries,
and (6) PM containing metals that is released in emissions from zinc plants. CEPA 1999 is
an Act respecting pollution prevention and the protection of the environment and human
health.
Metal releases to the environment continued to receive regulatory attention over subsequent
years. In addition to the early Canadian regulations on metals emissions mentioned above,
other CEPA (1999) regulations limiting metal emissions have since been promulgated. Some
metals are now being reported annually under the CEPA 1999 National Pollutant Release
Inventory (NPRI). The Canada-U.S. Great Lakes Water Quality Agreement (GLWQA,
1978) identified and set concentration limits in water for 10 metals and metalloids identified
as persistent toxic substances of concern in the Great Lakes: arsenic, cadmium, chromium,
copper, iron, lead, mercury, nickel, selenium and zinc. In 2002, Environment Canada
introduced the Metal Mining Effluent Regulations (MMER) under section 36 of the Fisheries
Act to regulate the deposit of mine tailings and other effluents produced during mining
5
operations into natural fish bearing waters. These regulations are applied to both new and
existing mines and are among the most comprehensive and stringent national standards for
mining effluents in the world.
Currently Canada is working on Chemicals Management Plan (CMP) that will improve the
degree of protection against hazardous chemicals including many metals. The Government
of Canada's Toxic Substances Management Policy puts forward a precautionary and
preventive approach to deal with substances that enter the environment and could harm the
environment and/or human health. It provides a framework for making science-based
decisions on the effective management of toxic substances.
Similarly there are numerous environmental regulations for metal emissions that are in place
by federal (e.g., U.S. Environmental Protection Agency; USEPA) and state agencies in the
United States. Within European jurisdiction, REACH (Regulation on Registration,
Evaluation, Authorisation and Restriction of Chemicals, 2007) streamlines and improves the
former legislative framework on hazardous chemicals of the European Union (EU). The
major aim of REACH is to ensure a high level of protection of human health and the
environment from the risks that are posed by anthropogenic emissions of chemicals.
1.4 Environmental Assessments
Environmental assessment of potential harm posed by the release of chemicals is currently
completed using one of the following methods: (1) hazard assessment (HA); (2) risk
assessment (RA); and (3) life cycle assessment (LCA).
1.4.1 Hazard Assessment
Hazard identification is defined as a measure of the intrinsic capacity of a substance to cause
an adverse response in a living organism (OECD 1995). Hazard assessment is differentiated
from risk assessment by not considering exposure or the probability of a hazardous outcome,
but rather it deals with the inherent properties of substances.
6
Hazard information has several uses (Adams and Chapman 2005): (1) environmental hazard
classification of substances; (2) ranking and/or selection of priority substances; (3) selection
of contaminated sites for further evaluation; (4) derivation of water, soil and sediment quality
guidelines or criteria for individual substances; and (5) ecological risk assessment, both site-
specific (i.e., local) and generic (i.e., regional) in conjunction with appropriate exposure data.
Hazard identification and classification procedures of commercial chemicals exercised in
most jurisdictions are based upon three criteria: Persistence (P), potential for
Bioaccumulation (B), and Toxicity or inherent Toxicity (T or iT). The PBT criteria are also
used in the regulatory context to rank and identify substances of concern. In Canada, a PBiT
approach is used for categorizing substances on the Domestic Substances List (DSL) to
determine if further screening is required. In the U.S., PBT criteria have been used to
identify substances of concern for waste minimization, emission reporting, and for the
identification of substances for stricter regulations (e.g., air, water and solid waste).
Depending upon the assessment findings, actions to reduce exposure may be taken. In the
framework of New Chemicals Policy of the European Union (EU), discussions are ongoing
on whether to use PBT criteria to identify substances of very high concern.
1.4.2 Risk Assessment
Risk assessment is the determination of quantitative or qualitative value of risk related to a
given exposure situation and a recognized hazard. Thus, quantitative risk assessment takes
hazard assessment one step farther by requiring calculation of two components of risk: the
magnitude of the potential harm (i.e., hazard), and the probability that the harm will occur.
Risk assessment combines results of environmental toxicology with fate and exposure
assessments. For example, assessment entails calculating the ratio of the toxicity effect
concentration for a given organism and scenario to the expected environmental concentration
to which that organism is exposed to is determined. A ratio greater than one indicates a
margin of safety within which limited environmental impact is expected.
7
1.4.3 Life Cycle Assessment
LCA is an objective process to evaluate the environmental burdens associated with a product,
process, or activity over its life cycle (cradle-to-grave) by identifying energy and materials
used and wastes released to the environment, and to evaluate and implement opportunities to
affect environmental improvements (SETAC 1990, Heijungs 1992). LCA is comprised of
four steps. The first stage involves defining the goal, scope and boundaries of the project. In
the second stage, a life cycle inventory (LCI) is compiled, consisting of the quantitative
inputs and outputs over the product’s life cycle. The third step translates inventory items into
environmental burdens based on a product’s life cycle. Finally, the results are reviewed to
minimize environmental burden in light of the goals set out in the first stage. Life cycle
assessment assesses environmental burden according to a functional unit (e.g., assessing the
environmental impact of a tooth brush according to 100 uses of a brush) of the product or
process. Characterization Factors (CFs) connect inventory measurements (e.g., an emission
of 1 mg benzene/1 functional unit of product) with an “incremental” impact (e.g., probability
of an adverse effect/1 functional unit of a product). For the impact category of chemical
toxicity, CFs are calculated using first, Fate Factor (FFs) that numerically relate emissions of
substances to their fate and secondly, toxicity Effect Factors (EFs) that connect fate with
incremental adverse effect on one or more receptors.
The method used to obtain CFs within LCIA is common to that of other chemical hazard
assessments such as ecological risk assessment and hazard ranking. The differences among
these methods lie in (1) the multimedia model, the use of generic environmental data in an
“evaluative environment” versus site-specific data to simulate an actual environment, (2) in
the effects assessment, the use of generic versus site-specific toxicity data, (3) estimation of
hazard versus risk, and (4) estimation of hazard or risk related to total chemical emissions or
an incremental increase in emissions.
1.5 Modelling Metal Movement
There is a crucial need to answer the deceptively simple question, “What is the fate and harm
caused by metals after they are released in the environment?” In other words, there is a need
8
to link metal emissions with resultant concentrations and distribution that will cause adverse
environmental effects. This linkage provides the pathways analysis necessary to conduct a
hazard or ecological risk assessment or to derive CFs, with the final goal of establishing
reasonable emission rates. Mathematical models provide a useful means of addressing this
question.
1.5.1 Modelling Challenges
Traditionally multi-media mass balance models have been used to relate an emission of
chemical into an evaluative environment with the outcomes being quantitative expressions of
chemical distribution, persistence and concentrations (e.g., EQC Mackay et al. 1996, EUSES
Vermeire et al. 1997). However, the methods used for the hazard assessment of metals are
based on the same PBT criteria that were developed based on the experience gained with
organic chemicals. For example, Mackay-type multimedia fugacity models quantify
persistence (P), as the residence time of chemical in the evaluative environment or
compartments therein. Mackay and co-workers (Mackay et al. 2003, Harvey et al. 2007)
have argued that mass balance models can quantify P in a systematic fashion that
incorporates chemical-specific fate processes rather than using empirical measures of
degradation or transformation which are not suitable for metals given their infinite
persistence. These models, such as the single box EQC (Mackay et al. 1996), the 24
ecoregion model ChemCAN (Webster et al. 2004), and SimpleBox (Brandes et al. 1996)
have been most extensively developed for organic compounds that exist as single species.
When used for metals, the models fail to account for the existence of multiple,
interconverting species and the sensitivity of the distribution of metals amongst species on
ambient chemistry (e.g., Diamond et al., 1992, Diamond 1999, Verdonck and Sprang 2005).
Consequently, persistence or chemical residence time can be miscalculated and the resultant
concentrations overestimate toxicity.
Several criticisms have been raised regarding the current practice of categorization according
to PBT for metals (e.g., Chapman and Wang 2000, Mackay et al. 2003). First, traditional
degradation mechanisms used for organic chemicals to evaluate persistence of metals have
been criticized as inappropriate (Canada/European Union 1996). Metals are naturally
9
persistent in the environment because they do not degrade to other elements of less
environmental concern. The use of P for organic chemicals stems from its proportionality to
biotic exposure and its indication of the potential for long range transport. For metals, an
alternative view is necessary to capture the fate processes that give rise to environmental
concentrations and hence biotic exposure. These fate and exposure processes depend on
metal speciation, which in turn depends on ambient chemistry and metal concentrations.
This point gives rise to the second criticism that under PBT, metals are treated as a single
species rather than multiple, interconverting species where speciation influences fate and
effects. Metals usually exist as several species that undergo reversible or irreversible inter-
conversion among, for example, dissolved species and sparingly soluble salts. Third, metal
exposure is a function of the emission rate plus background levels which can vary
geographically by orders-of-magnitude due to differences in geologic conditions. Fourth,
metal uptake and toxicity are highly sensitive to metal speciation, which as noted above, is
sensitive to ambient chemistry. The form and availability of the metal species can change
and are different for each metal element. Further, metal solubility is dependant on its forms
and therefore toxicity tests based on soluble salts may overestimate the bioavailability and
potential for toxicity for many metals, especially for the insoluble metal sulphide and oxide
forms. Fifth, assessing toxicity must account for the essentiality of many metals as
micronutrients, which implies that adverse effects can occur at low and high concentrations.
The uptake by and release of metals from organisms may be modulated or regulated by
physiological processes and exposure conditions, and some organisms can store certain
metals with no adverse physiological response. Finally, bioaccumulation of metals cannot be
estimated using octanol-water partition coefficients (Kow) unlike organic substances.
Bioconcentration and bioaccumulation factors (BCFs and BAFs) can be inversely related to
exposure concentration and are not reliable predictors of chronic toxicity or food chain
accumulation for most aquatic organisms and most metals (Chapman and Wang 2000). This
results in organisms from the cleanest environment (i.e., background) having the largest BCF
or BAF values, which is counterintuitive to the use of BCF and LogKow as originally derived
for organic substances (McGeer et al. 2003). Organic chemicals and metals have different
modes of exposure and toxicity (see Table 1.1).
10
Table 1.1: Summary of major differences in behaviour of organic chemicals compared to
metals and inorganic metal compounds in humans.
Non-Polar Organic Chemicals Metals
Persistence in body fat is common
because of lipid solubility (not capacity-
limited)
Often sequestered, bound to specific plasma
or tissue proteins (intrinsically capacity-
limited) or bone
Depending on substation pattern,
metabolism could be up-regulated, the
degree to which is species-specific
Metabolism is usually limited to oxidation
state transitions and alkylation/dealkylation
reactions
Due to complex metabolism, some non-
polar organics may be eliminated by
excretion in urine after
biotransformation or conjugation from
lipophilic forms to hydrophilic forms, or
in exhaled air if not metabolized
Predominantly eliminated in urine and also
feces because metal compounds are generally
of low molecular weight and are hydrophilic.
As a result of protein binding, a small
fraction of body burden may be eliminated
via hair and fingernails
Interactions with other structurally
similar (notably endogenous)
compounds may occur
Interactions among metals and between
metals and organics are numerous and occur
commonly during the processes of
absorption, excretion, and sequestration
Generally a lack of substance-specific
homeostatic mechanisms
Essential metals have homeostatic
mechanisms that maintain optimum tissue
levels over a range of exposures
Tissue uptake is most commonly a
blood flow-limited process, with passive
fugacity-driven partitioning into tissues
of low molecular weight compounds
although protein-carrier uptake also
occurs
Metals and their complexes are often ionized,
with tissue uptake (membrane transport)
usually via active uptake channels
Source: adapted from Goyer et al. (2004).
11
Several models are available to assess toxicological impacts in LCA. For example, using
EUSES-LCA model, Huijbregts et al. (2000) calculated ecotoxicity CFs for 181 substances,
in which 17 metals including two species of chromium and mercury are included. Although
the method of Huijbregts et al. (2000) provides a simple, reproducible and defensible method
of adjudicating numerous substances, when applied to metals the method is also subject to
the same criticisms as the PBT method as discussed above.
Most models estimate the movement and concentrations of total dissolved and particulate
bound metal (e.g., Diamond 1995). In reality, transition metals form many aqueous species
as a result of hydrolysis and complexation reactions (Kohler et al. 1996). Interactions among
species alter transport potentials of total metal (Jennings et al. 1982). In addition, use of total
dissolved metal rather than species concentrations can result in unrealistically conservative
discharge levels for environmental protection and does not provide an understanding of the
biological implications of metal discharges (e.g., Allen and Hansen 1996). The free metal
ion is considered the most bioavailable species, but other labile metal species may also be
bioavailable (Chapman et al. 1998). Hence, to estimate or predict the bioavailable fraction of
total metal in aquatic systems, estimating chemical speciation is essential (Chapman and
Wang 2000).
In practice, the PBT criteria are often evaluated independently for each environmental
medium (e.g., air, water, sediment, and soil), for both organic chemicals and metals. This
approach leads to problems of interpretation because it misses the linkages amongst media.
As a result, the conclusions that are drawn are often of questionable validity in terms of
predicting potential risk for the real world.
1.5.2 Research Developments
Several consensus statements, notably those of Pellston workshop (Adams and Chapman
2005), the Apeldoorn Declaration (2004), and the Clearwater Consensus (Diamond et al.
2010) have recommended alternative approaches to PBT for assessing the hazard associated
12
with metals using a framework consistent with that for organics and other chemicals to
provide equal and fair evaluation while ensuring that both the environmental and human
health are protected. Environment Canada has revised its PBiT approach to categorizing
substances on Canada’s DSL. This revised approach recognizes that metals and inorganic
compounds have infinite persistence and that they do not exert toxicity through
bioaccumulative processes. Given their infinite persistence, discrimination using the revised
approach for inorganics primarily depends on toxicity, since bioaccumulation data were not
considered. While this approach avoids some of the potential pitfalls of applying PBT
approaches for inorganics, relying on iT alone to provide discrimination neglects differences
in the environmental behaviour of inorganic substances. Furthermore this system applies
different criteria depending on the substance type since only iT is considered for inorganic
metal compounds.
Diamond and co-workers addressed the need to consider metal chemistry within a
multimedia model by developing a general, fugacity-type model that considers multiple
interconverting species such as metals (e.g., Diamond et al. 1992, Diamond 1999). Recently
they developed a model that loosely couples a geochemical speciation model such as
MINEQL+ (Schecher and McAvoy 1992) or WHAM (Tipping 1998) to their multi-species
fate model in order to capture the dependence of metal distribution and fate on ambient
chemistry (Bhavsar et al. 2004a, Bhavsar et al. 2004b, Gandhi et al. 2007). This model,
named TRANSPEC (TRANsport and SPECiation), has been applied to cationic metals such
as Cd, Cu, Pb, Ni and Zn (Bhavsar et al. 2008) as well as mercury in lakes and reservoirs
(Gandhi et al. 2007). TRANSPEC-II considers metal fate in soils in addition to aquatic
systems (Bhavsar et al. 2008a). The models have been used to explore the effect of ambient
chemistry and speciation calculations on estimates of fate and ecotoxicity effects (Bhavsar et
al. 2008b).
The results from these and other studies of metal chemistry, fate and effects, reinforces the
importance of metal speciation to fate and effects (e.g., Peijnenburg et al. 1997). The
distribution of metal between soluble and particulate phases controls the dominant transport
processes and hence overall fate (Diamond et al. 1990, Bhavsar et al. 2004a, Bhavsar et al.
2004b). This distribution between phases and speciation within the soluble phase controls
13
bioavailability. It is well accepted that the metal free ion in solution is bioavailable, where its
fractional contribution to total and soluble metal is a function of aqueous chemistry (Morel
and Hering 1993). The challenge to developing predictive models is that each metal differs
in ligand binding preference and strength, as well as sensitivity to aqueous phase chemistry,
notably solution pH, ionic strength, concentrations of major cations, anions, and other metals
etc. Geochemical models such as MINEQL+ and WHAM have enabled the calculation of
free metal ion assuming that equilibrium conditions prevail.
Ecological effects assessments must account for metal bioavailability and its dependence on
ambient chemistry. The Biotic Ligand Model (BLM), which has been developed for four
metals (Ag, Cu, Cd, and Zn), has advanced the estimation of metal bioavailability by
accounting for competitive interactions and differences in binding affinities amongst ligands
that include the fish gill sites as a biotic ligand (Di Toro et al. 2001). The model calculates
the fraction of free metal ions that binds with fish gills at a given water chemistry and can
cause adverse effects. The model enables the calculation of an effects ratio as the quotient of
free metal binding to the biotic ligand with a critical binding concentration at which no or a
minimal toxicological effect will occur at a given environmental chemistry. Both
components of the ratio account for metal bioavailability in the same way. For metals for
which the BLM has not yet been developed, the approach proposed as the Free Ion Activity
Model (FIAM) can be used. This approach assumes that free metal ion is the only
bioavailable fraction of the total metal and the chemical activity of free metal ion can be
related to toxic effects in organisms (Campbell 1995). In this model, the hazard quotient
(HQ) is obtained as the ratio of free metal ion concentration calculated for ambient
conditions (e.g., Predicted Environmental Concentration or PEC) and free metal ion
concentration in the toxicological benchmark (e.g., Predicted No Effects Concentration or
PNEC).
To estimate potential ecological hazard or risk, a hazard quotient is often obtained for a
particular biotic species that is either deemed to be representative, sensitive or a keystone
species. An alternative approach is to use toxicological information (e.g., No Observed
Effect concentration or NOECs, Ecological Concentrations that would cause adverse effects
in 50% of the exposed organisms or EC50) from a wide range of species, where the
14
information is expressed according to a statistically-derived Species Sensitivity Distribution
(SSD). From this non-linear SSD, one can calculate the toxicological benchmark that is
protective of a percentage of species rather than a single species or the probability of a
potential risk to a random species exposed to an environmental concentration (Traas et al.
2002). With the former interpretation, the results are expressed as the Potentially Affected
Fraction (PAF) of species and if chemical mixtures are considered, a multiple substances
PAF (msPAF). Huijbregts et al. (2002) and van De Meent and Huijbregts (2005) discuss the
use of PAF and msPAF within LCIA. Gaudet et al. (2002) used the SSD-approach to
develop national soil and sediment guidelines for Canada under the auspices of the Canadian
Council of Ministers of the Environment (CCME).
During the Pellston workshop (Adams and Chapman 2005), a consensus was reached that the
use of a multimedia model to estimate fate and effects in an ‘evaluative environment’ is
desirable because it is applicable to both metals and organic chemicals and would allow for
comparison of the hazards posed by both classes of substances. This approach estimates the
rate at which a metal or metal substances can enter a prescribed ecosystem (e.g., unit world
or virtual world) before reaching a concentration, at steady-state or after a defined time
period, in one of the compartments of the ecosystem (e.g., water, sediment, or soil) that
causes adverse effects to biota. Such an approach integrates metal environmental chemistry
and fate to link the emission with potential to cause toxic effects. Because the model is not
intended to represent a specific location or processes specific to a location but rather a
representative setting that is typical of the class of environments being evaluated, the model
structure and its intended use is similar to that of the European Union System for the
Evaluation of Substances (EUSES) (EC 1996, Vermeire et al. 1997). However,
implementation of the above approach requires the following information to consider while
modelling (Adams and Chapman 2005):
• the model must incorporate multiple media and be capable of assessing metal
speciation and other important fate properties by each compartment;
• the model must balance the competing needs of simplicity and transparency on one
side and realism on the other.
15
• the model should be capable of running in steady-state or dynamic modes;
• the number, nature and properties of the relevant compartments must be critically
analyzed to reflect the natural environment and its variability;
• the inter-media transport parameters, such as soil run-off and sediment deposition
rates, should be estimated reasonably;
• the model must analyze the effects of the mode of introducing loadings (emissions) to
the generic environment: to air, to water, to soil or all media;
• model parameterization is needed to reflect the wide range of ambient conditions
relevant to questions of metal fate and toxicity and to appreciate the implications of
choices made in the parameterization process.
1.6 Research Goals
The aim of my doctoral thesis work was to develop and evaluate a consistent and general
framework for adjudicating chemical hazard with a view to addressing the above criticisms
for metals. Specifically, the goal was to incorporate metal-specific considerations into the
fate and toxicity assessment of metals within the general context of chemical hazard
assessment. Although the new method is tailored towards metals, it should also be consistent
with the PBT approach used in several jurisdictions, ecotoxicity characterization within
LCIA, and other chemical hazard assessment methods.
The research stems from four main bodies of knowledge: (1) chemical hazard and risk
assessment and their basis in fate and exposure assessments of organic compounds using
multimedia mass balance models (e.g., Cowan et al. 1995), (2) metal chemistry, fate and
toxic effects (e.g., Diamond 1995, Peijnenburg et al. 1997, Chapman and Wang 2000), (3)
ecological effects assessment (e.g., Suter 1993), and (4) LCA, LCIA and the development of
CFs for use in LCIA (e.g., Huijbregts et al. 2000). The research was undertaken as part of a
multi-investigator project with colleagues in the Netherlands. These colleagues are co-
authors of the publications that form the basis of this thesis.
16
The thesis is organized into four chapters that are papers either published or in a stage of
publication, bookended by this Introduction and by a concluding chapter. Chapter 2 presents,
in detail, the development of the new generic method for assessing the hazard associated with
the release of substances into environment. The method is then applied to calculate and
compare CFs for freshwater ecotoxicity in LCIA using 12 different freshwater systems in
European Union. This chapter is published in the journal of Environmental Science and
Technology (Gandhi et al. 2010). The principal outcome of this exercise was the presentation
of revised freshwater ecotoxicity CFs for well studied cationic metals (copper, nickel and
zinc) for use in LCIA. While conducting this research, I explored the sensitivity of factors
such as variability in the chemistry of freshwater systems that affect metal bioavailability,
fate and toxicity. The method development and exploration of sensitivities were guided by
the goal of arriving at a single, harmonized method for assessing the hazard of inorganic and
organic chemicals and the need to have a generic and simple approach that is consistent with
the practice of LCIA.
The exploration of the sensitivity of metal ecotoxicity was explored in the context of the
Canadian environment. I achieved this by extending the single-box multimedia model EQC
(Mackay et al. 1996) and the regional multimedia contaminant fate model ChemCAN
(Webster et al. 2004) using the new metal modelling framework presented in Chapter 2.
Thus, Chapter 3 presents the spatially-differentiated version of the new modelling
framework, with its application to Canadian environment using the environmental and
landscape properties for the 24 ecoregions of Canada. This chapter is published in
Chemosphere (Gandhi et al. 2011a).
In Chapter 4, I examined the practical implications of using the new method for estimating
freshwater ecotoxicity in LCIA by means of two case studies. I calculated and compared
metal CFs and LCIA outcomes for freshwater ecotoxicity of each of the two case studies
using four models: USES-LCA 1.0 (Huijbregts et al. 2000), USES-LCA 2.0 (van Zelm et al.
2009), USEtoxTM
using the interim approach (Hauschild et al. 2008, Rosenbaum et al. 2008),
and the new method with the geochemical correction in USEtoxTM
(Gandhi et al. 2010). This
chapter is published in International Journal of Life Cycle Assessment in a special issue on
modelling ecological and health impacts using USEtoxTM
(Gandhi et al. 2011b).
17
In Chapter 5, I used the new model approach to develop a method for calculating Critical
Loads (CLs). A CL quantifies a chemical loading to a specified environment that will
minimize an adverse effect which in this case is freshwater ecotoxicity. The CLs approach
has been adopted in the regulatory arena to estimate, for example, emissions of acidifying
species. Within this analysis we discussed the application of the new method of estimating
metal fate and toxicity in the context of a Unit World Model (UWM) framework as
suggested during the Pellston workshop (Adams and Chapman 2005). The UWM framework
was used to explore the implications of the choice of parameter values, in the same vein as in
Chapter 4. However, this application involves a very detailed analysis using the measured
chemistries of three Canadian lakes. This chapter is being published in Environmental
Toxicology and Chemistry (Gandhi et al. 2011c).
Finally, I conclude my thesis with Chapter 6 that draws conclusions and discusses
recommendations for future work.
1.7 References
Adams W.J. and Chapman, P.M. (2005) Assessing the hazard of metals and inorganic metal
substances in aquatic and terrestrial systems: summary of a SETAC Pellston
workshop. Pensacola, FL. USA.
Apeldoorn. (2004) Declaration of Apeldoorn on LCIA of non-ferro metals. Results of a
workshop by a group of LCA specialists, held in Apeldoorn, NL, April 15th, 2004.
Achterberg E.P., Berg C.M.G.V.D., Boussemart M., and Davison W. (1997) Speciation and
cycling of trace metals in Esthwaite Water: A productive English lake with seasonal
deep-water anoxia. Geochimica et Cosmochimica Acta 61(24):5233-5253.
Allen H.E. and Hansen D.J. (1996) The importance of trace metal speciation to water quality
criteria. Water Environment Research 68(1):42-54.
18
Bhavsar S.P., Diamond M.L., Evans L.J., Gandhi N., Nilsen J., and Antunes P. (2004a)
Development of a coupled metal speciation-fate model for surface aquatic systems.
Environmental Toxicology and Chemistry 23(6):1376-1385.
Bhavsar S.P., Diamond M.L., Gandhi N., and Nilsen J. (2004b) Dynamic coupled metal
TRANsport-SPECiation model: Application to assess a zinc contaminated lake.
Environmental Toxicology and Chemistry 23(10):2410-2420.
Bhavsar S.P., Gandhi N. and Diamond M.L. (2008a) Extension of coupled multispecies
metal TRANsport and SPECiation (TRANSPEC) to soil. Chemosphere 70:914-924.
Bhavsar S.P., Gandhi N., Diamond M.L., Lock A.S., Spiers G. and Alfaro De La Torre M.C.
Effects of estimates from geochemical models on metal fate predicted by coupled
speciation-fate models. Environmental Toxicology and Chemistry 23:2410-2420.
Campbell P.G.C. (1995) Interactions between trace metals and aquatic organisms: A critique
of the free-ion activity model. In Tessier A, Turner DR, eds, Metal Speciation and
Bioavailability in Aquatic Systems. John Wiley, New York, pp 45-102.
Canada/European Union (1996) Technical workshop on biodegradation/persistence and
bioaccumulation/biomagnification of metals and metal compounds. Brussels,
Belgium.
Carignan R. and Tessier A. (1985) Zinc Deposition in Acid Lakes: The Role of Diffusion.
Science 228(4707):1524-1526.
Carvalho P.S.M.D., Zanardi E., Buratini S.V., Lamparelli M.C., and Martins M.C. (1998)
Oxidizing effect on metal remobilization and Daphnia Similis toxicity from a
Brazilian reservoir sediment suspension. Water Research 32(1):193-199.
Chapman P.M. and Wang F. (2000) Issues in ecological risk assessment of inorganic metals
and metalloids. Human and Ecological Risk Assessment 6(6):965-988.
Chapman P.M., Wang F., Janssen C., Persoone G., and Allen H.E. (1998) Ecotoxicology of
metals in aquatic sediments: binding and release, bioavailability, risk assessment,
19
and remediation. Canadian Journal of Fisheries and Aquatic Sciences 55(10):2221-
2243.
Cowan C.E., Versteeg D.J., Larson R.J., and Kloeppersams P.J. (1995) Integrated Approach
of Environmental Assessment of New and Existing Chemicals. Regulatory
Toxicology and Pharmacology 21(1):3-31.
Diamond M.L., Mackay D., and Welbourn P.M. (1992) Models of multi-media partitioning
of multi-species chemicals: the fugacity/aquivalence approach. Chemosphere
25:1907-1921.
Diamond M.L. (1995) Application of a mass balance model to assess in-place Arsenic
pollution. Environmental Science and Technology 29(1):29-42.
Diamond M.L. (1999) Development of a fugacity/aquivalence model of mercury dynamics in
lakes. Water, Air, and Soil Pollution 111:337-357.
Diamond M.L., Page C.A., Campbell M., McKenna S., and Lall R. (1999) Life-cycle
framework for assessment of site remediation options: Method and generic survey.
Environmental Toxicology and Chemistry 18(4):788-800.
Diamond M.L., Gandhi N., Adams W.J., Atherton J., Bhavsar S.P., Bulle C., Campbell
P.G.C., Dubreuil A., Fairbrother A., Farley K., Green A., Guinee J., Hauschild
M.Z., Huijbregts M.A.J., Humbert S., Jensen K.S., Jolliet O., Margni M., McGeer
J.C., Peijnenburg W., Rosenbaum R., van de Meent D., Vijver M.G. (2010) The
Clearwater consensus: the estimation of metal hazard in fresh water. International
Journal of Life Cycle Assessment 15:143-147.
Di Toro D.M, Allen H.E., Bergman H., Meyer J.S., Paquin P.R., and Santore C.S. (2001)
Biotic ligand model of the acute toxicity of metals. 1. Technical basis.
Environmental Toxicology and Chemistry 20:2383-2396.
[EC] European Commission. 1996. EUSES documentation: The European Union System for
the Evaluation of Substances. National Institute of Public Health and the
Environment (RIVM), The Netherlands.
20
Forstner U., Ahlf W., Calmano W., Kersten M., and Salomons W. (1986) Mobility of Heavy
Metals in Dredged Harbor Sediments. In Sediments and Water Interactions. Edited
by P. G. Sly. Springer-Verlag, USA, pp. 371-380.
Gandhi N., Bhavsar S.P., Diamond M.L., Kuwabara J.S., Marvin-DiPasquale M. (2007)
Development of mercury speciation, fate and bioaccumulation (BIOTRANSPEC)
model: application to the Lahontan Reservoir. Environmental Toxicology and
Chemistry 26:2260-2273.
Gandhi N., Diamond M.L., van de Meent D., Huijbregts M.A.J., Peijnenburg W., Guinee J.
(2010) New method for calculating comparative toxicity potential of cationic metals
in freshwater: Application to copper, nickel, and zinc. Environmental Science and
Technology 44:5195-5201.
Gandhi N., Huijbregts M.A.J., van de Meent D., Peijnenburg W., Guinee J., Diamond M.L.
(2011a) Implications of geographic variability on Comparative Toxicity Potentials
of Cu, Ni and Zn in freshwaters of Canadian ecoregions. Chemosphere 82:268-277.
Gandhi N., Diamond M.L., Huijbregts M.A.J., Guinee J., Peijnenburg W., van de Meent D.
(2011b) Implications of considering metal bioavailability in estimates of freshwater
ecotoxicity: Examination of two case studies. International Journal of Life Cycle
Assessment In press.
Gandhi, N., Bhavsar, S.P., Diamond, M.L. (2011c) Critical load analysis in hazard
assessment of metals using a unit lake model. Environmental Toxicology and
Chemistry In press.
Gaudet C.L., Bright D., Adare K., and Potter K. (2002) A rank-based approach to deriving
Canadian soil and sediment quality guidelines. In The Use of Species Sensitivity
Distributions in Ecotoxicology. Edited by L. Posthuma, Th.P. Traas and G.W. Suter.
CRC Press, Boca Raton, FL, USA, pp. 255-273.
21
Goyer R., Golub M., Choudhury H., Hughes M., Kenyon E., Stifelman M. (2004) Issue paper
on the human health effects of metals. U.S. EPA, August 19, 2004. Available online
at http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=59052
Jennings A.A., Kirkner D.J., and Theis T. L. (1982) Multicomponent equillibrium chemistry
in groundwater quality models. Water Resources Research 18(4):1089-1096.
Hamilton-Taylor J., Davison W., and Morfett K. (1996) The biogeochemical cycling of Zn,
Cu, Fe, Mn, and dissolved organic C in a seasonally anoxic lake. Limnology and
Oceanography 41(3):408-418.
Harvey C., Mackay D., and Webster E. (2007) Can the unit world model concept be applied
to hazard assessment of both organic chemicals and metal ions? Environmental
Toxicology and Chemistry 26:2129-2142.
Hauschild M.Z., Huijbregts M., Jolliet O., Macleod M., Margni M., van de Meent D.,
Rosenbaum R.K., and McKone T.E. (2008) Building a model based on scientific
consensus for Life Cycle Impact Assessment of chemicals: the search for harmony
and parsimony. Environmental Science and Technology 42:7032–7037.
Heijungs H. (1992) Environmental life cycle assessment of products. Guidlines and
backgrounds. Centre of Environmental Sciences, Leiden, The Netherlands.
Huijbregts M.A.J., Thissen U., Guniee J.B., Jager T., Kalfe D., van de Meent D., Ragas
A.M.J., Wegener Sleeswijk A., and Reijnders L. (2000) Priority assessment of toxic
substances in life cycle assessment, I: Calculation of toxicity potentials for 181
substances with the nested multi-media fate, exposure and effects model USES-
LCA. Chemosphere 41: 541-573.
Kohler M., Curtis G.P., Kent D.B., and Davis J.A. (1996) Experimental investigation and
modeling of Uranium(VI) transport under variable chemical conditions. Water
Resources Research 32(12):3539-3551.
22
Mackay D. and Diamond M.L. (1989) Application of the QWASI (Quantitative Water Air
Sediment Interaction) fugacity model to the dynamics of organic and inorganic
chemicals in lakes. Chemosphere 18: 1343-1365.
Mackay D., Di Guardo A., Paterson S., Cowan C.E. (1996) Evaluating the environmental
fate of a variety of types of chemicals using the EQC model. Environmental
Toxicology and Chemistry 15: 1627-1637.
Mackay D, Webster E., Woodfine D., Cahill T., Doyle P., Couillard Y., and Gutzman D.
(2003) Towards consistent evaluation of the persistence of organic, inorganic and
metallic substances. Human and Ecological Risk Assessment 9: 1445-1474.
McGeer J.C., Brix K.V., DeForest D.K., Brigham S.I., Skeaff J.M., Adams W.J., and Green
A. (2003) Bioconcentration factor for the hazard identification of metals in the
aquatic environment: a flawed criterion? Environmental Toxicology and Chemistry
22: 1017-1037.
Morel F.M.M. and Hering J.G. (1993) Principles and Applications of Aquatic Chemistry.
John Wiley & Sons, New York, NY, USA.
[OECD] Organization for Economic Cooperation and Development. (1995) Test methods for
hazard and risk determination of metals and inorganic metal compounds. Paris,
France.
[OECD] Organization for Economic Cooperation and Development. (2001) Harmonized
integrated hazard classification system for human health and environmental effects
of chemical substances. Available from http://www.oecd.org/ehs/class/HCL6.htm
Patrick W. H., Jr. and Verloo M. (1998) Distribution of soluble heavy metals between ionic
and complexed forms in a saturated sediment as affected by pH and redox
conditions. Water Science and Technology 37(6-7):165-172.
Peijnenburg W.J.G.M., Posthuma L., Eijsackers H.J.P., and Allen H.E. (1997) A conceptual
framework for implementation of bioavailibility of metals for environmental
management purposes. Ecotoxicological Environmental Safety 37: 163-172.
23
Petersen W., Wallmann K., Li P., Schroeder F., and Knauth H.D. (1995) Exchange of trace
elements at the sediment-water interface during early diagenesis processes. Marine
Freshwater Research 46:19-26.
Rosenbaum R.K., Bachmann T.M., Swirsky Gold L., Huijbregts M.A.J., Jolliet O., Juraske
R., Koehler A., Larsen H.F., MacLeod M., Margni M., McKone T.E., Payet J.,
Schumacher M., van de Meent D., Hauschild M.Z. (2008) USEtox—the
UNEP/SETAC toxicity model: recommended characterisation factors for human
toxicity and freshwater ecotoxicity in life cycle impact assessment. International
Journal of Life Cycle Assessment 13:532–546.
Sauvé S., Dumestre A., McBride M., and Hendershot W. (1998) Derivation of soil quality
criteria using predicted chemical speciation of Pb+2
and Cu+2
. Environmental
Toxicology and Chemistry 17(8):1481-1489.
Sauvé S., Hendershot W., Allen H.E. (2000). Solid-solution partitioning of metals in
contaminated soils: dependence on pH, total metal burden, and organic matter.
Environmental Science and Technology 34: 1125-1131.
Schecher W.D. and McAvoy D.C. (1992) MINEQL+ - A Chemical Equilibrium Modelling
System, Ver 4.0. Environmental Reserach Software, Hallowell, ME, USA.
Slotton D.G. and Reuter J.E. (1995) Heavy metals in intact and resuspended sediments of a
California Reservoir, with emphasis on potential bioavailability of copper and zinc.
Marine Freshwater Research 46:257-265.
Smith S.L. and Jaffe P.R. (1998) Modeling the transport and reaction of trace metals in
water-saturated soils and sediments. Water Resources Research 34(11):3135-3147.
Suter G.W. II. (1993) Ecological Risk Assessment. Lewis Publishers, Boca Raton, Florida,
USA.
Tipping E. (1998) Humic ion-binding model VI: An improved description of the interactions
of protons and metal ions with humic substances. Aquatic Geochemistry 4:3-48.
24
Traas Th.P., Van de Meent D., Hamers T., Aldenberg T., Posthuma L., and De Zwart D.
(2001) The Potentially Affected Fraction as a measure of ecological risk. In: The
Use of Species Sensitivity Distributions in Ecotoxicology. Edited by L. Posthuma,
Th.P. Traas and G.W. Suter. CRC Press, Boca Raton, FL, USA, pp. 315-344.
van de Meent D. and Huijbregts M.A.J. (2005) Calculating LCA effect factors from PAF-
based ecotoxicological response functions. Environmental Toxicology and
Chemistry 24:1573-1578.
van Zelm R., Huijbregts M.A.J., van de Meent D. (2009): USES-LCA 2.0 – a global nested
multi-media fate, exposure, and effects model. International Journal of Life Cycle
Assessment 14: 282-284.
Verdonck F. and Van Sprang P. (2005) Review of LCA toxicity potential for copper. A
report prepared by Euras.
Vermeire T.G. (1997) European Union system for the evaluation of substances (EUSES).
Principles and Structure. Chemosphere 34:1823-1836.
Webster E., Mackay D., Di Guardo A., Kane D., and Woodfine D. (2004) Regional
differences in chemical fate model outcome. Chemosphere 55:1361-1376.
Wen X. and Allen H.E. (1999) Mobilization of heavy metals from Le An River sediment.
The Science of the Total Environment 227:101-108.
25
2. New Method for Calculating Comparative Toxicity Potential of
Cationic Metals in Freshwater: Application to Copper, Nickel, and Zinc©
2.1 Abstract
Current practice in chemical hazard ranking and toxic impact assessments is to estimate fate
and toxicity assuming the chemical exists in dissolved and particulate phases and, for metals,
that all dissolved species are equally bioavailable. This introduces significant error since
metal effects are related to the truly dissolved phase and free metal ion within it, not the total
dissolved phase. We introduce a Bioavailability Factor (BF) to the calculation of hazard or
Comparative Toxicity Potentials (CTPs) (also known as Characterization Factors; CFs) for
use in Life Cycle Impact Assessment (LCIA). The method uses for calculation (1)
USEtoxTM
for environmental fate, (2) WHAM 6.0 for metal partitioning and speciation in
aquatic systems, and (3) Biotic Ligand Model (BLM) for average toxicity. For 12 EU water-
types, we calculated medians (range) of CTPs of 1.5 × 104 (1.5 × 10
2 to 1.2 × 10
5), 5.6 × 10
4
(9.4×103 to 4.1×10
5), and 2.1×10
4 (7×10
3 to 5.8×10
4) day.m
3/kg for Cu, Ni, and Zn,
respectively, which are up to ∼1000 times lower than previous values. The greatest
contributor to variability in CTPs was the BF, followed by toxicity Effect Factor (EF). The
importance of the choice of water-type is shown by changes in the relative ranking of CTPs,
which are equally influenced by water chemistry and inherent metal-specific differences.
© Contents of this chapter have been adopted from the publication in the Environmental Science & Technology:
Gandhi, N., Diamond, M.L., van de Meent, D., Huijbregts, M.A.J., Peijnenburg, W.J.G.M., and Guinée, J.
(2010) New Method for Calculating Comparative Toxicity Potential of Cationic Metals in Freshwater:
Application to Copper, Nickel, and Zinc. Environ. Sci. Technol. 44: 5195-5201.
Reproduced with permission from Rightslink Printable License. Copyright (2011) American Chemical Society,
Licence Number: 2647210880447. A copy of the licence agreement is appended. A link to the published paper
can be found at http://pubs.acs.org/doi/pdf/10.1021/es903317a.
I was primarily responsible for the model development, data collection, model application, analyis of model
results, and writing of this manuscript.
26
2.2 Introduction
Screening level hazard identification and risk classification procedures for commercial
substances currently conducted in most jurisdictions are based upon three criteria: persistence
(P), potential for bioaccumulation (B), and toxicity (T) (European Commission 1996,
Government of Canada 1999, Cissel and Cromwell 1999). In the Life Cycle Impact
Assessment (LCIA) of products, ecotoxicity is dealt with using Comparative Toxicity
Potentials (CTPs) (also known as Characterization Factors; CFs), which are quantitative
estimates of the ecotoxicological impacts of substances per unit emission via pathways of
exposure to defined environmental recipients (Huijbregts et al. 2000, Hauschild 2005). The
methods for adjudicating ecotoxicological impacts rely on translating emissions into potential
adverse effects of substances while accounting for environmental fate. These methods are
based on those developed for nonpolar organic substances (Adams et al. 2000, Fairbrother
2002). This approach ranks metals among the most ecotoxic both in terms of their effect
levels and time-integrated potential toxicity (Huijbregts et al. 2000, Payet and Jolliet 2002).
Several criticisms have been raised regarding the current methods used to evaluate metal
impacts (Chapman and Wang 2000, Paquin et al. 2002, Adams and Chapman 2005). First,
metals are naturally occurring and can occur in very high concentrations. Second, unlike
organic chemicals, metals do not degrade in the environment and therefore the traditional
measures of persistence used for organic substances do not apply to metals (Adams et al.
2000). Third, metal exposure is a function of the emission rate plus background levels which
can vary geographically and in different environmental media (e.g., soil, sediment, water) by
orders-of-magnitude (Chapman and Wang 2000). Fourth, metal uptake and toxicity can be
highly sensitive to speciation/complexation and species interconversion that vary according
to ambient chemistry. For these reasons, an alternative method is necessary to capture the
potential effects of metals in hazard ranking and impact assessments (Adams and Chapman
2005). Further, the method needs to be scientifically rigorous and yet sufficiently simple to
make it tractable such that these competing needs are similarly balanced for organics and
metals.
27
Strandesen et al. (2007) developed a fate and exposure model to characterize aquatic
ecotoxicological impacts caused by a metal and its multiple species in LCIA. Their model
estimated equilibrium partitioning of metal species between solid and dissolved phases (i.e.,
partition coefficient Kd (L/kg)) based on metal-specific and pH-dependent empirical
relationships regardless of the environmental compartment (e.g., water, soil, sediment).
Their method can be improved by accounting for the dependence of metal partitioning and
speciation within the dissolved phase on additional factors such as presence/concentrations of
major cations, anions, and types of suspended matter (minerals vs organic material). The
method should also consider differences in the sorption of metals in (suspended) sediment
versus soil, in large part because of fundamental differences in binding sites. Bhavsar et al.
(2004, 2008a) developed a coupled metal speciation and fate/transport (TRANSPEC) model
for fate assessment that incorporated metal species-specific partitioning and distribution as a
function of ambient chemistry. Harvey et al. (2007), in their Unit World Model (UWM) for
hazard assessment of chemicals, used Kd values from the literature which avoided decisions
on parameterizing ambient chemistry. They recommended adjusting the compartment
volumes to compensate for the non-degradability of metals, adjusting which compartments
receive emissions, and performing shorter versus longer term dynamic simulations for
metals. The latter two models lack the toxicity calculations for metals that require
normalizing toxicity test data for ambient chemistry.
In this paper, we propose a general modeling framework for metals in freshwaters that is
illustrated for LCIA; the framework can be used for hazard ranking and risk assessment as
well. The framework is based on: (1) fate estimated using a multimedia mass balance model
for metals (e.g., Bhavsar et al. 2004, 2008a), (2) aquatic metal chemistry estimated by means
of a geochemical speciation model (e.g., Schecher and McAvoy 1992, Tipping 1998), and (3)
aquatic ecotoxicological effects estimated by means of the Biotic Ligand Model (BLM; Di
Toro et al. 2001). The modeling framework incorporates metal-specific geochemical
behavior into the fate and toxicity assessments. Note that the issues of analyzing ecotoxicity
potentials for metals in Life Cycle Assessment (LCA) not only come from LCIA but also
from the Life Cycle Inventory (LCI) stage (Gloria et al. 2006, Pettersen and Hertwich 2008).
We limit the scope of this paper to address issues related to LCIA. We present revised CTPs
28
for Cu, Ni, and Zn that express the relative hazard associated with their release into an
evaluative freshwater environment for which we have specified 12 water chemistries.
2.3 Methods
We adopt definitions and assumptions proposed in the Clearwater Consensus (Diamond et al.
2010) for the estimation of metal hazard in freshwaters. The bioavailable fraction of chemical
is defined as “...the fraction of the total amount of a chemical present in a specific
environmental compartment that, within a given time span, is either available or can be made
available for uptake by (micro)organisms from either the direct surrounding of the
organism...” (Peijnenburg and Jager 2004). The amount of total chemical is the sum of its
amount in total dissolved (or soluble) and particulate phases. The total dissolved phase is
further divided into colloidal, which is mainly associated with Dissolved Organic Carbon
(DOC), and truly dissolved fractions.
2.3.1 Current Practice
CTP incorporates the assessment of fate, exposure, and toxicity (effect) of a substance. The
fate component is expressed as Fate Factor (FF, day), representing the residence time of the
chemical in a specific compartment, and is calculated using an environmental multimedia
fate model (Huijbregts et al. 2000). A substance’s ecotoxicological impact is represented by
its Effect Factor (EF, m3/kg), indicating the average toxicity of a chemical expressed as a
Potentially Affected Fraction (PAF) of organisms (Pennington et al. 2004). Below we
present equations for freshwater aquatic ecotoxicity, however, the method can be extended to
other environmental compartments (e.g., soil or sediment). The CTPi,s (day.m3/kg) of
substance s emitted to compartment i is
[2.1] ssi,si, EFFFCTP ⋅=
where i can be any compartment including freshwater. The compartment-specific FF is
defined as the change in steady-state total dissolved amount of a substance in an
29
environmental compartment due to the incremental change in its emission (Huijbregts et al.
2000). The FFi,s for the freshwater compartment is:
[2.2] s,i
s,d
s,im
V.CFF
∆
∆=
which accounts for the transport efficiency of substances from compartment i to freshwater
and its persistence in the freshwater compartment, ∆Cd,s is the incremental change in the
steady-state concentration of the total dissolved substance s (kg/m3), V is the volume of
freshwater compartment (m3), ∆mi,s is the incremental change in the emission of total
substance s (total dissolved and particulate phases) to compartment i (kg/day), and d refers to
the total dissolved fraction of that substance.
The EFs is calculated as (Pennington et al. 2004, Rosenbaum et al. 2008):
[2.3] s10
5.0
C
TU.
TU
PAF
C
PAFEF
s,ds,d
s µ≈
∆
∆
∆
∆=
∆
∆=
where ∆PAF is the incremental change in the potentially affected fraction of biological
species in a freshwater community due to exposure to the total dissolved fraction d of
substance s, ∆TU is the change in toxic unit (TU), e.g., acute or chronic EC50, of substance s,
and 10µ is the geometric mean of substance’s toxicity across aquatic species (kg/m
3). In
equation 2.3, ∆PAF/∆TU can be calculated either (1) as a function of the slope of a Species
Sensitivity Distribution (SSD) developed for a specific water chemistry in addition to the
choice of the underlying statistical distribution (lognormal, log-logistic) in the case of a non-
linear dose-response function (van de Meent and Huijbregts 2005), or (2) by simply
assuming a constant value (e.g., 0.5) for a linear dose-response function (Rosenbaum et al.
2008).
2.3.2 Proposed Framework
Most fate and effect models (e.g., USES-LCA (Huijbregts et al. 2000); IMPACT2002+
(Jolliet et al. 2003)) do not distinguish between total dissolved and truly dissolved (or free
30
metal ion) fractions of a substance. These models assume that the total dissolved fraction d of
metal in an environmental compartment represents the fraction of bioavailable species k. Our
current understanding is that toxicologically relevant species is the metal free ion for cationic
metals (Campbell 1995, Paquin et al. 2002). Further, the toxic impact of a metal is described
best by its concentration at the biotic ligand site, and the degree of complexation of metals by
these biotic ligands depends on water chemistry. Therefore, the above assumption results in
an incorrect estimation of the potential toxic impact of a metal.
To remedy this overestimation of bioavailability by accounting for metal species-specific fate
and effects, we propose the use of a Bioavailability Factor (BF) to translate between total
concentration of a substance and the bioavailable fraction. The BF of a substance is
[2.4] s,t
s,k
sC
CBF
∆
∆=
where BFs is the bioavailable fraction of substance s in freshwater (dimensionless), ∆Ck,s is
the incremental change in the bioavailable fraction k of the total substance s (kg/m3) and
∆Ct,s is the incremental change in concentration of the total substance t (total dissolved and
particulate). We assume that the bioavailable fraction is within the truly dissolved fraction.
The free metal ion, which is assumed to be toxicologically active (Campbell 1995), is within
the truly dissolved fraction. Thus, at low environmental metal concentrations and for
constant ambient chemistry, BF is simply the fraction of truly dissolved metal within total
metal at equilibrium. The definition of BF can be applied consistently to organics and
metals: fate and effect models for organic substances assume that the total dissolved fraction
of the substance is bioavailable whereas models that quantify the toxicity of metals (e.g.,
BLM, FIAM) express the result in terms of truly dissolved or free metal ion fractions.
To accommodate BF in the calculation of CTP, FF and EF are redefined as:
[2.5] s,i
s,t
s,im
VCFF
∆
∆=
31
[2.6] s10
5.0
C
TU.
TU
PAF
C
PAFEF
s,ks,k
s µ≈
∆
∆
∆
∆=
∆
∆=
where ∆Ct,s and ∆Ck,s are the incremental changes in the steady-state concentrations of the
total (total dissolved and particulate) and bioavailable fractions k of substance s (kg/m3),
respectively. Thus, CTP of substance s can be calculated as:
[2.7] sssi,si, EFBFFFCTP ⋅⋅=
The use of BF to calculate CTP departs from the current practice in two ways. First, it
defines FF in terms of total rather than the total dissolved fraction of a substance entering a
system, which allows a direct link with the life cycle inventory data. Second, the BF
explicitly acknowledges that the total dissolved fraction of a substance is not necessarily the
bioavailable fraction. The BF allows equivalent treatment of metals and organics using the
same method. Explicitly including the BF also allows practitioners to update the translation
between the total substance and its bioavailable fraction as science progresses.
2.4 Model Selection and Parameterization
We selected popular and widely available models to calculate metal speciation/complexation
and aquatic ecotoxicity, and to couple with the fate model currently used in LCIA practice.
2.4.1 Fate Model
Recently, as a result of collective efforts of LCIA experts, a consensus LCIA model
USEtoxTM
has been developed through rigorous parameterization and considering important
fate and exposure processes from the fate and effects models listed above (Rosenbaum et al.
2008). We adopted USEtoxTM
to calculate FFs and then CTPs for metals. USEtoxTM
considers particulate and total dissolved phases of metals that are defined according to a
specified partition coefficient, Kd (L/kg). Although the total dissolved phase equals the truly
dissolved plus colloidal phases, current practice ignores the colloidal phase of chemicals and
32
hence the total dissolved and truly dissolved phases are assumed to be equal (Huijbregts et al.
2000, Harvey et al. 2007, Rosenbaum et al. 2008).
The fate expressions in the consensus model USEtoxTM
were used with the following
exceptions. As done with TRANSPEC (Bhavsar et al. 2004, 2008a), the modified version of
USEtoxTM
considers multiple, interconverting species within particulate, truly dissolved and
colloidal phases where interconversion is implicitly calculated based on specified values of
Kd and metal species fractions (Diamond and Mackay 1992). This approach differs from that
of Toose and Mackay (2004) who assumed constant species concentration ratios that do not
vary with ambient chemistry. The model considers all metals in air to be in the particulate
phase that is subject to wet and dry deposition. Air-water diffusive exchange is neglected
although this could be included for volatile metal species such as mono- and
dimethylmercury as done in Gandhi et al. (2007). Aside from the physical removal processes
of burial and outflow, chemical removal processes such as metal precipitation and co-
precipitation in soil, sediment and groundwater compartments should be included. For
application to freshwater-types in this paper, we did not include metal precipitation. The
values of Kd and species fractions vary according to compartment-specific chemistry data,
and can be calculated using geochemical models such as MINEQL+ (Schecher and McAvoy
1992) or WHAM (Tipping 1998) for aquatic systems.
2.4.2 Geochemical Speciation-Bioavailability Model
BF for metals is operationally defined by geochemical speciation models. According to these
models, metal bound to particles and organic matter, even in their labile phases, is not
considered bioavailable. To calculate Kd (L/kg) among total dissolved and particulate
phases, and speciation/complexation including the BF, we used the Windermere Humic-
Aqueous Model (WHAM 6.0; Tipping 1998). WHAM 6.0 is an equilibrium based metal
speciation/complexation model comprised of the Humic Ion-Binding Model VI and an
inorganic speciation code for aqueous solutions. An important advantage of WHAM is its
sophisticated treatment of metal binding to humic and fulvic acids in both particulate and
total dissolved phases. In addition, WHAM can also estimate metal adsorption to oxides of
Fe and Mn. However, the use of WHAM is limited for metal precipitation and redox related
33
reactions due to its inability to track changes in the thermodynamic distribution of
precipitated/redox coupled species.
The values of Kd and aqueous species fractions, including the fraction of free metal ion, for
freshwater chemistries were calculated “off line” using WHAM and then the results were
coupled with the fate model. Note that the model does not consider non-reactive, insoluble
native metal products such as copper ingots or insoluble minerals. Evaluation of insoluble
metal compounds requires the application of an additional model or procedure in which the
dissolution rates of these compounds into dissolved metal species can be estimated (Skeaff et
al. 2000).
2.4.3 Aquatic Ecotoxicity Model
We use the BLM approach, in which the toxic effects of metals are assumed to be directly
related to the concentration of metals bound to the biotic ligand, which depends on water
chemistry parameters such as pH and presence of competing cations (Di Toro et al. 2001).
The BLM combines chemical equilibrium modelling for metal speciation with a toxicity
model that relates metal accumulation at a biotic ligand to a toxic effect. Di Toro et al.
(2001) describe the structure of BLM in detail.
Currently, acute and chronic BLMs are available for Cu, Ni and Zn, for three aquatic
organism classes (algae, daphnids and fish). Consistent with the practice for calculating EF
in USEtoxTM
(Rosenbaum et al. 2008), we used values of chronic EC50 (concentration of
substance that causes an adverse effect in 50% of the exposed population) to calculate values
of µ for each metal and each aquatic organism. BLM parameters of each metal were
extrapolated across-biotic species within an organism class (e.g., fish) (Vijver et al. 2008).
When extrapolating BLMs across organism class (e.g., fish), it was assumed that conditional
stability constants for cations (metal of interest and major ions) and the biotic ligand,
mechanism of binding, and modes of action are similar across the organism class represented
in the calculation of HC50. These extrapolations to additional organisms allowed accepting
BLM parameters developed for algae and daphnids to other species that are sporadically
represented in the effects databases, such as amphibians, mollusks, and insects.
34
We used WHAM 6.0 to estimate metal speciation for toxicity test waters and selected water-
type (w). Each value of EC50 for each metal and each biotic species was normalized using
the chemistry of water-type w, giving a value of EC50-w (Vijver et al. 2008). All values of
EC50-w were then used to construct a water-type specific SSD assuming a log-normal
distribution to calculate µ.
2.4.4 Overall Model Structure and Parameterization
The modified metal fate and toxicity (BLM) models, that incorporated metal speciation
calculations, were then assembled into the USEtoxTM
framework such that metal assessments
were conducted on the fully (loosely) coupled sequence of metal speciation, fate and toxicity
models. The overall structure and connections amongst the models ensured that consistent
descriptions of environmental chemistry and characteristics were used throughout.
We applied the full model to estimate the BF, FF, EF, and CTP of the cationic metals Cu, Ni
and Zn for their emission to freshwater systems. We chose these metals because of
abundance of toxicity data and availability of BLMs. We assumed that metal emissions enter
the freshwater compartment with specified water chemistry in the form of total metal that is
then distributed among particulate and total dissolved phases at equilibrium. To explore the
effect of variability in freshwater chemistry, we assembled 12 sets of water-types based on
measured environmental chemistries in European Union (EU) surface waters (Van Tilborg
2002, Heijerick et al. 2005). Since water chemistry parameters co-vary, it is recommended
that the values of all chemistry parameters should be taken from one system (Diamond et al.
2010). We analyzed several sets of measured water chemistries for EU and categorized them
based on high, medium and low values of pH, DOC and hardness. The following operational
measures were used for this classification: pH >7.3 as high, 5.5 < pH <7.3 as medium and pH
<5.5 as low; DOC>9.0 mg/L as high, 4.5 <DOC <9.0 mg/L as medium and DOC <4.5 mg/L
as low; and hardness >75 mgCaCO3/L as high, 75 <hardness <175 mgCaCO3/L as medium
and DOC <175 mgCaCO3/L as low. We selected water chemistries to represent the range of
EU water-types and encompassed different combinations of pH (5.5-8.3), DOC (1.6-18.2
mg/L), and water hardness (8-225 mgCaCO3/L) categorized at low, medium and high scales
as shown in Table 2.1. Note that the combinations of high hardness and low pH are not
35
common in natural waters and therefore we do not have water-types to represent such
chemistry. We believe these sets represent most freshwater chemistries in EU. For each
water-type, we assumed a background total dissolved metal concentration of 1 µg/L for Cu
and Ni and 10 µg/L for Zn based on measured background concentrations in the European
surface waters.
We used the default database of stability constants in WHAM 6.0 (Tipping 1998). Metal
complexation with DOC was considered by metal binding to humic and fulvic acids of
dissolved organic matter (DOM), where we assumed that 50% of DOM was DOC. DOM
was considered ‘colloidal’ and was assumed to consist of a specified fraction of active fulvic
acid (%AFA) for ion binding and an inert fraction. We used %AFA values of 65% for Cu
(Bryan et al. 2002), 40% for Ni (Deleebeeck et al. 2008) and 60% for Zn (Cheng et al. 2005)
as previously calibrated values for metal binding with DOM. We also adjusted the values of
the stability constant of Me-fulvic acid complexes as 1.75 for log KMa(Ni) (Deleebeeck et al.
2008) and 1.8 for log KMa(Zn) (Cheng et al. 2005) in the WHAM 6.0 default database. We
further assumed that sulfide species did not play significant role in metal speciation for
freshwater types analyzed here. In absence of measured data and for consistency with
USEtoxTM
, we assumed the default value for total suspended solids (TSS) concentration of
15 mg/L for all water types (Rosenbaum et al. 2008). Metal sorption/complexation with Fe,
Mn and Al oxyhydroxides, as well as particulate organic matter (POM), was calculated using
a multi-dentate ligand approach (Tipping 1998). We estimated concentrations of POM using
POM:DOM ratios of 10:1 for oligotrophic water chemistries (DOC<=2 mg/L), 8:1 for
mesotrophic (2 < DOC <= 10.2 mg/L) and 6:1 for eutrophic systems (DOC > 10.2 mg/L)
(Wetzel 1983). We also assumed concentrations of 150, 10 and 20 µg/L Fe, Mn and Al,
respectively, in absence of measured values in order to calculate their oxides for providing
metal oxide surfaces for metal adsorption. A temperature of 15°C and a CO2 partial pressure
of 10-3.5 were set for all water types and metals. We analyzed the model outputs to provide
the following results for each water-type and metal: fractions of free metal ions relative to
total metal, fraction of truly dissolved metal, fraction of metal bound to colloids (represented
by humic and fulvic acids), and fraction of metal bound to particles, and values of LogKd.
36
Table 2.1: Freshwater chemistry data used in the geochemical model, WHAM 6.0, to estimate Bioavailability Factors (BFs) for Cu,
Ni, and Zn in 12 EU water-types (background metal concentrations of 1 µg/L for Cu and Ni and 10 µg/L for Zn were used for all
water-types).
Water-types Example Ecosystem Reference pH DOC Hardness Ca Mg Na K SO4 Cl
pH DOC Hardness mg/L mgCaCO3/L mg/L mg/L mg/L mg/L mg/L mg/L
EU Water-type 1 High High High Streams and brooks (33) 7.4 18.2 224 75.8 8.5 58.4 0.1 67 102
EU Water-type 2 High Med High Canals, large lakes and small lakes (33) 8.1 8.4 221 56.6 19.5 65.8 0.1 67 120
EU Water-type 3 High Med Med Mole, United Kingdom (34) 7.6 6.1 132 42.48 6.22 26.67 3.52 48.03 32.97
EU Water-type 4 High Low High River Rhine, Germany (34) 8.1 2.0 190 60.52 9.48 25.06 3.25 38.43 41.48
EU Water-type 5 High Low Med Segrino, Italy (34) 8.2 1.7 169 58.51 5.59 2.60 0.78 9.61 20.92
EU Water-type 6 High Low Low Lake Monate, Italy (34) 8.2 2.5 48 13.59 3.50 2.30 0.74 13.83 24.82
EU Water-type 7 Med High Med Ankeween, Netherlands (34) 7.3 17.8 165 52.10 8.58 11.79 0.82 109.51 20.21
EU Water-type 8 Med Low Med Small springs (33) 6.7 2.2 78 20.3 6.7 17 0.1 67 31
EU Water-type 9 Med Low Low Somerain, Belgium (34) 6.4 1.6 28 6.69 2.65 7.20 2.82 85.50 5.99
EU Water-type 10 Low High Low Skarsjon, Sweden (34) 5.5 10.3 8 2.40 0.49 7.89 6.22 2.79 2.41
EU Water-type 11 Low Med Low Bihain, Belgium (34) 5.9 8.9 10 2.48 0.95 6.39 1.80 2.88 8.37
EU Water-type 12 Low Low Low Clywydog, United Kingdom (34) 6.3 2.72 10 2.20 1.12 4.09 0.51 4.80 6.98
Catogory
37
The fate model used default landscape data and transport parameters in USEtoxTM
, except the
values of Kd for 12 water-types that were calculated using WHAM. Kd values for sediment
and soil compartments were set to default values of USEtoxTM
.
For the toxicity effects assessment, we assembled literature values of EC50 for chronic (e.g.,
data for ≥72 hours) exposures reported along with the test water conditions (e.g., pH, DOC
etc.). Our metal toxicity database included laboratory tests data for a minimum of three
biotic species classes (algae, daphnids, and fish). Depending on the mechanism of toxicity,
the BLM for each metal and aquatic organism can be different and requires distinct
parameterization of conditional binding constants (LogKBL) with the biotic ligand (as
reviewed by (Niyogi and Wood 2004). The values of LogKBL used in this application were
taken from chronic BLM studies and are listed in Table 2.2.
Table 2.2: Values of conditional binding constants (LogKBL) of binding metals and other
competing cations with biotic ligand for chronic BLMs used in this model application.
LogKBL Copper Nickel Zinc
Daphniaa Fish
b Algae
c Daphnia
c Fish
c Daphnia
d Fish
d
LogK Me-BL 8.02 8.02 4.00 4.00 4.00 5.3 5.5
LogK MeOH-BL 8.02 7.32 - - - - -
LogK MeCO3-BL 7.44 7.01 - - - - -
LogK Ca-BL - 3.47 2.1 3.25 3.6 3.2 3.6
LogK Mg-BL - 3.58 3.3 - 3.6 2.7 3.1
LogK Na-BL 2.91 3.19 - 3.24 - 1.9 2.4
LogK H-BL 6.67 5.40 5.9 - 6.8 5.8 6.3
fCu-BL (50%) 0.226 0.26 0.00373 0.0015 0.0143 0.127 0.246
For Cu binding to algaee, LogK = - 1. 431pH + 2.05
For Zn binding to algaed, LogK = 0.538pH + 2.25
a De Schamphelaere and Janssen 2004;
b De Schamphelaere and Janssen 2005;
c Deleebeeck et al. 2009;
d De Schamphelaere et al. 2005;
e De Schamphelaere et al. 2003
38
2.5 Results and Discussion
Below we present results for each constituent model within the framework for a unit
emission of Cu, Ni and Zn to the 12 EU freshwater-types. We then compare CTPs obtained
using the proposed approach with those calculated using the current USEtoxTM
method.
2.5.1 Kd Values
For the 12 EU water-types, the average values of LogKd (L/kg) for Cu, Ni and Zn were 4.1
(4.0-4.3), 3.9 (3.6-4.2), and 4.9 (4.4-5.4), respectively (Figure 2.1a). The average values
were similar to, but the ranges narrower than those listed in US EPA database of metal Kd
values for suspended sediment for Cu, Ni and Zn of 4.7 (3.1-6.1, n=70), 4.6 (3.5-5.7, n=30),
and 5.1 (3.5-6.9, n=75), respectively (http://www.epa.gov/athens/publications/reports/
Ambrose_600_R_05_074_Partition_Coefficients.pdf). Huijbregts et al. (2000) used LogKd
values of 4.7, 3.9 and 5.0 L/kg for Cu, Ni and Zn, respectively, to derive CTPs for these
metals. Harvey et al. (2007) used the values of 4.72, 4.80 and 5.26 L/kg for Cu, Ni and Zn,
respectively, in a critical load analysis of metals in UWM. There is a broad range in some of
the values of LogKd and the relative ranking of the metals differs among these sources.
Despite the importance of pH for metal distribution, DOC (p<0.05) and total suspended
solids (TSS; p<0.05) but not pH were significantly correlated with values of Kd for Cu and
Zn. Neither DOC nor TSS were significantly correlated with values of Kd for Ni (p=0.395
for DOC; p=0.395 for TSS) for the 12 water-types. Multiple linear regression models were
unable to provide relationships of Kd with pH, TSS, and DOC (r2: 0.029-0.380; p>0.05) for
these metals. In comparison, Bhavsar et al. (2008b) showed that DOC affected Cu speciation
most whereas TSS largely controlled values of Kd for Ni and Zn modelled using WHAM for
three freshwater systems of varying trophic status.
39
1
100
10000
10000001
100
10000
1000000
1000000000
20
40
60
80
1000.000001
0.00001
0.0001
0.001
0.01
0.1
1
BF
(F
racti
on
)
2
3
4
5
6
Lo
gK
dB
F (F
ractio
n)
FF
(da
ys)
EF
(m3/k
g)
a
b
c
d
CT
P (d
ay.
m3/k
g) e
Cu Ni Zn
Figure 2.1: Model results for Cu, Ni, and Zn using the chemistry of 12 EU water-types
described in Table 2.1. (a) WHAM estimated metal partition coefficients, LogKd (L/kg), used
40
in fate model, (b) WHAM estimated BFs (Bioavailability Factors; dimensionless; calculated
as a fraction of total metal in the bioavailable form), (c) freshwater FFs (Fate Factors, days)
for emissions in freshwater compartment calculated using the default parameter values of
USEtoxTM
model and WHAM estimated values of Kd for each water-type, (d) BLM
estimated metal EFs (Effect Factors; m3/kg) corrected for chemistry of each water-type, and
(e) comparison of metal CTPs (Comparative Toxicity Potentials; day.m3/kg) for water-types
and those calculated using the default parameters for metal assessment in USEtoxTM
(●).
Note FFs are for total metal and represent the residence time of metals in freshwater due to a
unit emission.
2.5.2 Fate Factors
The FFs of each metal for the “evaluative freshwater environment” set up in the USEtoxTM
fell within a factor of four despite the two order-of-magnitude range in values of Kd
generated by the 12 water-types (Figure 2.1b). The results suggest that FFs are not very
sensitive to water chemistry. Regardless of the metal, the FFs varied systematically with
LogKd (Figure 2.2). This is not surprising because only two processes, export through
outflow and sedimentation, control the fate of metals in the water and Kd controls the fraction
of chemical subject to one or the other process (Diamond et al. 1990). This is not true for
organic chemicals which are subject to these two processes plus volatilization and
degradation. In USEtoxTM
, water compartments are modeled such that transport by outflow is
a more important route of removal than sedimentation. Due to the relative simplicity of
processes governing the fate of metals, we suggest that an empirical relationship based on Kd
can be developed to obtain FFs. This empirical relationship can be used to explain
differences in FF among metals and among geographic sites.
A troubling issue is the role of TSS in this analysis. Through Kd, TSS influences the fraction
of metal lost via export through outflow versus sedimentative pathways. Consequently,
speciation/complexation and fate calculations require a consistent value of TSS. USEtoxTM
uses a value of TSS of 15 mg/l for the freshwater compartment which is linked to a
corresponding net sedimentation rate. The use of this default value of TSS (and the net
sedimentation rate) in USEtoxTM
“decouples” the parameter values used in the fate
41
calculation (FF) from those used in the speciation/complexation calculation (to calculate Kd,
BF and EF).
0
20
40
60
80
100
3.0 3.5 4.0 4.5 5.0 5.5 6.0
LogKd
Tot M
e F
F (
da
ys)
Cu
Ni
Zn
Regress
Figure 2.2: Estimated freshwater Fate Factors (FFs, days) of Cu, Ni, and Zn for their unit
emissions into the freshwater compartment using the default setting of USEtoxTM
model and
WHAM estimated values of Kd for the 12 EU water-types (see Table 2.1). Here FFs
represent residence times for total metals in freshwater after emission.
2.5.3 Bioavailability Factors
BFs varied across ~ 4, 2 and 2 orders-of-magnitude for Cu, Ni and Zn, respectively (Figure
2.1c), mainly due to variations in DOC, TSS and pH among the 12 water-types. The model
estimated that 77-87% of total Cu was bound to DOC and thus not bioavailable, whereas ~7-
66% of total Ni was estimated to be bioavailable, despite the similarity of the ranges of Kd of
42
these two metals (Figure 2.3). The fractions of free metal ions within the truly dissolved
phase varied ~3-98%, ~20-99% and ~63-99% for Cu, Ni and Zn, respectively (results not
shown). The range in estimates of the free metal ion within the total dissolved phase for Cu
was larger than that of Ni and Zn, mainly because of the influence of DOC for Cu speciation.
For previously reported metal CTPs (Huijbregts et al. 2000), the fractions of total dissolved
metal for freshwater (and hence the toxicologically relevant fraction under current practice)
were set at 57, 90 and 38% for Cu, Ni and Zn, respectively. These default values, on
average, reflect high and low bioavailability for Ni and Zn, respectively, when compared
with BFs calculated for the 12 water-types in EU. The corresponding default values of total
dissolved fraction recently revised in USEtoxTM
are 33, 61 and 73% for Cu, Ni and Zn,
respectively.
1E-6
1E-4
1E-2
1E+0
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
LogKd
BF
(F
rac
tio
n)
Cu
Ni
Zn
Figure 2.3: Values of Bioavailability Factors (BFs; dimensionless) calculated as the fraction
of total metal in the bioavailable form for Cu, Ni, and Zn using the chemistries of selected
EU water-types shown as the function of LogKd.
43
2.5.4 Effect Factors
EFs expressed in terms of truly dissolved metal for Cu, Ni and Zn, that were obtained using
the BLMs, varied by ~2, ~1 and ~1 order(s) of magnitude respectively for the 12 water-types
(Figure 2.1d). The corresponding average toxicity (10µ) values for all water-types were 2.2
(0.6-6.3), 180 (35-445) and 110 (50-150) µg/L for Cu, Ni and Zn, respectively. This range
reflects the fact that average toxicity depends not only on the intrinsic sensitivity of a biotic
species to the metal but also on the bioavailable fraction and competition for biotic ligands,
which is a function of water chemistry. Since BLM is semi-mechanistic, and stronger in
treating the chemistry than physiological aspects of ecotoxicity, it does not tease apart the
biological impacts of metal exposure. Default values of HC50 (≈10µ) in USEtox
TM are 133.5,
880 and 1050 µg/L for Cu, Ni and Zn, respectively. In contrast, Harvey et al. (2007) used
critical concentrations (equivalent to 10µ) of 9, 52 and 120 µg/L for Cu, Ni and Zn,
respectively, in their critical load analysis. Thus, considering metal speciation reduces
average toxicity by 5-50 times, depending on the metal and selected water chemistry.
2.5.5 Comparative Toxicity Potentials
CTPs of metals varied over three orders-of-magnitude among the 12 water-types which
predominantly reflects variability in their BFs and EFs (Figure 2.1e). The medians (range) of
CTPs were 1.5x104 (1.5x10
2 - 1.2x10
5), 5.6x10
4 (9.4x10
3 - 4.1x10
5), and 2.1x10
4 (7x10
3 -
5.8x104) day.m
3/kg for Cu, Ni and Zn, respectively. In comparison, CFs of Huijbregts et al.
(2000) were 1200, 3200 and 92 eq. 1-4DCB or when converted to the same unit 1-3 orders of
magnitude higher than those calculated here for all metals in the EU water-types. However,
CTPs calculated using the most recent version of USEtoxTM
(released online January, 2010)
that considers chemical binding to DOC were 5.5x104, 1.5x10
4 and 3.9x10
4 day.m
3/kg, or
within a factor of two of CTPs calculated here.
Extreme low CTPs of Cu, Ni and Zn came from EU water-types 1 and 2 (medium/high DOC,
high pH and hardness), while high CTPs were calculated for EU water-type 9 (medium pH
and low DOC and hardness). Further, the relative ranking of CTPs for three metals differed
for several of the EU water-types (Figure 2.4).
44
EU Water-type 12
EU Water-type 11
EU Water-type 10
EU Water-type 9
EU Water-type 8
EU Water-type 7
EU Water-type 6
EU Water-type 5
EU Water-type 4
EU Water-type 3
EU Water-type 2
EU Water-type 1
ZnNiCu
EU Water-type 12
EU Water-type 11
EU Water-type 10
EU Water-type 9
EU Water-type 8
EU Water-type 7
EU Water-type 6
EU Water-type 5
EU Water-type 4
EU Water-type 3
EU Water-type 2
EU Water-type 1
ZnNiCu
Figure 2.4: Comparison of metal ranking according to values of Comparative Toxicity
Potentials (CTPs; day.m3/kg) calculated for the 12 EU water-types. The lowest value of CTP
among three metals within one water-type represents the lowest concern (or relative hazard)
and vice versa. The relative ranking is displayed as hatched for the lowest, brick for medium
and dotted pattern for the highest.
For the same range of variations in Kd values, FFs were within a factor of two but were up to
two orders of magnitude different for EFs, indicating that toxicity was much more sensitive
to water chemistry than fate. Above all, the range in BF was the greatest for all metals and
water-types. The values of coefficient of variance (CV) between the water-types and metals
ranged 0.1-0.4 for FFs, 0.7-1.5 for BFs, 0.5-1.3 for EFs, and 0.7-1.3 for CTPs (Table 2.3).
There is a negative covariance (-0.67) between BF and EF of Copper. Further, values of CV
ranged 0.5-1.1 amongst three metals. These results imply that (1) BF and EF indeed have the
45
largest influence on CTP, (2) that the consistent use of water chemistry values for FF, BF and
EF is particularly important for Cu, and (3) that the variability in CTPs between the water-
types is as large as the variability between metals investigated.
Table 2.3: Estimated Bioavailability Factors (BFs, dimensionless), Fate Factors (FFs, days),
Effect Factors (EFs, m3/kg) and Comparative Ecotoxicity Potentials (CTPs, day.m
3/kg) for
Cu, Ni and Zn for the 12 EU water types listed in Table 2.1. Coefficients of variance (CV)
are reported for each modelled parameter among water-types and metals. Note that FF
represents residence time of total metal in freshwater after its unit (1 kg/day) emission to
freshwater compartment.
Water-types Cu Ni Zn Metals
BF FF EF CTP BF FF EF CTP BF FF EF CTP CV
EU Water-type 1 9.69E-6 40.4 3.47E+5 1.36E+2 5.89E-2 42.6 3.91E+3 9.81E+3 6.63E-2 27.1 3.93E+3 7.07E+3 0.88
EU Water-type 2 4.38E-6 45.0 1.40E+6 2.76E+2 5.15E-2 52.0 3.49E+3 9.36E+3 7.14E-2 15.8 6.55E+3 7.41E+3 0.84
EU Water-type 3 3.32E-5 45.5 6.57E+5 9.94E+2 1.14E-1 51.4 5.38E+3 3.15E+4 1.32E-1 16.8 5.75E+3 1.27E+4 1.02
EU Water-type 4 1.04E-4 49.1 1.97E+6 1.01E+4 1.85E-1 74.0 4.21E+3 5.78E+4 3.19E-1 10.7 6.77E+3 2.32E+4 0.81
EU Water-type 5 1.06E-4 48.5 4.95E+6 2.54E+4 1.45E-1 80.1 4.73E+3 5.47E+4 3.09E-1 10.4 7.83E+3 2.51E+4 0.49
EU Water-type 6 1.74E-5 44.6 7.78E+6 6.04E+3 5.91E-2 56.8 8.64E+3 2.90E+4 9.79E-2 11.5 1.36E+4 1.53E+4 0.69
EU Water-type 7 1.17E-5 41.2 6.16E+5 2.97E+2 5.37E-2 42.5 4.80E+3 1.09E+4 6.03E-2 27.8 4.65E+3 7.79E+3 0.86
EU Water-type 8 1.14E-3 50.3 4.96E+5 2.84E+4 4.62E-1 67.5 6.81E+3 2.12E+5 5.24E-1 19.0 4.29E+3 4.27E+4 1.08
EU Water-type 9 3.14E-3 56.8 6.42E+5 1.15E+5 5.20E-1 75.7 1.03E+4 4.05E+5 5.73E-1 21.5 4.72E+3 5.83E+4 0.97
EU Water-type 10 1.76E-3 49.0 5.55E+5 4.80E+4 2.66E-1 49.5 1.41E+4 1.86E+5 3.19E-1 35.0 3.36E+3 3.74E+4 0.92
EU Water-type 11 5.68E-4 53.7 6.66E+5 2.03E+4 1.57E-1 51.8 1.36E+4 1.11E+5 2.04E-1 20.2 4.39E+3 1.80E+4 1.06
EU Water-type 12 9.37E-4 52.2 8.12E+5 3.97E+4 2.86E-1 56.0 1.36E+4 2.18E+5 3.56E-1 14.6 5.62E+3 2.91E+4 1.11
CV = SD/MEAN 1.49 0.10 1.31 1.34 0.81 0.22 0.53 1.10 0.70 0.40 0.46 0.67
2.6 Practical Implications
Current LCIA practice has been to use generic environmental data to typify a single
“evaluative environment”, i.e., one water-type for freshwater. However, we see that CTPs,
and the relative ranking amongst metals, are a product of that water-type. Thus, one’s choice
of a freshwater chemistry has an equally important influence on the CTPs compared to the
inherent differences in chemical properties, e.g., Kd-values, average toxicity in case of
46
metals. The dependence of metal CTP on both extrinsic freshwater chemistry and intrinsic
chemical properties differs from that of organic chemicals for which only intrinsic chemical
properties control CTP. This independence on extrinsic properties has not been tackled in
the spatially-generic LCA studies to date.
The consequences of these results raise yet more questions and issues. First, are these results
applicable to other metals and freshwater environments? Second, for geographically generic
LCA and chemical hazard assessment, which water chemistry and corresponding values of
Kd, BF and EF should be selected as a default? Third, the variability in results demands
developing new methods of coupling LCI information of metal emissions with the up-to-date
CTPs of metals in spatially-explicit contexts. Finally, LCI must report the species and
particle sizes of metals released into the environment. Since most particulate forms of metals
emitted undergo a slow dissolution process, the time horizon considered for the dissolution
process and the influence of metal mineralization on long-term bioavailability are topics of
further research.
2.7 References
Adams, W. J.; Conard, B.; Ethier, G.; Brix, K. V.; Paquin, P. R.; DiToro, D. M. The
challenges of hazard identification and classification of insoluble metals and metal
substances for the aquatic environment. Hum. Ecol. Risk Assess. 2000, 6, 1019-1038.
Adams, W. J.; Chapman, P. M. Assessing the hazard of metals and inorganic metal
substances in aquatic and terrestrial systems: summary of a SETAC Pellston
workshop; SETAC: Pensacola, FL, 2005.
Bhavsar, S. P.; Diamond, M. L.; Evans, L. J.; Gandhi, N.; Nilsen, J.; Antunes, P.
Development of a coupled metal speciation-fate model for surface aquatic systems.
Environ. Toxicol. Chem. 2004, 23, 1376-1385.
Bhavsar, S. P.; Gandhi, N.; Diamond, M. L. Extension of coupled multispecies metal
transport and speciation (TRANSPEC) model to soil. Chemosphere 2008a, 70, 914-
924.
47
Bhavsar, S. P.; Gandhi, N.; Diamond, M. L.; Lock, A. S.; Spiers, G.; De la Torre, M. C. A.
Effects of estimates from different geochemical models on metal fate predicted by
coupled speciation-fate models. Environ. Toxicol. Chem. 2008b, 27, 1020-1030.
Bryan, S. E.; Tipping, E.; Hamilton-Taylor, J. Comparison of measured and modelled copper
binding by natural organic matter in freshwaters. Comparative Biochemistry and
Physiology Part C 2002, 133, 37-49.
Campbell, P. G. C. Interactions between Trace Metals and Aquatic Organisms: A Critique of
the Free-Ion Activity Model. In Metal Speciation and Bioavailability in Aquatic
Systems; Tessier, A., Turner, D. R., Eds.; John Wiley: New York, 1995; Vol. 1, pp
45-102.
Chapman, P. M.; Wang, F. Y. Issues in ecological risk assessment of inorganic metals and
metalloids. Hum. Ecol. Risk Assess. 2000, 6, 965-988.
Cheng, T.; De Schamphelaere, K.; Lofts, S.; Janssen, C.; Allen, H. E. Measurement and
computation of zinc binding to natural dissolved organic matter in European surface
waters. Anal. Chim. Acta. 2005, 542, 230-239.
Cissel, D.; Cromwell, E. F. Office of Pollution Prevention and Toxics: Program Activities for
Fiscal Years 1998 and 1999; U.S. Environmental Protection Agency, Office of
Prevention, Pesticides and Toxic Substances: Washington, DC, 1999.
Deleebeeck, N. M. E.; De Schamphelaere, K. A. C.; Heijerick, D. G.; Bossuyt, B. T. A.;
Janssen, C. R. The acute toxicity of nickel to Daphnia magna: Predictive capacity of
bioavailability models in artificial and natural waters. Ecotoxicol. Environ. Safety
2008, 70, 67-78.
Deleebeeck, N. M. E.; De Laender, F.; Chepurnov, V. A.; Vyverman, W.; Janssen, C. R.; De
Schamphelaere, K. A. C. A single bioavailability model can accurately predict Ni
toxicity to green microalgae in soft and hard surface waters. Water Res. 2009, 43,
1935-1947.
48
De Schamphelaere, K. A. C.; Vasconcelos, F. M.; Heijerick, D. G.; Tack, F. M. G.; Delbeke,
K.; Allen, H. E.; Janssen, C. R. Development and field validation of a predictive
copper toxicity model for the green alga pseudokirchneriella subcapitata. Environ.
Toxicol. Chem. 2003, 22, 2454-2465.
De Schamphelaere, K. A. C.; Janssen, C. R. Development and field validation of a biotic
ligand model predicting chronic copper toxicity to Daphnia magna. Environ. Toxicol.
Chem. 2004, 23, 1365-1375.
De Schamphelaere, K. A. C.; Janssen, C. R. European Union Risk Assessment Report.
Copper, Copper II Sulphate Pentahydrate, Copper(I) Oxide, Copper(II) Oxide,
Dicopper Chloride Trihydroxide.; Report submitted to European Copper Institute
(ECI): Brussels, Belgium 2005.
De Schamphelaere, K. A. C.; Lofts, S.; Janssen, C. R. Bioavailability models for predicting
acute and chronic toxicity of zinc to algae, daphnids, and fish in natural surface
waters. Environ. Toxicol. Chem. 2005, 24, 1190-1197.
Diamond, M. L.; Mackay, D.; Cornett, R. J.; Chant, L. A. A model of the exchange of
inorganic chemicals between water and sediments. Environ. Sci. Technol. 1990, 24,
713-722.
Diamond, M. L.; Mackay, D.; Welbourn, P. M. Models of multimedia partitioning of
multispecies chemicals - the fugacity equivalence approach. Chemosphere 1992, 25,
1907-1921.
Diamond, M. L.; Gandhi, N.; Adams, W. J.; Atherton, J.; Bhavsar, S. P.; Bulle, C.;
Campbell, P. G. C.; Dubreuil, A.; Fairbrother, A.; Farley, K.; Green, A.; Guinee, J.
B.; Hauschild, M. Z.; Huijbregts, M. A. J.; Humbert, S.; Jensen, K. S.; Jolliet, O.;
Margni, M.; McGeer, J. C.; Peijnenburg, W. J. G. M.; Rosenbaum, R.; van de Meent,
D.; Vijver, M. G. The Clearwater consensus: the estimation of metal hazard in fresh
water. Int. J Life Cycle Assess. 2010, 15, 143-147.
49
Di Toro, D. M.; Allen, H. E.; Bergman, H. L.; Meyer, J. S.; Paquin, P. R.; Santore, R. C.
Biotic ligand model of the acute toxicity of metals. 1. Technical basis. Environ.
Toxicol. Chem. 2001, 20, 2383-2396.
European Commission. EUSES, the European Union System for the Evaluation of
Substances; European Chemicals Bureau: Ispra, Italy, 1996.
Fairbrother, A. Risk assessment: Lessons learned. Environ. Toxicol. Chem. 2002, 21, 2261-
2263.
Gandhi, N.; Bhavsar, S. P.; Diamond, M. L.; Kuwabara, J. S.; Marvin-Dipasquale, M.;
Krabbenhoft, D. P. Development of a mercury speciation, fate, and biotic uptake
(BIOTRANSPEC) model: application to Lahontan Reservoir (Nevada, USA).
Environ. Toxicol. Chem. 2007, 26, 2260-2273.
Gloria, T. P.; Russell, A. J.; Atherton, J.; Baker, S. R.; Cook, M. Ecological toxicity methods
and metals: An examination of two case studies. Int. J Life Cycle Assess. 2006, 11,
26-33.
Government of Canada. Canadian Environmental Protection Act (CEPA); Ottawa, ON,
Canada, 1999.
Harvey, C.; Mackay, D.; Webster, E. Can the unit world model concept be applied to hazard
assessment of both organic chemicals and metal ions? Environ. Toxicol. Chem. 2007,
26, 2129-2142.
Hauschild, M. Z. Assessing environmental impacts in a life-cycle perspective. Environ. Sci.
Technol. 2005, 39, 81A-88A.
Heijerick, D. G.; Bossuyt, B. T. A.; de Schamphelaere, K.; Indeherberg, M.; Mingazzini, M.;
Janssen, C. R. Effect of varying physicochemistry of European surface waters on the
copper toxicity to the green alga Pseudokirchneriella subcapitata. Ecotoxicology
2005, 14, 661-670.
50
Huijbregts, M. A. J.; Thissen, U.; Guinee, J. B.; Jager, T.; Kalf, D.; van de Meent, D.; Ragas,
A. M. J.; Sleeswijk, A. W.; Reijnders, L. Priority assessment of toxic substances in
life cycle assessment. Part I: Calculation of toxicity potentials for 181 substances with
the nested multi-media fate, exposure and effects model USES-LCA. Chemosphere
2000, 41, 541-573.
Jolliet, O.; Margni, M.; Charles, R.; Humbert, S.; Payet, J.; Rebitzer, G.; Rosenbaum, R.
IMPACT 2002+: A new life cycle impact assessment methodology. Int. J. Life Cycle
Assess. 2003, 8, 324-330.
Niyogi, S.; Wood, C. M. Biotic ligand model, a flexible tool for developing site-specific
water quality guidelines for metals. Environ. Sci. Technol. 2004, 38, 6177-6192.
Paquin, P. R.; Di Toro, D. M.; Mathew, R.; Santori, R. C.; Wu, K. B. Consideration of metal
availability and fate in life-cycle assessments. International Workshop on Life-Cycle
Assessment and Metals 2002, 248-257.
Payet, J.; Jolliet, O. Comparative assessment of the toxic impact of metals on aquatic
ecosystems: The AMI method. International Workshop on Life-Cycle Assessment
and Metals 2002, 188-191.
Peijnenburg, W. J. G. M.; Jager, T. Monitoring approaches to assess bioaccessibility and
bioavailability of metals: Matrix issues. Ecotoxicol. Environ. Safe. 2003, 56, 63-77.
Pennington, D. W.; Payet, J.; Hauschild, M. Aquatic ecotoxicological indicators in life-cycle
assessment. Environ. Toxicol. Chem. 2004, 23, 1796-1807.
Pettersen, J.; Hertwich, E. G. Critical review: life-cycle inventory procedures for long-term
release of metals. Environ. Sci. Technol. 2008, 42, 4639-4647.
Rosenbaum, R. K.; Bachmann, T. M.; Gold, L. S.; Huijbregts, M. A. J.; Jolliet, O.; Juraske,
R.; Koehler, A.; Larsen, H. F.; MacLeod, M.; Margni, M.; McKone, T. E.; Payet, J.;
Schuhmacher, M.; van de Meent, D.; Hauschild, M. Z. USEtox-the UNEP-SETAC
toxicity model: recommended characterisation factors for human toxicity and
51
freshwater ecotoxicity in life cycle impact assessment. Int. J. Life Cycle Assess. 2008,
13, 532-546.
Schecher, W. D.; McAvoy, D. C. MINEQL+ - A software environment for chemical-
equilibrium modelling. Comp. Env. Urban Sys. 1992, 16, 65-76.
Skeaff, J.; Delbeke, K.; Van Assche, F.; Conard, B. A critical surface area concept for acute
hazard classification of relatively insoluble metal-containing powders in aquatic
environments. Environ. Toxicol. Chem. 2000, 19, 1681-1691.
Strandesen, M.; Birkved, M.; Holm, P. E.; Hauschild, M. Z. Fate and distribution modelling
of metals in life cycle impact assessment. Ecol. Model. 2007, 203, 327-338.
Tipping, E. Humic ion-binding model VI: An improved description of the interactions of
protons and metal ions with humic substances. Aquat. Geochem. 1998, 4, 3-48.
Toose, L. K.; Mackay, D. Adaptation of fugacity models to treat speciating chemicals with
constant species concentration ratios. Environ. Sci. Technol. 2004, 38, 4619-4626.
van de Meent, D.; Huijbregts, M. A. J. Calculating life-cycle assessment effect factors from
potentially affected fraction-based ecotoxicological response functions. Environ.
Toxicol. Chem. 2005, 24, 1573-1578.
Van Tilborg, W. J. M. Natural background/ambient concentrations of metals and abiotic
conditions of fresh surface waters in relation to risk assessment of metals; WTBC:
2002.
Vijver, M. G.; de Koning, A.; Peijnenburg, W. Uncertainty of water type-specific hazardous
copper concentrations derived with Biotic Ligand Models. Environ. Toxicol. Chem.
2008, 27, 2311-2319.
Wetzel, R. G. Limnology; Saunders College Publishing: Philadelphia, 1983; Vol. 2;
Complete Revision.
52
3. Implications of Geographic Variability on Comparative Toxicity
Potentials of Cu, Ni and Zn in Freshwaters of Canadian Ecoregions©
3.1 Abstract
Current methods of estimating potential environmental impacts of metals in hazard and Life
Cycle Impact Assessment (LCIA) do not consider differences in chemistry and landscape
properties between geographic sites. Here, we developed and applied a model for regional
aquatic impact characterization of metals using an updated method for estimating
environmental fate factor (FF), bioavailability factor (BF) and aquatic ecotoxicity factor
(EF). We applied the model to analyze differences in Comparative Toxicity Potentials
(CTPs) of Cu, Ni and Zn for 24 Canadian ecoregions. The combined impacts of regional
variability in ambient chemistry (in particular DOC, pH and hardness) and landscape
properties (water residence time) can change the CTPs of these metals for freshwater by up
to three orders of magnitude and change the relative ranking of metal hazard between
ecoregions. Variation among Canadian freshwater chemistries and landscape characteristics
influence the FFs within two orders of magnitude, BFs within two orders of magnitude for Ni
and Zn and four orders of magnitude for Cu, and EFs within one order of magnitude.
Sensitivity of metal FFs to environmental parameters alone spans three orders of magnitude
when a constant water chemistry was used for all ecoregions. These results indicate that
application of regionalised metal CTPs can have a significant influence in the analysis of
ecotoxicological impacts in the life cycle assessment of products and processes.
© Contents of this chapter have been adopted from the publication in the Chemosphere:
Gandhi N., Huijbregts M.A.J., van de Meent D., Peijnenburg W.J.G.M., Guinée J. and Diamond M.L. (2011)
Implications of geographic variability on Comparative Toxicity Potentials of Cu, Ni and Zn in freshwaters of
Canadian ecoregions. Chemosphere 82: 268–277.
A link to the published paper can be found at linkinghub.elsevier.com/retrieve/pii/S0045653510010611
I was primarily responsible for the model development, data collection, model application, testing the
sensitivity of model parameters, analyis of model results, and writing of this manuscript.
53
3.2 Introduction
Within Life Cycle Assessment (LCA), Life Cycle Impact Assessment (LCIA) translates the
inventory of emissions from a product’s or process’ life cycle (from resource acquisition
through use and finally disposal or end-of-life management) into an environmental profile of
the product representing its potential contributions to a wide range of environmental impacts.
The method used to translate an emission into an impact relies on a substance-specific
Characterization Factor (CF), which is also known as Comparative Toxicity Potential (CTP)
for ecotoxicity impact category (Gandhi et al. 2010). Generic CFs have been developed
independently of spatial and temporal information because of LCA’s original goal of
expressing the potential, incremental environmental burden associated with the defined
functional unit of the product or process rather than an actual burden. These generic CFs
evaluate global impacts, such as global warming and ozone layer depletion, but there can be
large variations in potential environmental burdens for impact categories like acidification or
eutrophication that depend strongly on the receiving environment. Thus, Potting and
Hauschild (2006) concluded that the exclusion of spatial information can lead to erroneous
results in LCIA characterization.
To deal with this issue, several approaches for including spatial differentiation have been
proposed (e.g., Nigge 2001, Sleeswijk 2003; 2006). One approach is to use spatial
differentiation, which refers to incorporating important geographical features of continents,
countries, or regions throughout the world. Depending on impact categories and type of
chemicals, some variations in geographical features may justify a further differentiation to
archetypical situations within a region. The latter approach requires defining the
characteristics of a location or the type of environment into which the inventory flow occurs,
e.g., chemistry of a lake into which a chemical is discharged.
Several studies have addressed spatial differentiation in LCIA for the impact categories of
acidification and eutrophication (Potting et al. 1997, Huijbregts and Seppälä 2001, Hettelingh
et al. 2005, Seppälä et al. 2006), human and ecotoxicity (Pennington et al. 2005, Bare 2006,
Sleeswijk 2006, Humbert et al. 2009), respiratory effects caused by primary and secondary
particles (Humbert and Horvath 2006), and photochemical smog formation (Hauschild et al.
54
2006). The scale of spatial differentiation can vary from continental (Huijbregts et al. 2000;
Rosenbaum et al. 2008) to regional (Bare et al. 2003; 2006, Toffoletto et al. 2007) resolution.
The conclusions drawn from these studies are that spatial differentiation in LCA requires
data availability, methods and modelling tools, computational affordability, and ease of
interpreting results. Depending on the need and availability of information to carry out LCA,
and as mentioned above, two distinct types of spatial models have emerged: LCIA methods
proposing spatial differentiation in terms of archetypes (e.g., USES-LCA in Huijbregts et al.
2000, USEtoxTM
in Rosenbaum et al. 2008) and those based on GIS or political regions (e.g.,
GLOBOX in Sleeswijk 2006, RAINS in Schöpp et al. 1999, TRACI in Bare et al. 2003,
LIME in Itsubo and Inaba 2003). Both types of models can be connected or combined in one
model such that within countries one can distinguish archetypes (e.g., shield versus prairie
lakes in Canada) and within archetypes one can distinguish countries (e.g., the Rhine river
basin within Switzerland, Germany and Netherlands).
Several LCA studies have been conducted by Canadian industries. Most of the Canadian
LCA studies use generic CFs (Godin et al. 2004, Menard et al. 2004, Toffoletto et al. 2005)
calculated using either European- USEtoxTM
(Rosenbaum et al. 2008) or American-TRACI
(Bare et al. 2003) models. TRACI has the advantage of considering all North-American
territory in some of its deposition models (Bare et al. 2003). As a consequence, some of the
site-specific factors are not necessarily appropriate to the Canadian context.
Two regional models, ChemCAN (Webster et al. 2004) and LUCAS (Toffoletto et al. 2007),
have been developed specifically for the Canadian environment. ChemCAN is designed for
the hazard assessment of chemicals, whereas LUCAS estimates CFs for 10 different impact
categories of LCIA. Both models divide Canada into ecozones (comparable to geographic
archetypes in LCIA) that represent large ecological units, a spatial resolution intended to
characterize regional and, if data are available, local impacts. This level of spatial resolution
has the advantage of facilitating correspondence between site-dependant LCI results and site-
dependant LCIA models. Such models also allow the examination of uncertainty due to the
spatial variability in generic values of CFs. For example, LUCAS was used to complete a
LCIA of a contaminated site, where the authors showed significant differences between
55
several generic impact factors and those developed for Canadian environments (Toffoletto et
al. 2007).
All of these models (USEtoxTM
, TRACI, ChemCAN and LUCAS) have been most
extensively developed for organic compounds that exist as single chemical species. When
used for metals, the models fail to account for the existence of multiple, interconverting
species and the sensitivity of species distributions to ambient chemistry. Gandhi et al. (2010)
demonstrated the importance of considering chemistry-dependent speciation when
calculating CTPs of cationic metals for aquatic environment. Using 12 EU freshwater-types,
they found that considering the influence of water chemistry is most important for assessing
metals’ bioavailability and toxicity but, to a lesser extent, fate.
The goal of this work was to incorporate recent improvements in assessing the
ecotoxicological impacts of metals, in the regional differentiation of hazard analysis and
LCIA of chemicals. We used USEtoxTM
(Rosenbaum et al. 2008), a consensus LCIA model,
to which we added the method of calculating CTPs of metals proposed by Gandhi et al.
(2010) to calculate transport and transformation of Cu, Zn and Ni emitted to the 24
ecoregions of Canada defined in ChemCAN (Webster et al. 2004). Each ecoregion was
assigned a water-type, which represents generic chemistry for freshwaters within that
ecoregion based on the available measurements. An additional water-type was also assigned
to the entire Canadian geographic in order to analyze the differences in estimates of CTPs
should the regional details are omitted in LCIA studies. We also compared the model
estimates with previously reported CTPs for use in LCIA and analyzed the sensitivity of
model results to limnological characteristics such as water residence time, background metal
concentrations and concentrations of total suspended solid (TSS) that affect adsorption and
net sedimentation of metals.
3.3 Methods
3.3.1 Modelling Framework
The model is based on the following definitions that are recommended in the Clearwater
Consensus (Diamond et al. 2010) concerning metal aquatic ecotoxicity assessment for use in
56
LCIA. The amount of total chemical is the sum of its amount in particulate and total
dissolved phases, where the total dissolved phase is the sum of colloidal, mainly associated
with Dissolved Organic Carbon (DOC), and truly dissolved fractions (refer to Figure 1 in
Diamond et al. 2010).
CTPs have been developed for the total metal emitted into the environment (reported in LCI
in ‘‘elementary” form). In the past, CTP was calculated as the product of a fate factor (FF)
and an effects factor (EF) representing fate and potential toxicity of a chemical, respectively
(Huijbregts et al. 2000). The FF was calculated as the residence time of the total dissolved
fraction of the chemical in the environment, whereas the EF was calculated for the total
dissolved fraction, which was assumed to be bioavailable. The distribution between
particulate and total dissolved forms was estimated using the particle-dissolved distribution
coefficient (Kd). Gandhi et al. (2010) revised the method for calculating the CTP for all
chemicals, and metals in particular, in terms of the bioavailable fraction of chemical, which
for organics and metals is defined as the truly dissolved fraction. They introduced a
bioavailability factor (BF) that explicitly quantifies the relationship between total dissolved
and bioavailable fractions of a chemical. Their definitions of BF and CTP can be extended to
all environmental compartments (e.g., soil and sediment); however, our discussion focuses
on freshwater systems. According to the method of Gandhi et al. (2010), the CTP for
freshwater is calculated as:
[3.1] sssi,si, EFBFFFCTP ⋅⋅=
where CTPi,s (day.m3/kg) is the ecotoxicity potential of substance s emitted to compartment i,
FFi,s (day) is the fate factor, BFs (dimensionless) is the bioavailability factor, and EFs
(m3/kg) is the effect factor. Here an emission of substance s may be to any compartment i
(e.g., soil or sediment) including freshwater.
The compartment-specific FF is defined as the change in steady-state total concentration of a
substance in the environmental compartment due to the marginal change in emission in the
same or other connected compartment (Gandhi et al. 2010). For the freshwater:
57
[3.2] s,i
s,t
s,im
VCFF
∆
∆=
where FFi,s (day) accounts for the transport efficiency of a substance from compartment i to
freshwater, in case where the substance is emitted to other compartment than freshwater, as
well as its persistence in freshwater, ∆Ct,s (kg/m3) is the marginal change in the steady-state
concentration of the total t (total dissolved and particulate) substance s in freshwater, V (m3)
is the volume of the freshwater compartment, and ∆mi,s (kg/day) is the marginal change in the
emission rate of substance s to compartment i.
The BF (dimensionless) of a substance s is calculated as (Gandhi et al. 2010):
[3.3] s,t
s,k
sC
CBF
∆
∆=
where ∆Ck,s (kg/m3) is the marginal change in the bioavailable concentration k. BF was
consensually defined as truly dissolved fraction for metals ((Diamond et al. 2010). For
freshwaters, BF can be estimated using geochemical speciation models like Windermere
Humic Aqueous Model (WHAM; Tipping 1998) or MINEQL+ (Schecher and McAvoy
1992).
Consistent with the definition of bioavailability, the EF (m3/kg) of substance s for freshwater
is calculated based on truly dissolved concentrations (Gandhi et al. 2010):
[3.4] s10
5.0
C
PAFEF
s,k
s µ≈
∆
∆=
where ∆PAF is the marginal change in the potentially affected fraction of species due to
exposure to a toxic substance in the freshwater compartment, and 10µ (kg/m
3) is the
geometric mean of toxicity data for substance s. The EF of metals can be calculated using
models like Biotic Ligand Model (BLM; Di Toro et al. 2001) or Free Ion Activity Model
(FIAM; Campbell 1995). The BLM enables to calculate a value of µ specific to a water
chemistry assuming that a toxic effect in aquatic organism is caused by the free metal ion that
binds with a biotic ligand in competition with other cations, considering the chemistry of the
58
water. In contrast, µ is calculated using FIAM by assuming that a fixed activity of free metal
ion in water causes a toxic effect in case of cationic metals. Both models are capable of
calculating µ in terms of k (truly dissolved concentration) that is used in the definition of the
EF.
In USEtoxTM
, the value of 10µ (also referred as 10
µ(HC50) or HC50-EC50) is calculated using
values of EC50 (concentration of substance s causing adverse effect(s) in 50% of the exposed
organisms) from chronic ecotoxicity tests for multiple freshwater biotic species (Rosenbaum
et al. 2008). The HC50-EC50 is equivalent to the HC50 obtained from a log-normal Species
Sensitivity Distribution of EC50 (SSDEC50). In the absence of at least three values of chronic
EC50, the HC50 can be calculated using acute EC50 based on the correspondence between
acute and chronic test results (Rosenbaum et al. 2008). A factor incorporating typical acute-
to-chronic ratios should be included in this case.
3.4 Model Selection and Parameterization
3.4.1 Fate
We adapted USEtoxTM
to calculate FF for each ecoregion (Hauschild et al. 2008, Rosenbaum
et al. 2008) by incorporating landscape and transport parameters of 24 ecoregions of the
ChemCAN model (Table 3.1). These regional divisions of Canada were based on the
ecozones identified by Environment Canada, with consideration of the distribution of
population and industrial activity, political boundaries, drainage basins, and climate to give
areas of homogeneous ecological conditions (Webster et al. 2004). We kept the existing
formulations in USEtoxTM
to calculate the transport and transformation of chemicals.
USEtoxTM
uses a simplified formulation of net sedimentation that does not account for
sediment-to-water diffusive release of soluble metal resulting from post-diagenetic fate
processes. Although this assumption may underestimate chemical contribution from
sediment to water, the diffusive release of metals is generally low compared to other fate
processes (Alfaro-De la Torre and Tessier 2002, Bhavsar et al. 2004) except in cases where
there is historical accumulation of metals in sediment (e.g., Diamond 1995). The approach is
59
reasonable for LCIA since the method assesses the added environmental impacts of a
chemical emission and not the contribution from the historical deposition of that chemical.
Table 3.1: Summary of model parameters used to calculate Fate Factors (or residence times)
of metals in freshwater compartment of 24 ecoregions of Canada.
Ecoregion Ecoregion Description
Area
Total
Area
Freshwater
Area of
Natural
Soil
Area of
Agricultural
Soil
Area
Water
Volume
Air Temp
Wind
Speed
Freshwater
Advection
Residence Time Rain rate
Soil water
runoff
rate
Soil
erosion
rate
km2
% % % km2
km3 o
C m/s h m/h m/h m/h
1 Newfoundland 109700 1.7 0.49 0.49 1864.9 219400 4.8 5.9 3600 1.4E-04 5.6E-05 2.8E-08
2 Labrador 282400 6.7 0.47 0.47 18920.8 564800 -2.3 4.2 14160 1.1E-04 4.4E-05 2.2E-08
3 Atlantic Maritime 180900 0.8 0.50 0.50 1447.2 361800 5.5 4.5 1920 1.4E-04 5.6E-05 2.8E-08
4 Quebec - Mixed Wood Plain 85140 1.7 0.49 0.49 1447.4 170280 5.0 3.6 10560 1.2E-04 4.9E-05 2.5E-08
5 Quebec - Boreal Shield Region 629400 5.3 0.47 0.47 33358.2 1258800 0.8 3.5 14880 1.1E-04 4.4E-05 2.2E-08
6 Quebec - Northern Region 747200 5.5 0.47 0.47 41096.0 1494400 -5.1 4.7 16560 8.0E-05 3.2E-05 1.6E-08
7 Ontario - Mixed Wood Plain 142700 32.0 0.34 0.34 45664.0 285400 7.4 4.2 26160 1.0E-04 4.2E-05 2.1E-08
8 Ontario - Boreal Shield 657300 6.2 0.47 0.47 40752.6 1314600 -0.1 3.6 35280 8.2E-05 3.3E-05 1.6E-08
9 Ontario - Northern 231300 0.6 0.50 0.50 1387.8 462600 -2.0 3.9 3840 8.0E-05 3.2E-05 1.6E-08
10 Manitoba - Prairie 96010 4.4 0.48 0.48 4224.4 192020 1.8 4.6 106560 5.9E-05 2.4E-05 1.2E-08
11 Manitoba - Boreal Shield 331700 18.3 0.41 0.41 60701.1 663400 -1.6 3.9 185280 5.8E-05 2.3E-05 1.2E-08
12 Manitoba - Northern 220500 5.8 0.47 0.47 12789.0 441000 -4.7 4.3 58560 5.2E-05 2.1E-05 1.0E-08
13 Saskatchewan - Prairie 236800 1.4 0.49 0.49 3315.2 473600 3.7 5.0 125280 4.4E-05 1.8E-05 8.8E-09
14 Saskatchewan - Northern 412200 10.7 0.45 0.45 44105.4 824400 -1.8 3.3 154320 5.1E-05 2.0E-05 1.0E-08
15 Alberta - Prairie 247800 1.0 0.50 0.50 2478.0 495600 5.0 4.4 24720 5.3E-05 2.1E-05 1.1E-08
16 Alberta - Northern 404900 2.9 0.49 0.49 11742.1 809800 0.8 3.2 29040 5.0E-05 2.0E-05 1.0E-08
17 B.C. - Montane Cordillera 389700 2.3 0.49 0.49 8963.1 779400 6.0 3.4 6720 6.5E-05 2.6E-05 1.3E-08
18 B.C. - South Pacific Maritime 80190 1.8 0.49 0.49 1443.4 160380 7.7 3.1 3600 2.0E-04 8.0E-05 4.0E-08
19 B.C. - North Pacific Maritime 158200 1.2 0.49 0.49 1898.4 316400 7.8 4.3 1200 1.4E-04 5.6E-05 2.8E-08
20 B.C. - Northern Region 316600 2.1 0.49 0.49 6648.6 633200 0.9 3.1 6240 5.5E-05 2.2E-05 1.1E-08
21 Yukon Territory 481000 0.4 0.50 0.50 1924.0 962000 -5.2 3.0 3360 7.2E-05 2.9E-05 1.4E-08
22 Mackenzie River Valley 794700 8.0 0.46 0.46 63576.0 1589400 -6.4 2.9 83760 3.6E-05 1.4E-05 7.2E-09
23 Northwest Territories 426300 11.8 0.44 0.44 50303.4 852600 -9.7 3.7 109680 3.2E-05 1.3E-05 6.4E-09
24 Arctic and Subarctic 2176000 2.8 0.49 0.49 60928.0 4352000 -12.7 4.5 30000 2.5E-05 1.0E-05 5.0E-09
The fate calculations relied on the values of Kd and metal species fractions in freshwater that
were calculated using the chemistry of each ecoregion in speciation/complexation model as
described below. The net exchange of metals between water and sediment was modelled
using the fixed values of LogKd for sediments in all ecoregions, which were set at 3.5, 3.9
and 4.1 for Cu, Ni and Zn, respectively (Allison and Allison 2005). This is in line with
simplified treatment of the sediment compartment in USEtoxTM
. An improved alternative
would be to calculate LogKd values for sediment using the measured chemistry of sediment
and pore water (e.g., pH, organic matter and sulphides) and geochemical adsorption models
that incorporate precipitation of metal sulfides in anoxic environments.
60
3.4.2 Speciation/Complexation
To calculate Kd values, aqueous metal species fractions and BF of a metal for water
compartment, we used the Windermere Humic-Aqueous Model (WHAM 6.0; Tipping 1998).
WHAM 6.0 is an equilibrium based metal speciation/complexation model that includes the
Humic Ion-Binding Model VI and an inorganic speciation code for aqueous solutions. The
model calculates metal distribution in total dissolved and solid phases that are then used to
calculate Kd (L/kg). The model structure is designed to evaluate dissolved metal species and
not non-reactive native metal products such as copper ingots or insoluble minerals.
Calculations in WHAM 6.0 are completed “off line” and then the results are coupled with the
fate and ecotoxicity models.
3.4.3 Ecotoxicity
To calculate EFs, the metal ecotoxicity assessment assumes that toxicity is a function of the
free metal ion activity in solution (Campbell 1995) or more specifically a fraction of free
metal ion binding to the biotic ligand in competition with other cations present in water (Di
Toro et al. 2001). We used BLMs to calculate the concentration of metal at a specified
chemistry that has potential to cause an adverse effect. This approach assumes that metal
uptake in aquatic organisms is directly from water and does not address dietary uptake. We
also used WHAM 6.0 to estimate metal speciation for toxicity test waters and water-types for
ecoregions so that the speciation calculations would be comparable for both fate and toxicity
assessments.
3.4.4 Model Parameters
Several factors, such as water pH, DOC and hardness, control metal partitioning and
speciation in freshwaters and thus affect metal toxicity to aquatic organisms (e.g., Bryan et
al. 2002, Heijerick et al. 200,; Allen and Janssen 2006). The Clearwater Consensus
(Diamond et al. 2010) recommended that the values of all chemistry parameters for a region
should be taken from one system determined to be the central tendency of all parameters
61
rather than picking the central tendency of individual chemistry parameters from different
systems. Following this recommendation, we assembled water chemistry data for over 800
freshwater systems in Canada in order to assign a water-type for each ecoregion. When
several measurements were available for one ecoregion, we selected the data from system
that closely resembled the central tendency of all systems. In cases when few measurements
were available, we chose the system which contributed the largest fraction of freshwater to
that ecoregion. For example, due to paucity of data for the northern Canada we used data for
the large lakes Great Bear, Great Slave and Nettling to represent their ecoregions. We did
not consider chemistries of inland saline and subsaline lakes of the northern Great Plains,
which represent a significant fraction of aquatic systems in western Canada. This is because
models used for estimating metal bioavailability (WHAM) and toxicity (BLM) are calibrated
for freshwater systems and are not intended to use for high ionic strength environments.
Following the above guidelines, we derived 24 sets of freshwater chemistries (water-types)
corresponding to each Canadian ecoregion (Table 3.2). We also identified one water-type
that would be representative of entire Canada, i.e. all Canadian ecoregions (water-type 25 in
Table 3.2). The latter water-type was selected from the list of all Canadian water chemistries
such that the selected set of measured values closely resembled the central tendencies of each
water chemistry parameter for 24 systems. The selected water-types ranged in pH from 5.3-
8.5, DOC from 2.3-22 mg/L, and water hardness from 5-375 mgCaCO3/L.
Metal speciation, and hence BFs and CTPs, vary non-linearly with background metal
concentrations. Background concentrations (BCs) can be highly variable among metals and
geographically at local to global scales (Reimann and Garrett 2005). However, due to
insufficient measured metal BCs for these freshwaters, we assumed a background total
dissolved concentration of 1 µg/L for Cu and Ni, and 10 µg/L for Zn for each water-type.
Further, since the water-types did not include measurements of TSS, we assumed an average
TSS concentration of 15 mg/L in all ecoregions, the value used in USEtoxTM
(Rosenbaum et
al. 2008). Because of these important data gaps, we performed a sensitivity analysis to
analyze the effects of varying TSS and BCs on model estimates.
62
Table 3.2: Freshwater chemistry data used in the geochemical model, WHAM 6.0, to
estimate Bioavailability Factors (BFs) for Cu, Ni and Zn in 24 Canadian ecoregions. A
background metal concentration of 1 µg/L for Cu and Ni, and 10 µg/L for Zn were used for
all water-types.
Ecoregion Ecoregion Description Water-type Description pH DOC Hardness Na Mg Ca K Cl SO4
mg/L mgCaCO3/L mg/L mg/L mg/L mg/L mg/L mg/L
1 Newfoundland Newfoundland 5.76 5.8 7.7 1.7 0.6 2.1 0.21 2.4 2.3
2 Labrador Labrador 6.56 5.6 15.2 1.4 0.8 4.7 0.34 1.1 5.0
3 Atlantic Maritime Nova Scotia 5.31 7.3 8.0 3.4 1.0 1.5 0.30 5.2 4.6
4 Quebec - Mixed Wood Plain Bay of Qunite (Lake Ontario), St. Lawrence River 7.95 2.9 108.7 8.0 5.5 34.4 1.29 19.5 21.6
5 Quebec - Boreal Shield Region Quebec Shield Lakes 5.95 4.2 4.7 0.5 0.3 1.4 0.20 0.3 1.3
6 Quebec - Northern Region Northern Quebec Acid Lakes 5.92 4.6 9.5 0.5 0.6 2.9 0.19 0.3 5.7
7 Ontario - Mixed Wood Plain Lake Huron (Georgian Bay) 7.60 4.9 80.5 3.4 3.5 26.4 1.02 4.5 20.0
8 Ontario - Boreal Shield Sudbury Shield Lakes (Long Lake) 7.10 4.1 32.6 22.4 2.8 8.5 1.43 36.4 13.5
9 Ontario - Northern Northern Ontario Algoma Lakes 5.99 4.7 18.2 0.8 1.3 5.1 0.34 0.7 12.8
10 Manitoba - Prairie Typical Western freshwater - Manitoba Prairie 8.25 15.0 134.0 20.0 15.0 29.0 5.00 1.4 24.0
11 Manitoba - Boreal Shield Manitoba Boreal Shield - Lake Manitoba 7.80 5.0 378.3 256.5 57.5 57.0 20.50 408.0 170.5
12 Manitoba - Northern Average of Northern Canadian Shield Lakes 5.97 5.3 15.9 1.3 1.2 4.4 0.33 0.3 9.9
13 Saskatchewan - Prairie Saskatchewan - Prairie - Figure Eight Lake 8.50 18.0 106.0 2.0 10 26 9.00 1.0 14
14 Saskatchewan - Northern Saskatchewan - Northern - Lake Athabasca 7.40 3.0 29.8 3.0 3 7 1.00 5.0 5
15 Alberta - Prairie Alberta - Prairie - Baptiste Lake 8.10 17.0 132.6 22.0 11 35 4.00 2.0 15
16 Alberta - Northern Alberta - North Saskatchewan River 8.10 2.8 163.7 5.8 13 44.2 1.00 2.2 42.7
17 B.C. - Montane Cordillera Prairie - e.g., Moonshine Lake 8.00 22.0 280.1 48.0 36 53 6.00 1.0 209
18 B.C. - South Pacific Maritime Mackenzie River Valley - GB/GS Lakes 7.80 6.7 35.7 2.2 3 8.7 1.20 2.0 6
19 B.C. - North Pacific Maritime Northern B.C. Lakes 7.80 10.0 154.2 45.9 20.6 27.9 10.00 3.6 8.2
20 B.C. - Northern Region Northern B.C. Lakes 7.80 10.0 154.2 45.9 20.6 27.9 10.00 3.6 8.2
21 Yukon Territory Yukon Territory Lakes 7.90 12.7 90.0 8.8 10 19.6 1.90 7.3 40.6
22 Mackenzie River Valley Mackenzie River Valley - GB/GS Lakes 7.80 6.7 35.7 2.2 3.4 8.7 1.20 2.0 5.6
23 Northwest Territories Northwest Territories Lakes 7.30 4.4 17.5 0.8 1.7 4.2 0.60 0.7 3.5
24 Arctic and Subarctic Arctic and Subarctic - Nettiling Lake 6.90 2.5 9.7 16.7 1.4 1.6 1.00 1.4 4.4
25 Canada Canadian Average 7.00 6.0 35.2 5.0 3.1 9.0 1.10 3.0 11
Values of chronic EC50 endpoints were taken from the literature for a minimum of three
organism classes, algae, daphnia and fish, for each metal. We used the identical datasets of
biotic species and bioassays for all ecoregions that were only corrected for regional water-
types to estimate EFs. Thus, the method does not address regional specificity in the structure
of aquatic biotic communities and climatic effects of extreme temperatures, which can
influence toxicity (Chapman et al. 2006 and references therein). Although we believe that
this should be a necessary and an important inclusion in spatially explicit ecotoxicity models,
we were unable to accommodate such details here because of the scarcity of data on biotic
species from lakes in all ecoregions and the limitations of BLM to extrapolate to aquatic
species outside of its range of calibration.
To calculate µ, each chronic EC50 value was scaled to a region’s water-type using an
appropriate BLM. The values of conditional binding constants used for the parameterization
of chronic BLMs are listed in Table 2.2 (Chapter 2). Further details of calculating µ and EFs
can be found in Gandhi et al. (2010).
63
3.5 Results and Discussion
Below we present results for each parameter in the calculation of CTPs for Canadian
ecoregions and compare them with those previously reported for use in LCIA of toxic
releases. Recall that metal Kd, BFs and EFs were calculated for each water-type (water pH,
DOC, and concentrations of major cations and anions) chosen to reflect the “central
tendency” in water-types of the 24 ecoregions. The effect of variability in a metal’s Kd, BF
and EF due to variability among ecoregion water-types (effects of environmental chemistry
only) was assessed by comparing them to that of average Canadian water-type (25) and
extreme maximum and minimum values obtained from all ecoregions.
3.5.1 Metal Partitioning (Kd)
Values of Kd had the greatest impact on the FF as partitioning between solid and aqueous
phases determines the fraction of metal lost due to sedimentation versus export. Values of
LogKd ranged from 4.5-5.4, 4.5-5.3, and 4.5-5.2 L/kg for Cu, Ni and Zn across ecoregions,
respectively (Figure 3.1a). The calculated average LogKd values of 5.0, 5.0 and 4.9 L/kg
from these ecoregions match with those estimated for average Canadian chemistry (water-
type 25) for Cu, Ni and Zn, respectively. The ranges and means were virtually identical
regardless of metal, which is, in part, due to the assumed constant concentration and
composition of TSS (15 mg/L). Gandhi et al. (2010) estimated average values of LogKd for
Cu, Ni and Zn as 4.1 (3.98-4.28), 3.9 (3.61-4.24), and 4.9 (4.39-5.37) L/kg, respectively, for
12 different EU freshwater archetypes (also using a constant 15 mg/L of TSS). The EU
estimates for Cu and Ni are significantly different than those for Canadian waters. The
average values of metal Kd for suspended sediment reported in the USEPA database for Cu,
Ni and Zn are 4.7 (3.1-6.1, n=70), 4.6 (3.5-5.7, n=30), and 5.1 (3.5-6.9, n=75) L/kg,
respectively (Allison and Allison, 2005). Harvey et al. (2007) used the values of 4.72, 4.80
and 5.26 L/kg for Cu, Ni and Zn, respectively, in their critical load analysis of metals using a
‘Unit World’ model.
64
1E+0
1E+2
1E+4
1E+6
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
3.0
4.0
5.0
6.0
7.0
Lo
gK
d
Cu Ni Zn
1
10
100
1000
FF
(d
ay)
b
a
c
d
e
1E-8
1E-6
1E-4
1E-2
1E+0
Que
bec-
Borea
l Shi
eld
BF
(d
imen
sio
nle
ss)
1E+0
1E+3
1E+6
1E+9
EF
(m
3/k
g)
CT
P (d
ay.m
3/k
g)
Figure 3.1: Model results for Cu, Ni, and Zn using the chemistry and landscape
characteristics of 24 Canadian freshwater-types and the overall Canadian water-type
65
described in Table 3.2. (a) WHAM estimated metal partition coefficients, Kd (L/kg), used in
fate calculations, and (b) WHAM estimated BF (Bioavailability Factors; dimensionless)
calculated as a fraction of total metal that is bioavailable and is assumed to be within the
truly dissolved fraction of total metal (c) FFs (Fate Factors; days) for unit emission of each
metal in freshwater compartment using the fate parameter values of Canadian ecoregions and
WHAM estimated Kd for each ecoregion freshwater-type (d) BLM estimated EFs (Effect
Factors; m3/kg) that were corrected for chemistry of freshwater-type in each ecoregion, and
(e) CTP (Comparative Toxicity Potential; day.m3/kg), where the variability in values of CTP
reflects variability in chemistry of freshwater-types and landscape properties of Canadian
ecoregions.
The lowest values of LogKd for all metals were consistently estimated for ecoregion-13
(Saskatchewan, prairie hard water). The highest values of LogKd for Cu, Ni and Zn were for
ecoregions-16 (Alberta, northern soft-water), -23 (NWT, head-waters), -4 (Quebec, mixed
wood plain), respectively. Combinations of low DOC, low hardness and high pH resulted in
the greatest partitioning on particles and thus had greatest effects on estimates of LogKd.
3.5.2 Fate
The FFs, which represent the residence time of metals in surface waters, ranged by up to 2
orders of magnitude from 4-270, 4-306, and 4-306 days for Cu, Ni and Zn across the
ecoregions due to variability in water chemistry and landscape properties affecting water
residence time (Figure 3.1c). Since the values of metal-specific Kd, which is a key parameter
in estimating FFs (Diamond et al. 1990, Gandhi et al. 2010), were similar, these FFs mainly
reflect variations in water residence times in these regions. Bhavsar et al. (2008) showed that
residence times for soluble metals closely match with water residence time for a system. The
highest FF was estimated for ecoregion-7 (Ontario, mixed wood plain) reflecting the long
water residence time of Lakes Ontario and Superior. The average FFs were 40, 44 and 45
days for Cu, Ni and Zn in all ecoregions, which also matched with those estimated for water-
type 25 (average Canadian chemistry).
66
These results of greater variability in FFs for Canadian ecoregions compared to the FFs of
the 12 EU water types considered by Gandhi et al. (2010) can be attributed to the inclusion of
a wide variety of lake sizes, including the Great Lakes whereas the dimension of the
freshwater compartment was constant for EU freshwater analysis. Variability in freshwater
compartment volume did not change the ranking of metals, but it could change the relative
importance of freshwater toxicity relative to that of other compartments such as soil.
3.5.3 Bioavailability
BFs varied over 4, 2 and 2 orders of magnitude for Cu, Ni and Zn, respectively (Figure 3.2a)
which is similar to the range estimated for 12 EU water-types (Gandhi et al., 2010). The
average values of BF were 5.5X10-5
(6.1X10-3
-1.3X10-7
), 8.7X10-2
(4.9X10-3
-5.0X10-1
), and
1.1X10-1
(7.8X10-3
-5.6X10-1
) for Cu, Ni and Zn, respectively (Figure 3.1b). These values
correspond to BFs for average Canadian surface waters of Cu: 5.9X10-5
, Ni: 9.9X10-2
, and
Zn: 1.3X10-2
. The lowest BF for all metals was estimated for ecoregion-13 (Saskatchewan,
prairie alkaline water). This water-type had the lowest values of LogKd due to it high pH and
the second highest level of DOC among all Canadian water-types, illustrating the importance
of these variables on metal bioavailability in surface waters. The highest BF (i.e., most
bioavailable) for all metals was estimated for watertype-3 (Nova Scotia, Atlantic Maritime
acid water), which had the lowest pH and low hardness.
3.5.4 Ecotoxicity
Estimated EFs varied up to 1.5 orders of magnitude for Cu, Ni and Zn, respectively (Figure
3.1d). The average values of 10µ(EC50)
(HC50-EC50) for the 24 ecoregions were 1.9 (0.5-11.2),
200 (33-2305) and 90 (46-165) µg/L for Cu, Ni and Zn, respectively (Figure 3.1d). These
estimates are close to the average estimates of 10µ(EC50)
for 12 EU water-types of 2.2 (0.6-
6.3), 180 (35-445) and 110 (50-150) µg/L for Cu, Ni and Zn, respectively (Gandhi et al.
2010).
67
1.E-7
1.E-6
1.E-5
1.E-4
1.E-3
1.E-2
1.E-1
1.E+0
Cu Ni Zn
BF
(-)
1.E-1
1.E+0
1.E+1
1.E+2
1.E+3
Cu Ni Zn
FF
(d
ay
s)
a
b
1.E+3
1.E+4
1.E+5
1.E+6
1.E+7
1.E+8
Cu Ni Zn
EF
(m
3/k
g)
c
1.E+0
1.E+1
1.E+2
1.E+3
1.E+4
1.E+5
1.E+6
Cu Ni Zn
CT
P (
da
y.m
3/k
g)
d
1.E-7
1.E-6
1.E-5
1.E-4
1.E-3
1.E-2
1.E-1
1.E+0
Cu Ni Zn
BF
(-)
1.E-1
1.E+0
1.E+1
1.E+2
1.E+3
Cu Ni Zn
FF
(d
ay
s)
a
b
1.E+3
1.E+4
1.E+5
1.E+6
1.E+7
1.E+8
Cu Ni Zn
EF
(m
3/k
g)
c
1.E+0
1.E+1
1.E+2
1.E+3
1.E+4
1.E+5
1.E+6
Cu Ni Zn
CT
P (
da
y.m
3/k
g)
d
Figure 3.2: Model estimated ranges in (a) Bioavailability Factors (BF; dimensionless), (b)
Fate Factors (FFs; days) for unit emission of metals in freshwater compartment, (c) Effect
Factors (EFs; m3/kg) that represent average potential ecotoxicity, and (d) Comparative
Toxicity Potentials (CTPs; day.m3/kg) for Cu, Ni, and Zn calculated to examine variability in
chemistry of freshwater-types and landscape properties of Canadian ecoregions.
68
The lowest EF was estimated for ecoregion-11 (Manitoba, boreal shield), mainly because of
highest water hardness and relatively high pH, whereas the highest EF was estimated for
ecoregion-13 (Saskatchewan, prairie) possibly due to presence of very high DOC and pH
(Figure 3.1d). The corresponding estimates of 10µ(EC50)
for the average Canadian surface
water-type (25) were 0.7, 55 and 73 µg/L for Cu, Ni and Zn, respectively. Although these
estimates correspond reasonably with the averages of 24 ecoregions, they do not reflect the
large range in the regional values. The values of EFs calculated using the Canadian water-
type(25) are three times more protective, especially for Cu and Ni, than the corresponding
averages of the regional estimates of EF. The provisional values of chronic 10µ provided in
USEtoxTM
are 67, 440, and 604µg/L for Cu, Ni and Zn, respectively.
3.5.5 Comparative Toxicity Potential
CTPs of all metals varied over 3 orders-of-magnitude among the 24 freshwater-types (Figure
3.1e), which mainly reflected variability in their BFs. The median values of CTPs were
2x103 (15-3.6x10
4), 1.6x10
4 (4.3x10
2-2.1x10
5), and 2.2x10
4 (8.2x10
2-2.7x10
5) day.m
3/kg for
Cu, Ni and Zn, respectively (Figure 3.2d). These estimates for the three metals overlap and
match closely with the range and median values of CTPs for Cu, Ni and Zn of 1.5x104
(1.5x102-1.2x10
5), 5.6x10
4 (9.4x10
3-4.1x10
5), and 2.1x10
4 (7x10
3-5.8x10
4) day.m
3/kg for 12
EU water-types (Gandhi et al. 2010). Despite this overlap and similarity to 12 EU water-
types, the relative ranking in metal CTPs, which indicates relative potential for adverse
effects, was different for several ecoregions as discussed below.
The highest value of CTP represents the ecoregion of greatest potential for metal
ecotoxicological impacts (or relative hazard) and vice versa. The highest value of CTP for Cu
was estimated for ecoregion-12 (Manitoba, northern waters; Figure 3.1e), which was mainly
because of high bioavailability although the metal residence time was average. In contrast,
the highest values of CTP for Ni and Zn were estimated for ecoregion-7 (Ontario, mixed
wood plain), which is a product of average bioavailability, toxicity and the highest residence
time in its freshwater (mainly represented by the Great Lakes). Thus, bioavailability controls
the hazard for Cu, whereas fate controls the hazard for Ni and Zn, and both affect the final
outcome.
69
The lowest values of CTP for all metals were estimated for ecoregion-21 (Yukon, northern
freshwaters; Figure 3.1e), which is a product of low bioavailability, toxicity and the lowest
residence time in freshwater. Cu would be the lowest concern of the 3 metals in most
Canadian ecoregions except for ecoregion-3 (New Brunswick Atlantic maritime) for which
Zn had the lowest CTP. The low concern for Cu is largely because of its binding to DOC
which lowers its bioavailability. Ni and Zn had similar values of CTPs but their order
switched depending on regional water chemistry.
The provisional values of CTP for Cu, Ni and Zn calculated using USEtoxTM
are 5.5x104,
1.5x104, 3.9x10
4 day.m
3/kg. The values of CTP calculated by USES-LCA2 (van Zelm et al.
2009) are 4.3x104, 1.5x10
4 and 2.7x10
3 day.m
3/kg for Cu, Ni and Zn, respectively. The range
of CTPs presented here for 24 ecoregions are 1-3 orders of magnitude lower for all metals
than those reported by models that do not consider metal speciation into fate and ecotoxicity
calculations (e.g., Huijbregts et al. 2000).
3.6 Sensitivity Analysis
3.6.1 Freshwater Residence Time
Water chemistry affects metal partitioning and BFs, and therefore contributes to the
variability in FFs, EFs and CTPs. The variability in freshwater volume and therefore in
water residence time only affects FFs. Gandhi et al. (2010) showed that FFs varied by 3.5
times due to chemistry of water-types, when water residence time was held constant. We
carried out a sensitivity analysis of water residence time on FF by keeping the water
chemistry parameters constant and thus constant values of Kd, BFs and EFs for all
ecoregions. We used the LogKd values of 4.7, 4.6 and 5.1 for Cu, Ni and Zn, respectively,
which are the reported mean values by US EPA (Allison and Allison, 2005). The resultant
FFs varied over two orders of magnitude for all metals (Figure 3.3; Cu: 4-400, Ni: 5-470, Zn:
3-265), which translated into a similar range of variability of CTPs (results not shown). The
results suggest that the water residence time is an important factor that merits inclusion in a
regional model for LCIA and hazard assessment.
70
1
10
100
1000
Fate
Facto
rs (
days)
Cu Ni Zn
Figure 3.3: Sensitivity of metal Fate Factors (FFs; days) to landscape properties of
freshwater compartments of Canadian ecoregions.
3.6.2 Background Metal Concentrations
We varied metal background concentrations (BCs) over two orders of magnitudes (0.1x and
10x original values) to analyze the sensitivity of model results to the base case BCs of 1, 1
and 10 µg/L for Cu, Ni and Zn, respectively. This range covers most natural freshwater
systems within Canada (Table 3.3). Figure 3.4 shows that for these variations in BCs CTPs
ranged about -100 (-2x) to 2100% (42x), -50 (-1x) to 150% (3x), and -50 (-1x) to 200% (4x)
from the base cases for Cu, Ni and Zn, respectively (Figure 3.4). A decrease in BCs by a
factor of 10 resulted in CTPs that were up to a factor of 2, 1.5 and 1.5 lower for Cu, Ni and
Zn, respectively. However, a 10 times increase in BCs resulted in larger changes, i.e. CTPs
increased by up to 42, 3 and 4 times for Cu, Ni and Zn, respectively. Note that metal fate is
largely unaffected over this range of metal BCs (Figure 3.4). Metal EFs (toxicity) did not
change since the water chemistry remained the same as of the base case for each ecoregion
(results not shown). The increase in CTP was mainly due to an increase in the BF, which
ranged about 0.5 to 23, 0.2 to 2.5, and 0.25 to 3 times from the base cases for Cu, Ni and Zn,
respectively (Figure 3.4). These results imply that in systems with relatively high background
71
No. System Province TSS Range TSS Cu Ni Zn
mg/L mg/L µg/L µg/L µg/L
1 Fraser River B.C. 34 10 - 76 26 2 4
2 Skeena River B.C. 35 19 2 7
3 Great Bear River NWT 2 1 - 6 1 1 1
4 Slave River NWT, Alberta 34 7 - 96 7 8 46
5 Mackenzie River NWT, Northern Alberta, Sasketchewan 35 4 - 81 4 5 14
6 Saskatchewan River Alberta, Sasketchewan, Parts of Manitoba 30 2 - 90 2 3 6
7 Churchill River Alberta, Sasketchewan, Parts of Manitoba 3 2 - 10 1 1 3
8 Nelson River Northern Ontario 9 4 - 27 2 1 3
9 St. Lawrence River Quebec 4 2 - 10 1 1 5
10 Roseau River Southern Manitoba 40 1 - 690 1 - 3 1 - 2 5
concentrations (>>1 µg/l), the current CTPs and BFs for Cu could be underestimated.
However, this could be offset by the acclimation of aquatic biota to elevated metal levels in
these systems, which is not considered in current calculations of EFs. Our analysis of the
results suggests that the sensitivity of BF to the BC of Cu is mainly due to the binding
capacity of DOC that is affected by pH in a given ecoregion (Figure 3.5). The greatest
changes in BFs of up to 23 times were observed for ecoregions -16 (Alberta, northern soft-
water) and -4 (Quebec, mixed wood plain) with low DOC and high pH values (Figure 3.5).
The fraction of metal in truly dissolved form significantly increased at the expense of
colloidal metal when the BC of Cu was increased from 1 to 10 µg/L; however, that fraction
remained relatively constant when BC was decreased to 0.1 µg/L for each ecoregion. Note
that the fractions of metal in the total dissolved (truly dissolved + colloidal) were similar to
the respective base case scenarios (Figure 3.5). The influence of DOC binding on Ni and Zn
speciation is generally not significant and hence has minimal effect on their BFs. Other
water chemistry parameters also affect BF estimates of metals and therefore a more
comprehensive analysis of speciation is necessary to generalize which combination of
chemistry parameters (e.g., water-type) will have the highest influence on a metal’s
bioavailability.
Table 3.3: Measurements of metal background concentrations and total suspend sediment
concentrations in freshwaters across Canada. This information was used to estimate ranges
in parameter values to conduct the sensitivity analysis of model results for bioavailability,
fate and ecotoxicity potentials of metals in freshwaters of 24 Canadian ecoregions.
72
-500%
0%
500%
1000%
1500%
2000%
2500%
Cu
-50%
0%
50%
100%
150%
200%
Ni
-50%
0%
50%
100%
150%
200%
Zn
Bioavailability
-15%
-10%
-5%
0%
5%
10%
15%
20%
25%
Cu
-15%
-10%
-5%
0%
5%
10%
15%
20%
25%
Ni
-15%
-10%
-5%
0%
5%
10%
15%
20%
25%
Zn
Fate
-500%
0%
500%
1000%
1500%
2000%
2500%
Cu
-50%
0%
50%
100%
150%
200%
Ni
-50%
0%
50%
100%
150%
200%
Zn
CTP
Ecoregions1 3 5 7 9 11 13 15 17 19 21 23 25 1 3 5 7 9 11 13 15 17 19 21 23 251 3 5 7 9 11 13 15 17 19 21 23 25
BC 0.1X BC 10X
Figure 3.4: Sensitivity of modelled metal BFs (bioavailability), FFs (fate) and CTPs (Comparative Toxicity Potentials) to
background concentrations of Cu, Ni and Zn in freshwaters of Canadian ecoregions. The numbers on x-axis represent Canadian
ecoregions as listed in Table 3.1. The results on y-axis are displayed as percentage changes from the respective base case for each
modelled parameter and ecoregion.
73
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Diss Fr Coll Fr Part Fr
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Canadian Ecoregions
Fra
ction
of
tota
l C
u
a
b
c
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Diss Fr Coll Fr Part Fr
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Canadian Ecoregions
Fra
ction
of
tota
l C
u
a
b
c
Figure 3.5: Percentage of total Cu in dissolved, colloidal and particulate phases estimated
for Cu background concentrations of (a) 0.1 µg/L, (b) 1 µg/L (base case scenario), and (c) 10
µg/L.
74
3.6.3 Total Suspended Sediment Concentrations
In the absence of a comprehensive set of measured TSS data for a sensitivity analysis, we
used TSS concentrations of 5 and 35 mg/L for all ecoregions which is representative of
cross-country conditions (Table 3.3). Model results were compared with the base case
estimates that used TSS of 15 mg/L. CTPs ranged -10 to 1050% (0.9-21X), -20 to 1050%
(0.8-21X), and -50 to 1250% (0.5-25X) from their base cases for Cu, Ni and Zn, respectively,
predominantly due to changes in metal fate (Figure 3.6). Values of TSS affect metal fate by
changing (a) the fraction of metal that is associated with/adsorbed to particles (Kd), and (b)
the net sedimentation rate. The former effect is non-linear and controlled by adsorption
isotherms whereas the latter effect is linear and proportional to TSS concentration. In contrast
to the results of variations in BC (Figure 3.4), metal bioavailability was largely unaffected
(within ±1% of the base case) over this range of TSS (Figure 3.6) mainly because the metal
background concentrations, expressed as total dissolved concentrations in speciation
calculations, were the same for all ecoregions. Thus, this analysis may have underestimated
the effect of metal sorption to particles because the dissolved metal concentrations, which
also correspond to constant EFs, were held constant to calculate equilibrium partitioning on
particles. As mentioned above, FFs and thus CTPs increased by a factor of up to 25 as a
result of increasing TSS concentration from 15 to 35 mg/L. This is contrary to expectation
since the increase in TSS concentration increases the sedimentation rate of metals and thus
decreases the metal residence time in water. However, the estimates of metal adsorbed to
TSS (Kd) decreased by 10 times which “cancelled” the linear effects of net sedimentation of
metals.
75
-1%
0%
1%
Cu
-5%
0%
5%
Ni
-10%
-5%
0%
5%
10%
Zn
Bioavailability Fate CTP
Ecoregions
1 3 5 7 9 11 13 15 17 19 21 23 25 1 3 5 7 9 11 13 15 17 19 21 23 251 3 5 7 9 11 13 15 17 19 21 23 25
TSS 5 mg/L TSS 35 mg/L
-200%
0%
200%
400%
600%
800%
1000%
1200%
Cu
-200%
0%
200%
400%
600%
800%
1000%
1200%
Ni
-200%
0%
200%
400%
600%
800%
1000%
1200%
1400%
Zn
-200%
0%
200%
400%
600%
800%
1000%
1200%
Cu
-200%
0%
200%
400%
600%
800%
1000%
1200%
Ni
-200%
0%
200%
400%
600%
800%
1000%
1200%
1400%
Zn
Figure 3.6: Sensitivity of modelled metal BFs (bioavailability), FFs (fate) and CTPs (Comparative Toxicity Potentials) to total
suspended sediment (TSS) concentrations in freshwaters of Canadian ecoregions. The numbers on x-axis represent Canadian
ecoregions as listed in Table 3.2. The results on y-axis are displayed as percentage changes from the respective base case for each
modelled parameter and ecoregion.
76
3.7 Summary
Current practice in LCIA and chemical hazard screening relies on generic values of
chemicals’ CTP that do not account for regional differences in environmental variability. We
added regional variability from ChemCAN, a fugacity-based multimedia model for 24
Canadian ecoregions, to the USEtoxTM
model used here to estimate freshwater CTPs. Our
results show that there could be up to three orders-of-magnitude difference in CTPs from
region-to-region driven largely by differences in metal bioavailability (due to differences in
water hardness, DOC and pH) and fate (due to differences in water residence time). More
importantly our results showed that the relative ranking in metal CTPs and thus potential for
adverse effects changed among metals for several ecoregions. Since this analysis did not
include regional differences in metal background and total suspended sediment
concentrations, we completed a sensitivity analysis that showed that variability of 10- and 7-
times from the assumed values of BCs (1 µg/L for Cu, Ni and 10 µg/L for Zn) and TSS (15
mg/L) led to differences in CTPs of up to 20 and 10-times, respectively. These differences
were mainly due to metal speciation and adsorption characteristics that largely affected
estimates of bioavailability followed by fate, and not ecotoxicity.
There are several implications of this study. First, considering geographic variability in
chemical hazard and LCIA can provide a more accurate impact analysis of chemical toxicity
potentials. Second, practitioners need to choose environmental chemistry parameters for
geographic regions around the world to improve the assessment of chemical hazard. This is
because such choices affect the outcome of an impact assessment where the impact is not just
based on the relative ranking of chemicals but the position of individual metals within the
ranking Alternatively, the choice of water chemistry in a generic analysis could be provided
for regional considerations where necessary. Finally, if these improvements are incorporated
into LCA then it raises an important research question of what is the highest but still practical
geographic resolution needed and feasible for LCIA. It is imperative that a consistent spatial
differentiation for different impact categories is implemented for all commonly accepted
impact categories such that it also includes geographically differentiated land use and water
use impact assessment indicators. A greater practical challenge is to develop methods and
77
databases to connect such spatially explicit analysis of LCIA with similar details in Life
Cycle Inventory (LCI).
3.8 References
Alfaro-De la Torre, M.C., Tessier, A., 2002. Cadmium deposition and mobility in the
sediments of an acidic oligotrophic lake. Geochim. Cosmochim. Acta 66, 3549-3562.
Allen, H.E., Janssen, C.R., 2006. Incorporating bioavailability into criteria for metals. In:
Twardowska, I., Allen, H.E., Haggblom, M.M., Stefaniak, S. (Eds.). Viable Methods
of Soil and Water Pollution Monitoring, Protection and Remediation. Springer,
Dordrecht, pp. 93-105.
Allison, J.D., Allison, T.L., 2005. Partition coefficients for metals in surface water, soil, and
waste. U.S. EPA Report EPA/600/R-05/074, p. 93.
Bare, J., Gloria, T., Norris, G., 2006. Development of the method and U.S. normalization
database for Life Cycle Impact Assessment and sustainability metrics. Environ. Sci.
Technol. 40, 5108-5115.
Bare, J.C., 2006. Risk assessment and Life-Cycle Impact Assessment (LCIA) for human
health cancerous and noncancerous emissions: Integrated and complementary with
consistency within the USEPA. Human Ecol. Risk Assess. 12, 493-509.
Bare, J.C., Norris, G.A., Pennington, D.W., McKone, T., 2003. TRACI - The Tool for the
Reduction and Assessment of Chemical and Other Environmental Impacts. J. Indust.
Ecol. 6, 49-78.
Bhavsar, S.P., Diamond, M.L., Evans, L.J., Gandhi, N., Nilsen, J., Antunes, P., 2004.
Development of a coupled metal speciation-fate model for surface aquatic systems.
Environ. Toxicol. Chem. 23, 1376-1385.
Bhavsar, S.P., Gandhi, N., Diamond, M.L., Lock, A.S., Spiers, G., De la Torre, M.C.A.,
2008. Effects of estimates from different geochemical models on metal fate predicted
by coupled speciation-fate models. Environ. Toxicol. Chem. 27, 1020-1030.
78
Bryan, S.E., Tipping, E., Hamilton-Taylor, J., 2002. Comparison of measured and modelled
copper binding by natural organic matter in freshwaters. Comparative Biochemistry
and Physiology C-Toxicology & Pharmacology 133, 37-49.
Campbell, P.G.C., 1995. Interactions between Trace Metals and Aquatic Organisms: A
Critique of the Free-Ion Activity Model. In: Tessier, A., Turner, D.R. (Eds.). Metal
Speciation and Bioavailability in Aquatic Systems. John Wiley, New York, pp. 45-
102.
Chapman, P.M., McDonald, B.G., Kickham, P.E., McKinnon, S., 2006. Global geographic
differences in marine metals toxicity. Marine Pollut. Bulletin 52, 1081-1084.
Di Toro, D.M., Allen, H.E., Bergman, H.L., Meyer, J.S., Paquin, P.R., Santore, R.C., 2001.
Biotic ligand model of the acute toxicity of metals. 1. Technical basis. Environ.
Toxicol. Chem. 20, 2383-2396.
Diamond, M.L., 1995. Application of a mass-balance model to assess in-place arsenic
pollution. Environ. Sci. Technol. 29, 29-42.
Diamond, M.L., Gandhi, N., Adams, W.J., Atherton, J., Bhavsar, S.P., Bulle, C., Campbell,
P.G.C., Dubreuil, A., Fairbrother, A., Farley, K., Green, A., Guinee, J.B., Hauschild,
M.Z., Huijbregts, M.A.J., Humbert, S., Jensen, K.S., Jolliet, O., Margni, M., McGeer,
J.C., Peijnenburg, W.J.G.M., Rosenbaum, R., van de Meent, D., Vijver, M.G., 2010.
The Clearwater consensus: the estimation of metal hazard in fresh water. Int. J Life
Cycle Assess. 15, 143-147.
Diamond, M.L., Mackay, D., Cornett, R.J., Chant, L.A., 1990. A model of the exchange of
inorganic chemicals between water and sediments. Environ. Sci. Technol. 24, 713-
722.
Gandhi, N., Diamond, M.L., van de Meent, D., Huijbregts, M.A.J., Peijnenburg, W., Guinee,
J., 2010. New method for calculating comparative toxicity potential of cationic metals
in freshwater: Application to copper, nickel, and zinc. Environ. Sci. Technol. 44,
5195-5201.
79
Godin, J., Menard, J.F., Hains, S., Deschenes, L., Samson, R., 2004. Combined use of life
cycle assessment and groundwater transport modeling to support contaminated site
management. Human Ecol. Risk Assess. 10, 1099-1116.
Harvey, C., Mackay, D., Webster, E., 2007. Can the unit world model concept be applied to
hazard assessment of both organic chemicals and metal ions? Environ. Toxicol.
Chem. 26, 2129-2142.
Hauschild, M.Z., Huijbregts, M., Jolliet, O., MacLeod, M., Margni, M., van de Meent, D.V.,
Rosenbaum, R.K., McKone, T.E., 2008. Building a model based on scientific
consensus for life cycle impact assessment of chemicals: The search for harmony and
parsimony. Environ. Sci. Technol. 42, 7032-7037.
Hauschild, M.Z., Potting, J., Hertel, O., Schopp, W., Bastrup-Birk, A., 2006. Spatial
differentiation in the characterisation of photochemical ozone formation - The
EDIP2003 methodology. Int. J. Life Cycle Assess. 11, 72-80.
Heijerick, D.G., Janssen, C.R., De Coen, W.M., 2003. The combined effects of hardness, pH,
and dissolved organic carbon on the chronic toxicity of Zn to D-magna: Development
of a surface response model. Archives of Environmental Contamination and
Toxicology 44, 210-217.
Hettelingh, J.P., Posch, M., Potting, J., 2005. Country-dependent characterisation factors for
acidification in Europe - A critical evaluation. Int. J. Life Cycle Assess. 10, 177-183.
Huijbregts, M.A.J., Seppala, J., 2001. Life cycle impact assessment of pollutants causing
aquatic eutrophication. Int. J. Life Cycle Assess. 6, 339-343.
Huijbregts, M.A.J., Thissen, U., Guinee, J.B., Jager, T., Kalf, D., van de Meent, D., Ragas,
A.M.J., Sleeswijk, A.W., Reijnders, L., 2000. Priority assessment of toxic substances
in life cycle assessment. Part I: Calculation of toxicity potentials for 181 substances
with the nested multi-media fate, exposure and effects model USES-LCA.
Chemosphere 41, 541-573.
80
Humbert, S., Horvath, A., 2006. Geographically differentiated LCIA in North America:
influence of the scale for assessing the impacts of power plants' emissions. SETAC
North America, 27th Annual Meeting, Montreal, PQ, Canada.
Humbert, S., Manneh, R., Shaked, S., Wannaz, C., Horvath, A., Deschenes, L., Jolliet, O.,
Margni, M., 2009. Assessing regional intake fractions in North America. Sci. Tot.
Environ. 407, 4812-4820.
Itsubo, N., Inaba, A., 2003. A New LCIA Method: LIME has been completed. Int. J. Life
Cycle Assess. 8, 305.
Menard, J.F., Lesage, P., Deschenes, L., Samson, R., 2004. Comparative life cycle
assessment of two landfill technologies for the treatment of municipal solid waste.
Int. J. Life Cycle Assess. 9, 371-378.
Nigge, K.M., 2001. Generic spatial classes for human health impacts, part I: Methodology.
Int. J. Life Cycle Assess. 6, 257-264.
Pennington, D.W., Margni, M., Ammann, C., Jolliet, O., 2005. Multimedia fate and human
intake modeling: Spatial versus nonspatial insights for chemical emissions in Western
Europe. Environ. Sci. Technol. 39, 1119-1128.
Potting, J., Hauschild, M.Z., 2006. Spatial differentiation in life cycle impact assessment - A
decade of method development to increase the environmental realism of LCIA. Int. J.
Life Cycle Assess. 11, 11-13.
Potting, J., Schopp, W., Blok, K., Hauschild, M., 1997. Comparison of the acidifying impact
from emissions with different regional origin in life-cycle assessment. International
Conference on Mapping Environmental Risks and Risk Comparison (RISK 97),
Amsterdam, Netherlands, pp. 155-162.
Reimann, C., Garrett, R.G., 2005. Geochemical background—concept and reality. Sci. Tot.
Environ. 350, 12-27.
Rosenbaum, R.K., Bachmann, T.M., Gold, L.S., Huijbregts, M.A.J., Jolliet, O., Juraske, R.,
Koehler, A., Larsen, H.F., MacLeod, M., Margni, M., McKone, T.E., Payet, J.,
Schuhmacher, M., van de Meent, D., Hauschild, M.Z., 2008. USEtox-the UNEP-
81
SETAC toxicity model: recommended characterisation factors for human toxicity and
freshwater ecotoxicity in life cycle impact assessment. Int. J. Life Cycle Assess. 13,
532-546.
Schecher, W.D., McAvoy, D.C., 1992. MINEQL+ - A software environment for chemical-
equilibrium modeling. Comput. Environ. Urban Sys. 16, 65-76.
Schöpp, W., Amann, M., Cofala, J., Heyes, C., Klimont, Z., 1999. Integrated assessment of
European air pollution emission control strategies. Environ. Model. Software 14, 1-9.
Seppälä, J., Posch, M., Johansson, M., Hettelingh, J.P., 2006. Country-dependent
characterisation factors for acidification and terrestrial eutrophication based on
accumulated exceedance as an impact category indicator. Int. J. Life Cycle Assess. 11,
403-416.
Sleeswijk, A.W., 2003. General prevention and risk minimization in LCA - A combined
approach. Environ. Sci. Pollut. Res. 10, 69-77.
Sleeswijk, A.W., 2006. GLOBOX - A spatially differentiated multimedia fate and exposure
model. Int. J. Life Cycle Assess. 11, 141-141.
Tipping, E., 1998. Humic ion-binding model VI: An improved description of the interactions
of protons and metal ions with humic substances. Aquat. Geochem. 4, 3-48.
Toffoletto, L., Bulle, C., Godin, J., Reid, C., Deschenes, L., 2007. LUCAS - A new LCIA
method used for a Canadian-specific context. Int. J. Life Cycle Assess. 12, 93-102.
Toffoletto, L., Deschenes, L., Samson, R., 2005. LCA of ex-situ bioremediation of diesel-
contaminated soil. Int. J. Life Cycle Assess. 10, 406-416.
van Zelm, R., Huijbregts, M.A.J., van de Meent, D., 2009. USES-LCA 2.0-a global nested
multi-media fate, exposure, and effects model. Int. J. Life Cycle Assess. 14, 282-284.
Webster, E., Mackay, D., Di Guardo, A., Kane, D., Woodfine, D., 2004. Regional differences
in chemical fate model outcome. Chemosphere 55, 1361-1376.
82
4. Implications of considering metal bioavailability in estimates of
freshwater ecotoxicity: Examination of two case studies ©
4.1 Abstract
Goal, Scope and Background. Previous methods of assessing Characterization Factors
(CFs) of metals in Life Cycle Impact Assessment (LCIA) models were based on multi-media
fate, exposure and effect models originally developed to address the potential impacts of
organic chemicals. When applied to metals, the models neglect the influence of ambient
chemistry on metal speciation, bioavailability and toxicity. Gandhi et al. (2010) presented a
new method of calculating CFs for freshwater ecotoxicity that addresses these metal-specific
issues. In this paper, we compared and assessed the consequences of using the new method
versus currently available LCIA models for calculating freshwater ecotoxicity, as applied to
two case studies previously examined by Gloria et al. (2006): (1) the production of copper
(Cu) pipe, and (2) a zinc (Zn) gutter system.
Methods. Using the same inventory data as presented by Gloria et al. (2006), we calculated
and compared the LCIA outcomes for freshwater ecotoxicity of each case study using four
models: USES-LCA 1.0, USES-LCA 2.0, USEtoxTM
using the previous approach and
USEtoxTM
using the new method. Since the new method requires specification of water
chemistry for the freshwater compartment, we explored the effect of using seven freshwater
archetypes. We analyzed the freshwater ecotoxicity outcomes of the two case studies with
respect to the different models, infinite versus 100 year time-scales for calculating impacts
after metal emissions, and water chemistries representing spatial environmental variability.
© Contents of this chapter have been adopted from the manuscript in press in a special issue of International
Journal of Life Cycle Assessment:
Gandhi, N., Diamond, M.L., Huijbregts, M.A.J., Guinée, J., Peijnenburg, W.J.G.M., and van de Meent, D.
(2011b) Implications of considering metal bioavailability in estimates of freshwater ecotoxicity: Examination of
two case studies. International Journal of Life Cycle Assessment In press.
I was primarily responsible for the data collection, model applications, analyis of model results, and writing of
this manuscript.
83
Results and Discussion. Significant differences in CFs, overall freshwater ecotoxicity score
(Σ CF x emissions) and the contributions of individual metals to the overall score were traced
back to differences in modelling methods (e.g., variations in compartments included in the
fate model), the choice of metal partition coefficients versus those explicitly calculated based
on water chemistry (USEtox(new)), and the calculation of effect factors (EFs). Metal CFs
calculated using USES-LCA 1.0 ranked Co>Ni>Cd≈Cu>Zn>Pb, but changed using
USEtox(new) to Cd>Co>Ni>Zn>Cu>Pb for the archetype of hard alkaline water and
Cd>Ni>Co>Cu≈Zn>Pb for the archetype of soft, acidic water. For the Cu pipe, total
freshwater ecotoxicity scores for metal emissions into air and water ranged from 0.01-0.02
for USES-LCA1.0, ~1 for USEtoxTM
(previous) to 0.0002-0.01 1, 4-DCB eq. for
USEtox(new) depending on the archetype. Whereas Cu followed by Ni emissions contributed
most to total freshwater ecotoxicity estimated by USES-LCA1.0, Cu, Cd, Ni and Zn ,
emissions were all important contributors towards freshwater ecotoxicity with
USEtoxTM
(new) , with differences in contributions dependent on the freshwater archetype.
For the Zn gutter case study, the total scores varied from 10 for USEtoxTM
(previous) to 0.008
for USES-LCA 2.0 and 0.02 to 0.11 eq. to 1, 4-DCB for USEtox(new). Zn contributed ~98%
towards the freshwater ecotoxicity scores of metals in all models. For both case studies,
differences in ecotoxicity scores were not significant for the infinite vs 100 year time-scale.
Conclusions. Accounting for metal bioavailability and speciation by using USEtox(new)
when calculating CFs decreased by1 to 4 orders-of-magnitude the total metal freshwater
ecotoxicity scores (Σ CF x emissions) attributable to metal emissions tallied for Cu pipe and
Zn gutter system case studies (Gloria et al. 2006). This broad range came from the model
used in comparison to USEtox(new) and the choice of freshwater archetype. Additionally,
contributions of each metal to the total score of the Cu pipe case study changed significantly
from the use of previous CFs (Huijbregts et al. 2000) versus the revised CFs (Gandhi et al.
2010).
Practical Implications. Metal CFs calculated using the method proposed by Gandhi et al.
(2010) significantly lowers the total freshwater ecotoxicity impact of metal emissions. It is
suggested that this lower estimate of potential impact from metal emissions is consistent with
our understanding of metal chemistry. The magnitude of the potential freshwater ecotoxicity
84
of metals depends on the chemistry of the modelled freshwater compartment, similarly to the
dependence of acidification potential on regionally-variant freshwater chemistry.
Keywords: Life Cycle Impact Assessment (LCIA); metals; Comparative Toxicity Potential
(CTP); speciation-complexation; bioavailability; ambient chemistry
4.2 Introduction
Non-ferrous metals, in their numerous inorganic and organic forms, often rise to the top of
toxicity concerns in Life Cycle Assessment (LCA). This is not necessarily because of the
inherent hazard of metals, but because the tools used for Life Cycle Impact Assessment
(LCIA) do not incorporate the complex and seemingly idiosyncratic behaviour of metals. A
concern that metal hazard has been overestimated using available screening tools was
expressed in several fora. The Lausanne review workshop (Jolliet et al. 2006) and the
Apeldoorn Declaration (Apeldoorn 2004), expressed the consensus view among multi-
sectoral participants that metal-specific properties, speciation and bioavailability must be
considered in the assessment of chemical hazard of metal emissions. At issue is that the tools
used to assign characterization factors (CFs) indicative of the relative ecotoxicological
hazard of metals, are based on the behaviour of organic compounds. For organic
compounds, there is a relatively simple dichotomy between bioavailable and non-
bioavailable chemical in an evaluative environment. In contrast, metal bioavailability
depends on metal-specific speciation which is sensitive to ambient chemistry. Examples of
this dependency are the distribution of cationic metals between dissolved and particle phases
as a function of pH and the very strong affinity of Cu binding to dissolved organic matter
(DOM) (e.g., Adams and Chapman 2005). These differences between the aqueous chemistry
of non-ferrous metals and organics result in the inconsistent assessment of the relative hazard
of non-ferrous metals and organics.
The Clearwater Consensus (Diamond et al. 2010), which again assembled a multi-sectoral
group of experts in metal chemistry and LCA, formulated a series of recommendations aimed
at eliminating the inconsistencies between the assessment of non-ferrous metals and organics
85
in LCIA, hazard and risk assessment. A key recommendation was to estimate bioavailability
by considering metal-specific speciation.
The recommendations of the Clearwater Consensus were incorporated into a new framework
proposed by Gandhi et al. (2010) in which bioavailability is explicitly introduced into
calculations of metal impact. The revised method addresses metal-specific issues in three
stages by estimating: (1) metal bioavailability factors (BFs) using an equilibrium,
geochemical speciation model that also calculates metal adsorption to humic material (e.g.,
Windermere Humic Adsorption Model – WHAM 6.0; Tipping 1998), (2) fate factors (FFs) in
USEtoxTM
(Rosenbaum et al. 2008) by using metal Kd values calculated separately in the
geochemical model, and (3) effect factors (EFs) by applying the Biotic Ligand Model (BLM;
Di Toro et al. 2001) to correct for metal bioavailability in toxicity assesments of aquatic
organisms (e.g., chronic EC50). The expression of this three-stage analysis is a final
Comparative Toxicity Potential (CTP) that is a numerical indicator of ecotoxicity. The CF is
determined by normalizing the CTP of a substances relative to other chemicals under defined
model conditions (Gandhi et al. 2010; 2011).
Gloria et al. (2006) explored the consequences of using metal CFs from different LCIA
models that do not explicitly account for metal speciation and bioavailability. They
examined two case studies: (1) use of Cu pipe for supplying domestic water, and (2) an
average Zn based gutter system for residential use. Their analysis used Life Cycle Inventory
(LCI) data for total emissions of various chemicals as part of the cradle-to-grave analysis of
both case studies. Gloria et al. (2006) reported the outcomes of the ecotoxicity impact
category as a result of using five different LCIA models to calculate freshwater ecotoxicity
potential of metals: USES-LCA 1.0, Eco-indicator 99 (EI 99), IMPACT 2002, EDIP 97, and
CalTOX-ETP.
The goal of this paper was to examine the effect on the outcome of the metal case studies
presented by Gloria et al. (2006) of using the new method for estimating freshwater
ecotoxicity. We calculated and compared metal CFs and LCIA outcomes for freshwater
ecotoxicity of each of the two studies using four models: USES-LCA 1.0 (Huijbregts et al.
2000), USES-LCA 2.0 (van Zelm et al. 2009), USEtoxTM
using the interim approach
86
(Rosenbaum et al. 2008), and the new method with the geochemical correction in USEtoxTM
(Gandhi et al. 2010). We also examined the implications of the choice of freshwater
chemistry in this context.
4.3 Methods
4.3.1 Case Studies
We revisited the case studies examined by Gloria et al. (2006), first a Cu-based product of Cu
pipe and second a Zn-based gutter system for an average residential application. For Cu pipe
used in a typical American house over its lifetime, Gloria et al. (2006) relied on the
comprehensive inventory compiled by Ecobalance (2000a, b) under commission to the
International Copper Association, that represented data from 31 sites and accounted for 58%
of the Cu production by refinery and 74% by solvent extraction. The inventory covered the
stages from mining and milling to pipe manufacturing. The second case study was for zinc-
copper-titanium alloy gutters which are used in 70% of the residential market in the
Netherlands. The Zn used is a special high grade (SHG) Zn with a purity of 99.995%. The
inventory data gathered under commission by TNO Environment, Energy and Process
Innovation (TNO-MEP) was from three manufacturers, representing 90% of the Dutch
market. The case study included the gutter, four end pieces of zinc-copper-titanium alloy,
and 16 support brackets made of galvanized steel with a 30 µm Zn layer applied to both
sides.
Using the same inventory data (Tables 4.1-4.2) reported by Gloria et al. (2006), we
calculated metal impact scores for freshwater ecotoxicity (Σ CF x emissions) using four
models listed above. We used these models because they share a similar structure and fate
processes that isolate the comparison of metal CFs and thus LCIA results for freshwater
ecotoxicity related to these case studies. We did not make any modifications to the
modelling structure, default parameterization of fate and exposure processes, or toxicity
endpoints in the models.
87
Table 4.1: LCI data for 1 kg of copper pipe with emission estimates by the processing stage
(source: Gloria et al. 2006).
No. Chemical Emission Total Emission
To (kg)
1 Ammonia (NH3) Air 5.15E-03
2 Arsenic (As) Air 6.06E-06
3 Cadmium (Cd) Air 2.26E-06
4 Carbon Dioxide (CO2, fossil) Air 3.69E+00
5 Carbon Monoxide (CO) Air 1.00E-02
6 Chlorides (CI-) Air 3.58E-07
7 Chromium (Cr III, Cr VI) Air 2.15E-06
8 Cobalt (Co) Air 1.32E-07
9 Copper (Cu) Air 1.23E-04
10 Hydrocarbons (except methane) Air 1.58E-03
11 Hydrocarbons (unspecified) Air 2.81 E-04
12 Lead (Pb) Air 2.05E-05
13 Mercury (Hg) Air 5.09E-08
14 Metals (unspecified) Air 1.08E-05
15 Methane (CH4) Air 9.49E-03
16 Nickel (Ni) Air 1.39E-06
17 Nitrogen Oxides (NOx as NO2) Air 1.85E-02
18 Nitrous Oxide (N2O) Air 3.64E-04
19 Particulates (PM 10) Air 2.34E-03
20 Particulates (unspecified) Air 1.35E-02
21 Silver (Ag) Air 1.07E-07
22 Sulfur Oxides (SOx as SO2) Air 1.52E-02
23 Sulfuric Acid (H2SO4) Air 6.02E-03
24 VOC (Volatile Organic Compounds) Air 6.06E-04
25 Zinc (Zn) Air 1.07E-05
26 Acids (H+) Water 6.22E-06
27 BOD5 (Biochemical Oxygen Demand) Water 2.21 E-04
28 Cadmium (Cd++) Water 4.64E-09
29 Chromium (Cr III, Cr VI) Water 2.76E-08
30 Cobalt (Co I, Co II, Co III) Water 9.70E-11
31 COD (Chemical Oxygen Demand) Water 9.52E-04
32 Copper (Cu+, Cu++) Water 1.96E-06
33 Cyanide (CN-) Water 1.86E-08
34 Lead (Pb++, Pb4+) Water 1.25E-07
35 Mercury (Hg+, Hg++) Water 3.44E-10
36 Nickel (Ni++, Ni3+) Water 6.93E-07
37 Silver (Ag+) Water 3.64E-12
38 Sulfate (SO4--) Water 1.41E-04
39 Suspended Matter (unspecified) Water 4. 14E-04
40 TOC (Total Organic Carbon) Water 5.04E-07
41 Zinc (Zn++) Water 8.45E-07
88
Table 4.2: LCI data for zinc gutter system reported by Gloria et al. (2006; original source:
Eggels et al. 2000).
No. Chemical Emission Total Emission
To (kg)
1 Ammonia (NH3) Air 1.41E-02
2 Carbon Dioxide (CO2, fossil) Air 8.70E+01
3 Carbon Monoxide (CO) Air 1.40E-01
4 Chlorides (CI-) Air 1.30E-05
5 Chromium (Cr III, Cr VI) Air 9.90E-07
6 Copper (Cu) Air 1.00E-06
7 Hydrocarbons (except methane) Air 4.22E-02
8 Hydrogen Chloride (HCl) Air 7.90E-03
9 Hydrogen Fluoride (HF) Air 1.60E-04
10 Lead (Pb) Air 3.76E-04
11 Metals (unspecified) Air 3.80E-04
12 Methane (CH4) Air 2.50E-01
13 Nickel (Ni) Air 9.90E-07
14 Nitrogen Oxides (NOx as NO2) Air 4.24E-01
15 Nitrous Oxide (N2O) Air 7.60E-05
16 Particulates (unspecified) Air 3.86E-01
17 Sulfur Oxides (SOx as SO2) Air 6.58E-01
18 VOC (Volatile Organic Compounds) Air 8.53E-03
19 Zinc (Zn) Air 1.94E-03
20 Aluminum Water 6.50E-03
21 Ammonium (NH4+) Water 1.40E-04
22 BOD5 (Biochemical Oxygen Demand) Water 2.18E-03
23 Chlorine (Cl-) Water 1.03E-01
24 COD (Chemical Oxygen Demand) Water 5.79E-03
25 Dissolved solids Water 1.80E-01
26 Hydrocarbons (CxHy) Water 2.00E-04
27 Iron (Fe) Water 7.00E-03
28 Nitrogen – tot Water 2.80E-03
29 Sodium (Na) Water 3.10E-02
30 Sulfate (SO4--) Water 4.16E-01
31 Suspended Matter (unspecified) Water 5.48E-01
32 Zinc (Zn++) Water 1.00E-03
89
4.3.2 Model Applications
4.3.2.1 USES-LCA 1.0
For five emission compartments, i.e. air, freshwater, sea water, industrial soil and agricultural
soil, USES-LCA 1.0 (Huijbregts et al. 2000) calculates chemical-specific CFs over an
infinite time horizon based on FFs and EFs for each compartment. FFs and EFs express the
change in the total dissolved concentration of a chemical in a compartment due to a change in
its emission (see Table 4.3). The nested multimedia fate model Simplebox 2.0 (Brandes et
al. 1996) was the basis of USES-LCA 1.0. Model compartments are defined on the
continental and global scales. For the case studies, we used the default environmental
properties of the freshwater compartment at the continental scale (Table 4.3). EFs were
calculated as the inverse of the Predicted No Effect Concentration (PNEC) of a chemical
(Huijbregts et al. 2000). A reference chemical 1,4-dichlorobenzene (DCB) was used to
normalize freshwater ecotoxicity of metals using the midpoint calculation method (see Table
4.3). Huijbregts et al. (2000) list CFs calculated for 19 metals normalized to the CF of 1,4-
DCB.
4.3.2.2 USES-LCA 2.0
USES-LCA 1.0 was updated to USES-LCA 2.0 (van Zelm et al. 2009). The latter model
considers 10 emission compartments, including urban air, rural air, freshwater, and
agricultural soil. USES-LCA 2.0 calculates chemical-specific CFs using FFs and EFs in
multiple compartments to assess ecotoxicological impacts over a default infinite time
horizon. Similarly to USES-LCA 1.0, FFs and EFs represent the change in total dissolved
concentration of a chemical in an environmental compartment due to the change in its
emission (Table 4.3). The updated model Simplebox 3.0 (Den Hollander et al. 2004) forms
the basis of USES-LCA 2.0. For the case studies, the default landscape properties of the
freshwater compartment at the continental scale were used to calculate metal CFs. Unlike
USES-LCA 1.0, EFs were calculated using a slope factor (typically 0.5), and a chemical-
specific toxic potency factor that reflects the toxicity of a chemical averaged over multiple
90
species (van de Meent and Huijbregts 2005). This value can be interpreted as the dissolved
concentration at which 50% of the species considered are “protected”. Again, 1,4-DCB was
used as a reference substance in the midpoint calculations to estimate freshwater ecotoxicity
of each metal (Table 4.3).
For USES-LCA 1.0 and USES-LCA 2.0 models, we also analyzed the effects of varying
time-scales to analyze environmental impacts after a change emission by comparing results
for 100 years versus an infinite time scale. This analysis is important because metals are
infinitely persistent in comparison to organic compounds which degrade over time (Pettersen
and Hertwich 2008, Huijbregts et al. 2001). One of the approaches proposed in LCA is to
separate short-term and long-term to infinite time horizons over which environmental
impacts are considered after emissions (e.g., Udo de Haes et al. 1999). This distinction of
impact period can allow LCA practitioners to use different fate expressions derived from
experimental results along with kinetic modelling for estimating mineralization and/or
weathering i.e. the metal release from and incorporation into solid mineral phases over a
specified time period. However, in reality this is rarely or perhaps never has been done
because of the paucity of kinetic data and the need to introduce more complexity into the
available models. The short-term period is often set at 100 years (e.g., Huijbregts et al. 2000,
Finnveden 1999), which is the time span we chose. The default model formulations and
parameter values for the 100 year time-period were taken from the original USES-LCA 1.0
and USES-LCA 2.0 models (Table 4.4).
4.3.2.3 USEtoxTM
As a result of the Life Cycle Initiative launched by the United Nations Environment Program
(UNEP) and the Society for Environmental Toxicology and Chemistry (SETAC) to
harmonize several LCIA toxicity characterisation models, the consensus model USEtoxTM
has been introduced as a parsimonious and transparent tool to provide CFs for ecotoxicity
and human health (Hauschild et al. 2008, Rosenbaum et al. 2008). The model formulation
was jointly finalized by the developers of CalTOX, IMPACT 2002, USES-LCA, BETR,
EDIP, WATSON and EcoSense. USEtoxTM
assesses toxicological effects of a chemical
91
emitted into a model compartment by considering the three steps of environmental fate,
exposure and effects. A chemical can be emitted into one of the five compartments (e.g., air,
freshwater, marine water, and natural and agricultural soil) at both continental and global
scales nested in the model structure. Urban air is added as a separate compartment at the
continental scale.
The model calculates CTP based on FF and EF (Table 4.3). The FF is calculated as the
change in total dissolved concentration of a chemical after its emission and represents the
compartment-specific residence time in days. The fate calculations differ from the previous
models by the inclusion of, for example, intermittent rain and an urban air compartment. The
calculation of EF is similar to that in USES-LCA 2.0. A geometric mean of laboratory-
derived single species EC50 values (water concentration at which 50% of a population
displays an effect), also known as HC50, is used to represent the concentration – response
relationship. Different than USES-LCA 2.0, measured chronic EC50 values are preferred,
however, in case of insufficient chronic data, acute data are used by applying an acute-to-
chronic extrapolation factor that is set to a default value of 2 (Rosenbaum et al., 2008). For a
consistent comparison of model results with those from USES-LCA 1.0 and USES-LCA 2.0,
we used 1,4-DCB as a reference chemical to normalize metal CFs (Table 4.3). Model
parameter values used in the calculation of metal CFs are summarized in Table 4.4.
4.3.2.4 USEtox New Method
As mentioned above, Gandhi et al. (2010, 2011) incorporated the recommendations of the
Clearwater Consensus to develop a modelling method for metals that accounts for the effect
of geochemical speciation on freshwater fate and toxicity. They did this by introducing a
bioavailability factor BF, into the calculation of CTP and thus CF. The method of Gandhi et
al. (2010, 2011) and incorporated in USEtoxTM
allows the LCA practitioner to specify water
chemistry by choosing among freshwater archetypes. The method has been evaluated for the
cationic metals Cd, Co, Cu, Ni, Pb, and Zn. Thus, to calculate a metal CTP for a water
archetype, the model calculates BF and LogKd values in WHAM 6.0 or another geochemical
model, FF using value of LogKd obtained from WHAM 6.0 in USEtoxTM
, and EF using the
92
average chronic ecotoxicity of a metal using a BLM (Di Toro et al. 2001) to normalize metal
bioavailability in toxicity tests relative to the water chemistry of the archetype. Details of the
modelling method are described by Gandhi et al. (2010). We used 1,4-DCB as a reference
chemical to normalize metal CFs for each freshwater chemistry (Table 4.3).
In this analysis we calculated CTPs and CFs for seven freshwater archetypes that we are
proposing based on our analysis of global freshwater systems (Gandhi et al. in prep-a). These
freshwater archetypes are diverse in terms of chemistry parameters (Table 4.5) and are
environmentally abundant. We assumed that the amount of total metal listed in the LCI was
the sum of its amount in total dissolved (or soluble) and particulate phases. The total
dissolved phase was further divided into the colloidal phase, which is mainly associated with
Dissolved Organic Carbon (DOC), and the truly dissolved fraction. We further assumed that
the bioavailable fraction of metal is within the truly dissolved fraction and is predominantly
the free metal ion (see Figure 1 in Diamond et al. 2010). We used the default database of
stability constants for metal complexes in WHAM 6.0 to calculate values of BF and LogKd
for each freshwater archetype (reported in Table 4.6). Default landscape properties of the
freshwater compartment at the continental scale in USEtoxTM
were used to calculate FFs.
Chronic metal-specific BLMs were used to calculate EC50, and then HC50 and EFs for each
archetype. BLMs are, however, either under development or not available for many metals,
several of which are listed in LCI data of the case studies. For those metals such as Cd, Co
and Pb, we used the Free Ion Activity Model (FIAM; Campbell 1995) to replace BLMs. In a
separate exercise (Gandhi et al. in prep-b), we showed that the estimates of EF from FIAM
are comparable to those from BLM for metals for which BLMs are currently available (e.g.,
Cu, Ni and Zn). Our results showed that the largest gain in accuracy using the new method is
achieved by correcting for metal bioavailability (which if not corrected can result in
differences of up to ~3-4 orders of magnitude) than by the choice of method by which the
bioavailability correction is made (which can change the results by within 1 order of
magnitude).
93
Table 4.3: Comparisons of calculation methods and model parameters of four LCIA models used in the freshwater ecotoxicity
assessment of case studies.
Model CTP/CF FF EF BF
USES-LCA 1.0
CTPMe = FF × EF
CFMe = CTPMe/CTP1,4-DCB
FF was calculated for total dissolved
metal using the default continental
freshwater landscape parameters in the
model. Values of LogKd (suspended
particles – water partitioning
coefficient) and total dissolved
fractions of the metals are listed in
Table 4.4.
EF was calculated for total dissolved
metal using the PNEC (EF =
1/PNEC). PNEC is the predicted no
effect concentration of toxicity data
which are listed for each metal in
Table 4.4.
Not Applicable
USES-LCA 2.0
CTPMe = FF × EF
CFMe = CTPMe/ CTP1,4-DCB
FF was calculated for total dissolved
metal using the default continental
freshwater landscape parameters in the
model. LogKd (suspended particles –
water partitioning coefficient) and total
dissolved fractions of the metals are
listed in Table 4.4.
EF was calculated for total dissolved
metal using the HC50 values (EF =
0.5/HC50). HC50 is the geometric
mean of toxicity data which are listed
for each metal in Table 4.4.
Not Applicable
USEtoxTM
CTPMe = FF × EF
CFMe = CTPMe/ CTP1,4-DCB
FF was calculated for total dissolved
metal using the default continental
freshwater landscape parameters in the
model. LogKd (suspended particles –
water partitioning coefficient) and total
dissolved fractions of the metals are
listed in Table 4.4.
EF was calculated for total dissolved
metal using the HC50 values (EF =
0.5/HC50). HC50 is the geometric
mean of toxicity data which are listed
for each metal in Table 4.4.
Not Applicable
USEtox(new
method)
CTPMe = FF × EF × BF
CFMe = CTPMe/ CTP1,4-DCB
FF was calculated for total metal using
the default continental freshwater
landscape parameters in the USEtoxTM
.
LogKd of metals were calculated using
WHAM 6 with the chemistry of
freshwater archetype (Table 4.5).
EF was calculated for truly dissolved
metal using the HC50 values (EF =
0.5/HC50). HC50 is the geometric
mean of toxicity data corrected for
bioavailability of metals in water
chemistry archetypes using BLM.
Estimated HC50 values for each metal
are listed in Table 4.5.
BF was calculated as a
fraction of truly
dissolved metal in total
metal estimated using
WHAM 6 model and the
water chemistry of
freshwater archetypes
(Table 4.5).
94
Table 4.4: Default model parameter values required for the calculation of metal CFs as reported in the original USES-LCA 1.0,
USES-LCA 2.0 and USEtoxTM
(see Table 4.3).
Metal
USES-LCA 1.0 USES-LCA 2.0 USEtoxTM
LogKd
(L/kg)
Diss Fr (-)
PNEC
(mg/L)
LogKd
(L/kg)
Diss
Fr (-) HC50
(mg/L)
LogKd
(L/kg)
Diss
Fr (-) HC50
(mg/L)
Ag - - - 5.0 0.40 2.5E-02 5.0 0.41 7.7E-02
Cu 4.7 0.44 1.1E-03 4.7 0.57 1.6E-01 4.7 0.33 4.4E-01
Cd 5.1 0.24 3.4E-04 5.1 0.34 8.6E-01 4.9 0.45 1.9E+00
Co 3.6 0.91 2.6E-03 3.6 0.95 4.0E+00 4.6 0.59 2.0E+00
Cr (III) 5.5 0.12 3.4E-02 5.5 0.19 3.9E+00 4.6 0.59 7.9E+00
Cr (VI) 5.5 0.12 8.5E-03 5.5 0.19 3.9E+00 5.1 0.35 8.5E+00
Ni 3.9 0.83 1.8E-03 3.9 0.89 8.8E-01 4.2 0.81 1.4E+00
Hg 5.2 0.19 2.3E-04 5.2 0.28 6.3E-02 4.4 0.73 3.2E+00
Pb 5.8 0.058 1.1E-02 5.8 0.094 3.8E+00 5.3 0.19 3.1E-01
Zn 5.0 0.27 6.6E-03 5.0 0.38 1.2E+00 2.7 0.61 7.5E+00
95
Table 4.5: Ambient chemistry for freshwater archetypes used to calculate CFs of metals using the new framework proposed by
Gandhi et al. (2010).
Freshwater Category Example Ecosystem pH DOC Hardness Ca Mg Na K SO4 Cl
archetypes pH DOC Hardness mg/L mgCaCO3/L mg/L mg/L mg/L mg/L mg/L mg/L
Archetype 1 High Med High Canals, large & small lakes 8.1 8.4 221 56.6 19.5 65.8 0.1 67 120
Archetype 2 High Med Med Mole, United Kingdom 7.6 6.1 132 42.48 6.22 26.67 3.52 48.03 32.97
Archetype 3 High Low Med Segrino, Italy 8.2 1.7 169 58.51 5.59 2.60 0.78 9.61 20.92
Archetype 4 Med High Med Ankeveen, Netherlands 7.3 17.8 165 52.10 8.58 11.79 0.82 109.51 20.21
Archetype 5 Med Low Med Small springs 6.7 2.2 78 20.3 6.7 17 0.1 67 31
Archetype 6 Med Low Low Somerain, Belgium 6.4 1.6 28 6.69 2.65 7.20 2.82 85.50 5.99
Archetype 7 Low Med Low Bihain, Belgium 5.9 8.9 10 2.48 0.95 6.39 1.80 2.88 8.37
Table 4.6: Estimated metal bioavailable fractions (BFs, dimensionless), LogKd (L/kg) and average chronic toxicity (HC50; mg/L)
values corrected for the speciation of metals in various freshwater archetypes used in the analysis of USEtox(new) method as
discussed in Table 4.3.
Freshwater
Archetypes LogKd BF HC50 LogKd BF HC50 LogKd BF HC50 LogKd BF HC50 LogKd BF HC50 LogKd BF HC50
Archetype-1 4.2 4.4E-06 3.6E-04 4.1 5.2E-02 1.4E-01 5.0 7.1E-02 7.6E-02 4.9 2.0E-01 1.7E-02 4.2 9.1E-01 1.5E-01 5.2 4.4E-03 1.9E-01
Archetype-2 4.2 3.3E-05 7.6E-04 4.1 1.1E-01 9.3E-02 4.9 1.3E-01 8.7E-02 4.9 3.1E-01 1.5E-02 4.5 8.6E-01 2.3E-01 5.6 1.8E-03 9.1E-02
Archetype-3 4.1 1.1E-04 1.0E-04 3.6 1.4E-01 1.1E-01 5.4 3.1E-01 6.4E-02 5.4 5.0E-01 1.1E-01 4.2 9.8E-01 2.0E-01 5.9 3.0E-02 2.9E-01
Archetype-4 4.3 1.2E-05 8.1E-04 4.2 5.4E-02 1.0E-01 4.6 6.0E-02 1.1E-01 4.5 1.9E-01 1.4E-02 4.1 7.7E-01 2.1E-01 4.7 8.9E-04 2.9E-01
Archetype-5 4.1 1.1E-03 1.0E-03 3.8 4.6E-01 7.3E-02 4.8 5.2E-01 1.2E-01 4.7 6.9E-01 1.1E-02 4.7 9.2E-01 2.9E-01 6.6 5.8E-03 6.9E-02
Archetype-6 4.0 3.1E-03 7.8E-04 3.7 5.2E-01 4.9E-02 4.7 5.7E-01 1.1E-01 4.8 7.1E-01 1.4E-02 5.3 7.8E-01 4.5E-01 7.0 2.1E-03 8.9E-02
Archetype-7 4.0 5.7E-04 7.5E-04 4.1 1.6E-01 3.7E-02 4.8 2.0E-01 1.1E-01 4.8 2.4E-01 1.5E-02 4.6 6.1E-01 5.9E-01 6.1 8.7E-03 9.2E-02
Cu Ni Zn Cd Co Pb
96
For both BLM and FIAM calculations, aquatic species-specific chronic effects data (e.g.,
EC50) for major trophic- or taxonomic groups were taken from the literature. The geometric
mean or HC50 of species-specific EC50 values corrected for metal bioavailability was used to
calculate an archetype-specific EF (Tables 4.3, 4.6). We used the same trophic/taxonomic
group to obtain chemistry-corrected values of HC50 for all archetypes, e.g., the ecosystem
structure was assumed to be the same for all archetypes. This is a weakness of the modelling
approach since ecosystem composition strongly depends on local environmental conditions,
such as aquatic chemistry and tolerance developed by organisms to continuous exposure to
metals over the long term (e.g., Forbes and Calow 2002).
4.3.3 Scope and Assumptions
We limited the scope of this study to evaluating only the freshwater ecotoxicity potential of
metals quantified in the LCIs of the two case studies. This exercise was not intended to
compare the relative human health and environmental performance of Cu pipe or Zn gutters
and/or to provide/support information for decision-making directly related to these case
studies. Rather we investigated the effect of considering metal-specific chemistry in the
context of CFs for freshwater ecological toxicity. In line with this scope, we also did not
consider the emissions of organic chemicals in this analysis. Finally, we further limited the
scope of the study to metals – Cd, Cu, Co, Pb, Ni, and Zn - for which USEtox(new) is
currently applicable. This limitation was imposed by insufficient toxicity test data to run
either chronic BLM or FIAM for As, Au, Ag, Cr, and Hg that would permit us to correct for
the bioavailability of these metals in the freshwater archetypes. It is important to note here
that none of the models considered in this study can calculate CFs for Al and Fe, and
therefore these metals were also excluded from the analysis.
The following are the major assumptions common to all model applications for analyzing
these case studies. Both inventories specify most metal emissions to air followed by
emissions to water. Differing percentages of metal emissions to air are transferred to water
according to model FFs. Here we use freshwater ecotoxicity CFs for the fraction of metals
emitted to air that is transferred to water, as well as metals emitted directly to water. The
97
ecotoxicity of metals emitted to other compartments are not included in this discussion
because modelling methods for these other compartments have not been updated.
Further, we assumed that metals listed in the LCIs of both case studies were emitted in a
soluble and labile form (e.g., Me+2
) that is readily available to complex with various
inorganic and organic natural ligands present in the environment. This is a critical
assumption in terms of assessing total bioavailability and fate of emitted metals because
often a fraction of metal is emitted as a non-reactive, insoluble native metal or metal
composite product that is not subject to multimedia transport and is not bioavailable.
Therefore, this assumption may lead to overestimation of both BF and FF, and thus CTP.
The ecotoxicity characterization of insoluble metal compounds requires the use of an
additional model or procedure to estimate metal dissolution (e.g., Skeaff et al. 2000),
however incorporating such details in LCIA calculations is often constrained by the LCI data
that do not specify the forms of metal emitted into the environment.
4.4 Results and Discussion
4.4.1 Comparison of Metal CFs
First we compared the previously reported metal CFs using USES-LCA 1.0 (e.g., Huijbregts
et al. 2000) recommended for use in LCIA with those estimated using the USEtox(new)
method. The previous method ranked metals amongst the most toxic chemicals in terms of
both effect thresholds and time-integrated toxicity (Huijbregts et al. 2000, Payet and Jolliet
2002). The range of archetype-specific CFs calculated using the USEtox(new) method were
consistently lower by up to 3 orders of magnitude (e.g., Cu), than the previous values (Table
4.7). The greatest difference in the new and previous CFs was found for Cu, followed by Ni,
Co and Pb (within 2 orders of magnitude), whereas the least difference was observed for Cd
and Zn (1 order of magnitude). The range of variability of new metal CFs was greatest for
Cu, followed by Ni and Pb (2 orders of magnitude) and finally for Zn, Co and Cd (within 1
order of magnitude). A larger range in variability of CFs for Cu, Ni and Pb illustrates the
importance of considering metal speciation and bioavailability while conducting LCIA of
metal emissions.
98
Table 4.7: Comparison of previously reported (USES-LCA 1.0; Huijbregts et al. 2000) and
a range of archetype-specific metal CFs (kg eq. 1,4-DCB) calculated using the method of
Gandhi et al. (2010, 2011) for use in metal LCIA.
Metal Previous CFs
USES-LCA 1.0
Range of archetype-specific CFs
USEtox(new)
Cd 1.5 x 103 6 x 10
1 – 4.7 x 10
2
Co 3.4 x 103 1.7 x 10
1 – 8.8 x 10
1
Cu 1.2 x 103 2.8 x 10
-1 – 1.2 x 10
2
Pb 9.6 6.8 x 10-2
– 4.6
Ni 3.2 x 103 9.5 – 4.1 x 10
2
Zn 9.2 x 101 7.5 – 5.9 x 10
1
Next we compared the CFs estimated using the other models considered in this study. Metal
CFs varied ~3 orders of magnitude for all metals relevant to the case studies within one
model, and ~3 orders of magnitude for each metal across the models (Table 4.8). The lowest
CF was consistently estimated for Pb, however, the highest CF differed depending on the
model. For example, the highest CFs from USES-LCA 1.0 and USES-LCA 2.0 were for Co
and Cu, respectively. Cu had the highest CF in USEtoxTM
, whereas Cd consistently had the
highest CF in USEtox(new) calculations for all freshwater archetypes. Note that we did not
consider Ag for this analysis in absence of chronic BLMs that prevented us from calculating
its freshwater archetype-specific EFs in the USEtox(new) model calculations. However, if
Ag were included in this analysis then its CF was highest in USES-LCA 2.0 and USEtoxTM
(~ 3.5 times higher than Cu; results not shown).
The differences amongst CFs can be related back to the choice of Kd’s for all models except
USEtox(new) for which Kd’s were explicitly calculated according to specified water
chemistries. More importantly, EFs differed as a result of differing calculation methods used
that spanned the use of PNEC for USES-LCA 1.0 to HC50 for the other models.
USEtox(new) corrects the HC50 for chemistry-specific bioavailability, unlike the other
models. For example, values of LogKd for Co was 3.6 for the USES-LCA models but was
99
4.6 for USEtoxTM
, and for Zn was 5.0 in both USES-LCA models and 2.7 for USEtoxTM
(Table 4.4). In comparison, values of LogKd calculated for USEtox(new) for Co and Zn
were 4.1-5.3 and 4.6-5.4, respectively, depending on the archetype (Table 4.6). Examples of
differences among models for PNEC and HC50 are also listed in Tables 4.4 and 4.6. For Ni
these values ranged from 0.0018 (USES-LCA 1.0), 0.88 (USES-LCA 2.0), 1.4 (USEtoxTM
)
to 0.0372 to 0.1 (USEtox(new)).
The consideration of time scale (i.e., infinite vs. 100 yrs) in both USES-LCA 1.0 and USES-
LCA 2.0 calculations did not significantly change the magnitude of CFs and thus relative
ranking of metals for their potential to cause toxicity in freshwater (Tables 4.8, 4.9). In
general, CFs for 100 years impact period were similar or slightly lower than those for infinite
time. This insignificant difference may be because both models treat long-term (infinite time
scale) release of metals in the same way as for the short-term emissions; all models fail to
consider slow, kinetically driven geochemical processes over time such as weathering and
mineralization. However, time scales become important when considering CFs for metals
and organics where the latter have finite persistence versus the infinite persistence of metals.
Next we analyzed the magnitude and relative ranking of metal CFs as a result of varying
freshwater chemistry (Tables 4.8, 4.9). The relative ranking of chemicals is often more
important than absolute values due to comparative nature of LCIA analysis. According to
the previous CFs (USES-LCA 1.0; Huijbregts et al. 2000), Co and Ni were most toxic
followed by Cd and Cu, whereas Pb and Zn were the least toxic among these metals (Table
4.7). The metal ranking In USEtox(new) was a function of the effects of freshwater
chemistry on metal speciation and bioavailability. USEtox(new) method ranked Cd and Pb
as the most and least toxic metals, respectively (Table 4.9). The order of metal ranking
between these two extremes changed from one archetype to another (Table 4.9), but Ni and
Co were generally more toxic than Cu and Zn. For example, in archetype-1 (hard, alkaline
water) the trend in CF and thus the ecotoxicity potential was Cd > Co > Ni > Zn > Cu > Pb,
whereas the pattern in archetype -5 (soft, acidic water) was Cd > Ni > Co > Cu ≈ Zn > Pb
(Table 4.9). The CFs in USEtox(new) method are largely controlled by the impact of
bioavailability on EF which depends on metal speciation in a specified freshwater chemistry
(Gandhi et al. 2010, 2011). The bioavailability of metals is often higher in systems with low
100
pH, DOC and hardness. The effect of these chemistry parameters is different for each metal
as it depends on a metal’s inherent geochemical characteristics. For example, Cu speciation
is governed by the presence and amount of DOC in the freshwater, whereas Zn speciation is
mainly controlled by the acidity of water. Use of the new, archetype-specific CFs could
significantly alter the outcome of LCIA studies that previously scored high ecotoxicity
impacts for metal emissions. Further, the change in relative ranking of metals from one
archetype to another also could change the type of metal that would need attention for
reducing the overall freshwater ecotoxicity impact for a process/system.
The other differences for CFs is the large range and variability of estimates (e.g., USES-
LCA with values in the order of 103, USEtox
TM in 10X, and USEtox(new) ranging from10
-2
to 10-4
) which influences not only the ranking of metals considered here, but also the ranking
of these metals with respect to organic compounds (see Table 4.10).
The differences in metal CFs for various models and freshwater archetypes can be easily
summarized if we calculate total (sum) ecotoxicity of metals assuming a unit emission of
each metal (Figure 4.1). These results show that for the same emission data, the overall
ecotoxicity of these metals differs according to the model used. In USES-LCA 1.0, Ni and
Co contributed the ~35% each towards total ecotoxicity, followed by Cd (~15%) and Cu
(~10%). In USES-LCA 2.0, the total ecotoxicity was mainly due to Cu and Ni. Thus, Cu
became significantly more important in the LCIA using USES-LCA 2.0. In contrast to both
USES-LCA models, Cu and Zn contributed more towards total freshwater ecotoxicity when
using USEtoxTM
. Here the contribution of Zn is surprisingly high at 35%, which decreases
the relative contribution of Ni towards total ecotoxicity. The high CF estimate of Zn in
USEtoxTM
model is due to the use of a low LogKd value which results in a higher dissolved
fraction (61%) relative to the other models.
In USEtox(new) method, Cd contributed up to 65% to total ecotoxicity scores, followed by
Co and Ni in all archetypes in contrast to Zn, Cu and Pb that contributed minimally to total
toxicity (Figure 4.1).
101
Table 4.8: Comparison of metal CFs (kg eq. 1,4-DCB) estimated for freshwater ecotoxicity using four LCIA models, time-scales of
infinity and 100 years of environmental impacts after metal emissions, and for seven freshwater types as mentioned in the text.
Metal
USES-LCA 1 USES-LCA 2
USEtoxTM
USEtox(new)
Infinite 100 Yrs Infinite 100 Yrs Arche
type-1
Arche
type-2
Arche
type-3
Arche
type-4
Arche
type-5
Arche
type-6
Arche
type-7
Cd 1.52E+03 1.51E+03 9.05E+00 7.13E+00 9.88E+00 1.02E+02 1.65E+02 1.56E+02 1.67E+02 4.73E+02 4.57E+02 1.51E+02
Co 3.41E+03 3.38E+03 3.30E+01 3.31E+01 4.17E+00 8.34E+01 5.19E+01 8.72E+01 7.79E+01 4.17E+01 1.66E+01 3.31E+01
Cu 1.16E+03 1.15E+03 1.18E+02 1.01E+02 5.62E+01 2.81E-01 1.01E+00 2.58E+01 3.01E-01 2.89E+01 1.17E+02 2.07E+01
Pb 9.62E+00 9.62E+00 4.14E-01 2.93E-01 3.81E-01 2.43E-01 8.42E-02 1.64E+00 9.28E-02 4.15E-01 6.77E-02 5.56E-01
Ni 3.24E+03 3.22E+03 9.84E+01 9.59E+01 1.51E+01 9.51E+00 3.20E+01 5.57E+01 1.11E+01 2.16E+02 4.12E+02 1.12E+02
Zn 9.17E+01 9.11E+01 7.52E+00 6.01E+00 3.92E+01 7.53E+00 1.29E+01 2.56E+01 7.93E+00 4.34E+01 5.93E+01 1.83E+01
Table 4.9: The relative importance of metals (in an increasing order) based on the numerical ranking of estimated CFs towards the
freshwater ecotoxicity in LCIA.
USES-LCA1 USES-LCA2
USEtoxTM
USEtox(new)
Infinite 100
Yrs Infinite
100
Yrs
Arche
type-1
Arche
type-2
Arche
type-3
Arche
type-4
Arche
type-5
Arche
type-6
Arche
type-7
Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb Pb
Zn Zn Zn Zn Co Cu Cu Zn Cu Cu Co Zn
Cu Cu Cd Cd Cd Zn Zn Cu Zn Co Zn Cu
Cd Cd Co Co Ni Ni Ni Ni Ni Zn Cu Co
Ni Ni Ni Ni Zn Co Co Co Co Ni Ni Ni
Co Co Cu Cu Cu Cd Cd Cd Cd Cd Cd Cd
102
Table 4.10: Relative ranking in the order of low to high ecotoxicity potential for organic chemicals and metals based on the CFs
calculated in each model. Note that CFs for the organics used in the relative ranking of USEtox(new) approach are from USEtoxTM
.
Ranking USES-LCA 1 USEtoxTM
Infinite Infinite Archetype-1 Archetype-2 Archetype-3 Archetype-4 Archetype-5 Archetype-6 Archetype-7
1 Trichloromethane Trichloromethane Lead Lead Lead Lead Lead Lead Lead
2 Benzene Benzene Copper Copper Zinc Copper Copper Cobalt Zinc
3 Ethylbenzene Ethylbenzene Zinc Zinc Copper Zinc Trichloromethane Trichloromethane Copper
4 1,3,5-Trichlorobenzene Formaldehyde Nickel Nickel Trichloromethane Nickel Cobalt Zinc Cobalt
5 Lead Lead Tetrachloromethane Tetrachloromethane Nickel Trichloromethane Zinc Benzene Trichloromethane
6 Pentachlorobenzene 2 Chlorophenol Benzene Cobalt Benzene Benzene Benzene Copper Benzene
7 Zinc 1,3,5-Trichlorobenzene Cobalt Benzene Cobalt Cobalt Ethylbenzene Ethylbenzene Nickel
8 Hexachlorobenzene Naphthalene Cadmium Cadmium Cadmium Cadmium Nickel Formaldehyde Cadmium
9 Formaldehyde Cobalt Ethylbenzene Ethylbenzene Ethylbenzene Ethylbenzene Formaldehyde Nickel Ethylbenzene
10 Phenanthrene Cadmium Formaldehyde Formaldehyde Formaldehyde Formaldehyde Cadmium Cadmium Formaldehyde
11 Naphtalene Benzo[a]pyrene 2 Chlorophenol 2 Chlorophenol 2 Chlorophenol 2 Chlorophenol 2 Chlorophenol 2 Chlorophenol 2 Chlorophenol
12 Copper Nickel 1,3,5-Trichlorobenzene 1,3,5-Trichlorobenzene 1,3,5-Trichlorobenzene 1,3,5-Trichlorobenzene 1,3,5-Trichlorobenzene 1,3,5-Trichlorobenzene 1,3,5-Trichlorobenzene
13 Cadmium Phenanthrene Naphtalene Naphtalene Naphtalene Naphtalene Naphtalene Naphtalene Naphtalene
14 2 Chlorophenol Pentachlorobenzene Benzo[a]pyrene Benzo[a]pyrene Benzo[a]pyrene Benzo[a]pyrene Benzo[a]pyrene Benzo[a]pyrene Benzo[a]pyrene
15 Nickel Zinc Phenanthrene Phenanthrene Phenanthrene Phenanthrene Phenanthrene Phenanthrene Phenanthrene
16 Cobalt Copper Pentachlorobenzene Pentachlorobenzene Pentachlorobenzene Pentachlorobenzene Pentachlorobenzene Pentachlorobenzene Pentachlorobenzene
17 Atrazine Atrazine Atrazine Atrazine Atrazine Atrazine Atrazine Atrazine Atrazine
18 DDT Hexachlorobenzene Hexachlorobenzene Hexachlorobenzene Hexachlorobenzene Hexachlorobenzene Hexachlorobenzene Hexachlorobenzene Hexachlorobenzene
19 Benzo[a]pyrene DDT DDT DDT DDT DDT DDT DDT DDT
USEtox(new) - archetypal approach
103
Figure 4.1: Relative contribution of metals towards total freshwater ecotoxicity potential
(toxicity impact indicator) based on the model-specific CFs (1,4-DCB eq.) if an unit emission
of each of these metals occurs to the freshwater environment.
0%
20%
40%
60%
80%
100%
Infinite 100
Yrs
Infinite 100
Yrs
Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
USES-LCA1 USES-LCA2 USEtox
(prev)
USEtox (new)
Cd Co Cu Pb Ni Zn
4.4.2 Freshwater Ecotoxicity of Case Studies
4.4.2.1 Case Study: Cu Pipe
The results for the Cu case study were significantly different among the six model
approaches. During the processing of Cu pipe, the total estimated incremental metal
emissions to freshwater ranged >3 orders of magnitude from <0.1 g to <0.1 mg in the order
of Cu > Pb > Zn > As > Ni > Cd ≈ Cr > Co > Ag > Hg (Figure 4.2). This incremental
emission was dominated by the transport of each metal from air to freshwater. Metal
104
emissions to air were consistently higher by a factor of 2 (Ni) to 10000 (Ag) than to
freshwater in the LCI. These emissions were transported from air to freshwater.
1.E-8
1.E-7
1.E-6
1.E-5
1.E-4
1.E-3
Ars
enic
Cad
miu
m
Chr
omiu
m
Cob
alt
Cop
per
Lead
Mer
cury
Nic
kel
Silver
Zinc
Figure 4.2: Total estimated metal emission (kg) that will eventually end up in the freshwater
compartment due to the release of metals to air and water during the processing of Cu pipe
considered in the case study.
Total freshwater ecotoxicity scores for the emissions of metals to both air and water
compartments (Table 4.1) also ranged >3 orders of magnitude depending on the model and
water chemistry considered, but were dominated by emissions to air that were transferred to
freshwater (Figure 4.3). Cu emissions mostly to air and then to water contributed 70 – 94%
towards the total freshwater ecotoxicity scores for USES-LCA 1.0, USES-LCA 2.0 and
USEtoxTM
models (Table 4.11). Consistently, Co and Pb emissions to water contributed the
least (<0.01%) towards the total ecotoxicity scores for all models. Although Cu contributed
the most for USES-LCA 1.0 and USES-LCA 2.0 models, there were significant differences
105
in results for the two impact periods considered (Table 4.11). Cu emissions to air dominated
the overall ecotoxicity scores for the infinite time scale, whereas Cu emissions to both air and
water contributed equally at ~40% for the 100 year impact period. Cu, Ni emissions to water
also contributed significantly (13-25%) towards ecotoxicity in the analysis of 100 years of
metal impact.
Table 4.11: Percentage contribution of metal emissions to air and water towards the total
freshwater ecotoxicity estimated for Cu pipe case study. Note that six metals for which new
CFs are currently available were considered in this analysis.
USEtox
Infinite 100 Yrs Infinite 100 Yrs (prev) Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
Cd (air) 1.94% 1.34% 0.11% 0.09% 0.29% 51.33% 42.02% 5.31% 62.29% 12.22% 3.58% 6.15%
Co (air) 0.25% 0.44% 0.04% 0.07% 0.01% 2.71% 0.85% 0.19% 1.88% 0.07% 0.01% 0.09%
Cu (air) 80.87% 44.44% 80.07% 44.91% 93.40% 13.55% 24.55% 83.81% 10.77% 71.27% 87.13% 80.65%
Pb (air) 0.15% 0.03% 0.09% 0.02% 0.12% 1.78% 0.31% 0.81% 0.50% 0.15% 0.01% 0.33%
Ni (air) 2.60% 1.11% 0.68% 0.49% 0.28% 3.01% 5.10% 1.19% 2.61% 3.49% 2.02% 2.87%
Zn (air) 0.56% 0.25% 0.41% 0.27% 5.88% 19.89% 17.20% 4.54% 15.50% 5.87% 2.42% 3.92%
Cd (water) 0.02% 0.08% 0.00% 0.01% 0.00% 0.26% 0.21% 0.03% 0.32% 0.06% 0.02% 0.03%
Co (water) 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%
Cu (water) 6.73% 25.84% 14.05% 39.77% 0.02% 0.30% 0.55% 1.88% 0.24% 1.60% 1.95% 1.81%
Pb (water) 0.00% 0.01% 0.00% 0.01% 0.00% 0.02% 0.00% 0.01% 0.00% 0.00% 0.00% 0.00%
Ni (water) 6.65% 25.57% 4.16% 13.33% 0.00% 3.64% 6.17% 1.43% 3.15% 4.22% 2.44% 3.47%
Zn (water) 0.23% 0.88% 0.39% 1.02% 0.00% 3.51% 3.04% 0.80% 2.74% 1.04% 0.43% 0.69%
USES-LCA1 USES-LCA2 USEtox (new)Metals
The relative contribution of each metal to ecotoxicity was different for all the models (Figure
4.4a,b). Cu contributed 72-89% towards the total ecotoxicity for USES-LCA 1.0 and 2.0,
and USEtoxTM
due to its emission to air and (less so) water. The same was true for
USEtox(new) for the four archetypes – 3, 5, 6 and 7 (Table 4.11). These systems generally
have high bioavailability of Cu mainly due to low DOC and pH (Table 4.5). In contrast, Cd
(42-62%) followed by Zn (15-20%) emissions to air contributed the most towards total
ecotoxicity for archetypes-1, 2, and 4 (Table 4.11, Figure 4.4a). These archetypes are
characterized by high pH, DOC and hardness. Similarly to other models, metal emissions to
air dominated the total ecotoxicity; however Ni and Zn emissions to water were also
important contributors to USES-LCA 1.0 and USEtox(new) (Table 4.11, Figure 4.4b).
106
∑ F
resh
wate
r E
co
tox
icit
y (eq. 1
,4-D
CB
)
0.00001
0.0001
0.001
0.01
0.1
1
Infinite 100 Yrs Infinite 100 Yrs Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
USES-LCA1 USES-LCA2 USEtox (prev)
USEtox (new)
∑ Water ∑ Air
Figure 4.3: LCIA results presented as the total freshwater ecotoxicity score of metals
estimated for the Cu case study. Here the ecotoxicity was estimated for total emission of
metals to freshwaters due to release of metals in both air and water compartments during the
processing of Cu pipe (see Table 4.1).
4.4.2.2 Case Study: Zn Gutter
The results of the Zn gutter case study were more consistent and clear amongst models than
those for the Cu case study since the inventory was 92% Zn emissions when Al and Fe were
excluded from this analysis. Total emissions of each metal to freshwater spanned >4 orders
of magnitude (<10 g to <1 mg) from Al and Fe, through to Zn and Pb; emissions of Cr, Cu
and Ni were low (Table 4.2, Figure 4.5). Zn emissions were 2 times higher to air than to
freshwater. Cr, Cu, Pb and Ni were emitted to air only, whereas Al and Fe were emitted to
freshwater only (Table 4.2). As mentioned earlier, none of the models considered in this
study was able to calculate CFs for Al and Fe and therefore they were omitted.
107
0%
20%
40%
60%
80%
100%
Infinite100 Yrs Infinite 100 Yrs Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
Cd Co Cu Pb Ni Zn
0%
20%
40%
60%
80%
100%
Infinite100 Yrs Infinite 100 Yrs Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
USES-LCA1 USES-LCA2 USEtox (prev)
USEtox (new)
a
b
Figure 4.4: Relative contribution of each metal emitted in (a) air and (b) water as listed in
LCI towards the total freshwater ecotoxicity score for the LCIA of Cu case study.
108
1.E-7
1.E-5
1.E-3
1.E-1
Aluminum Chromium Copper Iron Lead Nickel Zinc
Figure 4.5: Total estimated metal emission (kg) that will reach the freshwater compartment
due to the release of metals to air and water in the case study of Zn gutter system.
The variability in total ecotoxicity spanned nearly 4 orders of magnitude both due to
differences in the model approaches and freshwater chemistry of the receiving environment
(Figure 4.6). The highest and lowest total ecotoxicity scores were estimated by USEtoxTM
and USES-LCA 1.0 models, respectively. Both USES-LCA 1.0 and USES-LCA 2.0 had
comparable estimates for the infinite and 100 years time-scales of impacts. For
USEtox(new) the highest scores were for archetypes-6 followed by -5 (Figure 4.4); these
freshwaters are characterized by low pH and low hardness (Table 4.5). The lowest
ecotoxicity scores were estimated for freshwater archetypes-1 and -4 (Figure 4.4) which have
high hardness and pH values above circumneutral and thus low CFs of Zn, Pb and Cu (Table
4.8). Zn contributed ~ 98% towards the total scores in all models and freshwater archetypes
because it completely dominated the inventory (Figure 4.7). Cu, Pb and Ni were
109
approximately equal contributors within the fraction (< 2%) contributed by other metals. For
USES-LCA models, Zn emissions to water contributed the most towards total ecotoxicity,
whereas in USEtoxTM
Zn emission to air contributed the most (Figure 4.7). In contrast, Zn
emissions to both air and water contributed equally towards total ecotoxicity for all
freshwater archetypes. The differences in freshwater chemistry had negligible effects on the
overall LCIA of the Zn case study.
∑ F
resh
wate
r E
coto
xic
ity
(eq. 1
,4-D
CB
)
0.001
0.01
0.1
1
10
Infinite 100 Yrs Infinite 100 Yrs Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
USES-LCA1 USES-LCA2 USEtox (prev)
USEtox (new)
∑ Water ∑ Air
Figure 4.6: LCIA results presented as the total freshwater ecotoxicity score of metals
estimated for the Zn case study. Here the ecotoxicity was estimated for total emission of
metals to freshwaters due to release of metals in both air and water compartments as listed in
LCI data for Zn gutter system (see Table 4.2).
110
0%
20%
40%
60%
80%
100%
Infinite 100
Yrs
Infinite 100
Yrs
Arch-1 Arch-2 Arch-3 Arch-4 Arch-5 Arch-6 Arch-7
USES-LCA1 USES-LCA2 USEtox
(prev)
USEtox (new)
Cu (air) Pb (air) Ni (air) Zn (air) Zn (wat)
Figure 4.7: Contribution of each metal listed in LCI of Zn gutter system towards the total
freshwater ecotoxicity score in the analysis of its LCIA.
4.4.3 Comparisons with Previous Case Study Results
Comparisons of the model results from this study with those previously reported by Gloria et
al. (2006) revealed up to 4 orders of magnitude differences in total ecotoxicity estimates for
both case studies depending on the model used. For the Cu pipe case study, although Cu
consistently had the highest ecotoxicity in all models considered by Gloria et al. (2006), Zn
(in CalTOX), Cd (in EDIP 97), and Zn and Cd (in EI 99 HA) had low but significant
contributions. The results of USEtox(new) in this exercise also suggested important
contributions to ecotoxicity of Zn, Cd and Ni, in addition to Cu. For the Zn gutter case study,
Gloria et al. (2006) only considered Pb and Zn in their comparative analysis due to several
limitations of the models to provide CFs for other metals.
111
4.4.4 Improvements in USEtox(new) Approach
As with all models of natural systems, the performance of the USEtox(new) approach in the
context of LCIA has not been, and nor it can be, rigorously evaluated (Oreskes et al. 1994).
Verification and validation of LCIA results based on these models is particularly not possible
since LCA is only concerned with the total emission of a substance associated with the
functional unit of a product over its life cycle, which is regarded as a pulse (in kg) and lacks a
time dimension (Guinée and Heijungs 1993). The model can be partially evaluated if applied
in a site-specific risk assessment rather than LCIA since the goal of the LCIA exercise is to
estimate the marginal change in the adverse effect as a function of the marginal change in
emission to an evaluative system. However, the sub-models used in USEtox(new) to
calculate metal speciation/bioavailability (WHAM 6.0) and toxicity (BLMs) have been
evaluated in literature as far as practically possible.
Although CF is a linear function of BF, FF and EF, it is challenging to assess the overall
performance of USEtox(new) model in the context of varying water chemistry since both BF
and EF vary non-linearly as a function of the chemistry of receiving environment. For an
evaluative environment, metal CFs are mainly controlled by BF and then EF (Gandhi et al.
2010, 2011). CFs calculated using USEtox(new) adequately addressed these effects by using
WHAM and BLM. For example, according to the chronic toxicity data in the literature that
were assembled to derive EFs for USEtox(new), the ranking of geometric mean values of
chronic EC50’s was Cd > Cu > Pb > Ni > Zn > Co. However, because the range of EC50
values for individual metals such as Zn was up to 4 orders-of-magnitude depending on water
chemistry, this ranking is not absolute; there is considerable overlap among the ranges of
each metal (Table 4.12). Thus, changes in this order of metal toxicity ranking can occur in
the EF of USEtox(new) as a function of pH, DOC and water hardness that control metal
speciation to varying degrees according to the geochemical behaviour of a metal. For
example, Zn toxicity decreased by a factor of 3 to 8 when pH increased from 6.5 to 9
whereas the effect of pH on Ni toxicity becomes significant only at pH > 8.0-8.2 (De
112
Schamphelaere et al. 2006). These observations are captured in the HC50 estimates and thus
CFs of USEtox(new) (Table 4.6).
Table 4.12: Literature derived ranges and geometric averages of measured chronic toxicity
test data, expressed as total dissolved concentration, for metals considered in the case studies.
Metal Range-EC50
(µg/L)
Geomean-EC50
(µg/L)
Data Source
Cd 2.1 - 1900 17 EU Risk Assessment Report
Co 21.7 - 5050 250 Bill Stubblefield & co-workers
Cu 1.1 - 320 58 EU Risk Assessment Report
Pb 7.6 - 1685 106 Martin Grosell & co-workers
Ni 3.3 - 4138 187 EU Risk Assessment Report
Zn 0.1 - 2050 216 EU Risk Assessment Report
4.5 Conclusions
Apeldoorn, Lausanne and Clearwater meetings recommended that metal-specific speciation
must be considered when evaluating or ranking the ecotoxicity of organic compounds and
metals in the contexts of hazard, risk assessment and LCA. We evaluated the implications of
considering metal speciation, and specifically bioavailability, on estimates of potential
freshwater ecotoxicity, by introducing the method of Gandhi et al. (2010) into USEtoxTM
(new) in comparison to previous methods (USES-LCA 1.0, 2.0, USEtoxTM
(previous)) that
did not account for metal speciation. The comparison was made using the inventories of two
case studies of Cu pipe and Zn gutters (Gloria et al. 2006). By accounting for metal
bioavailability, we estimated 1 to 4 orders-of-magnitude lower overall freshwater ecotoxicity
scores (Σ CF x emissions, calculated using USEtox(new)) for both case studies, in
comparison to estimates from the other models, and 1 to 2 orders of magnitude lower
ecotoxicity for the Cu pipe case study with previously published CFs calculated using USES-
LCA 1.0 (Huijbregts et al. 2000). The range in these differences is due to the choice of
113
freshwater chemistry, as illustrated through the use of 7 freshwater archetypes in the
USEtox(new) calculations. Contributions of each metal to the total score also changed due to
the consideration of metal bioavailability and speciation in USEtox(new). The latter could be
summarized by the change in rank order of metal CFs of USES-LCA 1.0 as
Co>Ni>Cd≈Cu>Zn>Pb, versus USEtox(new) as Cd>Co>Ni>Zn>Cu>Pb for the archetype of
hard alkaline water and Cd>Ni>Co>Cu≈Zn>Pb for the archetype of soft, acidic water. For
both case studies, differences in ecotoxicity scores were not significant for two time-scales:
infinite versus 100 years of impacts after emissions.
4.6 Practical Implications
The main implication of this study is that more realistically considering metal bioavailability
and its dependence on freshwater chemistry using the method of Gandhi et al. (2010) can
decrease estimates of overall metal ecotoxicity by up to several orders of magnitude, as
illustrated in the case studies. As argued by Gandhi et al. (2010), this revised assessment of
freshwater ecotoxicity of metals is consistent with our current understanding of metal
chemistry and ecotoxicity. These lower estimates could reduce contributions of metals, in
general, to overall freshwater toxicity estimates evaluated through LCIA, as well as the
ranking of individual metals and metals relative to organic compounds. The magnitude of
the reduction depends on the freshwater archetype chosen since CFs can vary by up to 2
orders of magnitude for one metal amongst archetypes.
4.7 References
Apeldoorn (2004): Declaration of Apeldoorn on LCIA of non-ferrous metals. Results of a
workshop by a group of LCA specialists, held in Apeldoorn, NL, April 15th, 2004.
Int J Life Cycle Ass 9:334.
Brandes LJ, Den Hollander H, Van de Meent D (1996): SimpleBox 2.0: a nested multimedia
fate model for evaluating the environmental fate of chemicals, Report No.
114
719101029, National Institute of Public Health and the Environment (RIVM),
Bilthoven, The Netherlands.
Campbell PGC (1995). Interactions between Trace Metals and Aquatic Organisms: A
Critique of the Free-Ion Activity Model. In Metal Speciation and Bioavailability in
Aquatic Systems; Tessier, A., Turner, D. R., Eds.; John Wiley: New York, 1995; Vol.
1, pp 45-102.
De Schamphelaere K, Van Laer L, Deleebeeck N, Muyssen B, Degryse F, Smolders E,
Janssen C (2006): Nickel speciation and ecotoxicity in European natural surface
waters: development, refinement and validation of bioavailability models. Report by
Laboratory for Environmental Toxicology and Aquatic Ecology, Ghent University,
Gent, Belgium.
De Schamphelaere K, Heijerick DG, Janssen CR (2006): Development and validation of
Biotic Ligand Model for predicting chronic zinc toxicity to fish, daphnids and algae.
Report by Laboratory for Environmental Toxicology and Aquatic Ecology, Ghent
University, Gent, Belgium.
Den Hollander HA, Van Eijkeren JCH, Van de Meent D (2004): SimpleBox 3.0: multimedia
mass balance model for evaluating the fate of chemicals in the environment. Report
No. 601200003, National Institute for Public Health and the Environment (RIVM),
Bilthoven, The Netherlands.
Diamond M, Gandhi N, Adams W, Atherton J, Bhavsar S, Bullé C, Campbell P, Dubreuil A,
Fairbrother A, Farley K, Green A, Guinée J, Hauschild M, Huijbregts M, Humbert S,
Jensen K, Jolliet O, Margni M, McGeer J, Peijnenburg W, Rosenbaum R, van de
Meent D, Vijver M (2010): The Clearwater consensus: the estimation of metal
hazard in fresh water. Int J Life Cycle Assess 15: 143-147.
Di Toro D, Allen H, Bergman H, Meyer J, Paquin P (2001): Biotic Ligand Model of the
acute toxicity of metals: 1. Technical basis. Environ Toxicol Chem 20 (10): 2383–
2396.
115
Ecobalance (2000a): Life Cycle Assessment of copper tube for the supply of water in
residential structures: Final report – Peer review. Ecobalance – Price Waterhouse
Coopers, Bethesda, MD
Ecobalance (2000b): Life Cycle Assessment of copper tube for the supply of water in
residential structures: Peer review methodology report. Ecobalance – Price
Waterhouse Coopers, Bethesda, MD
Finnveden G (1999): Methodological aspects of life cycle assessment of integrated solid
waste management systems. Resour Conserv Recycl 26(3-4): 173–187.
Forbes VE, Calow P (2002): Species sensitivity distributions revisited: A critical appraisal.
Hum Ecol Risk Assess 8(3): 473-492.
Gandhi N, Diamond ML, van de Meent D, Huijbregts MAJ, Peijnenburg WJGM, Guinée J
(2010): New method for calculating comparative toxicity potential of cationic metals
in freshwater: Application to copper, nickel, and zinc. Environ Sci Technol 44 (13):
5195-5201.
Gandhi N, Huijbregts MAJ, van de Meent D, Peijnenburg WJGM, Guinée J, Diamond ML
(2011): Implications of geographic variability on ecotoxicity potentials of Cu, Ni and
Zn in freshwaters of Canadian ecoregions. Chemosphere 82 (2): 268-277.
Gloria TP, Russell AJ, Atherton J, Baker SR, Cook M (2006): Ecological toxicity methods
and metals: An examination of two case studies. Int J Life Cycle Assess 11 (1), 26-33.
Guinée JB, Heijungs R (1993): A proposal for the classification of toxic substances within
the framework of Life Cycle Assessment of Products. Chemosphere 26 (10), 1925-
1944.
Hauschild MZ, Huijbregts M, Jolliet O, Macleod M, Margni M, van de Meent D, Rosenbaum
RK, McKone TE (2008): Building a model based on scientific consensus for Life
Cycle Impact Assessment of chemicals: the search for harmony and parsimony.
Environ Sci Technol 42:7032–7037.
116
Huijbregts MAJ, Thissen U, Guinée J, Jager T, Kalf D, van de Meent D, Ragas AMJ,
Sleeswijk AW, Reijnders L (2000): Priority assessment of toxic substances in life
cycle assessment. Part I: Calculation of toxicity potentials for 181 substances with the
nested multi-media fate, exposure and effects model USES-LCA. Chemosphere 41:
541-573.
Huijbregts MAJ, Guinée JB, Reijnders L (2001): Priority assessment of toxic substances in
LCA. Part III: Export of potential impact over time and space. Chemosphere 44: 59-
65.
Jolliet O, Rosenbaum R, Chapmann PM, McKone T, Margni M, Scheringer M, van Straalen
N, Wania F (2006): Establishing a framework for life cycle toxicity assessment:
Findings of the Lausanne review workshop. Int J Life Cycle Assess 11: 209-212.
Oreskes N, Shraderfrechette K, Belitz K (1994). Verification, validation, and confirmation of
numerical-models in the earth sciences. Science 263: 641-646.
Payet J, Jolliet O (2002). Comparative assessment of the toxic impact of metals on aquatic
ecosystems: The AMI method. International Workshop on Life-Cycle Assessment
and Metals, 188-191.
Pettersen J, Hertwich EG (2008): Critical review: life-cycle inventory procedures for long-
term release of metals. Environ Sci Technol 42: 4639-4647.
Rosenbaum RK, Bachmann TM, Swirsky Gold L, Huijbregts MAJ, Jolliet O, Juraske R,
Koehler A, Larsen HF, MacLeod M, Margni M, McKone TE, Payet J, Schuhmacher
M, van de Meent D, Hauschild MZ (2008): USEtox—the UNEP/SETAC toxicity
model: recommended characterisation factors for human toxicity and freshwater
ecotoxicity in life cycle impact assessment. Int J Life Cycle Assess 13:532–546.
Skeaff J, Delbeke K, Van Assche F, Conard BA (2000): Critical surface area concept for
acute hazard classification of relatively insoluble metal-containing powders in aquatic
environments. Environ Toxicol Chem 19: 1681-1691.
117
Tipping, E. (1998): Humic ion-binding model VI: An improved description of the
interactions of protons and metal ions with humic substances. Aquat Geochem 4: 3-
48.
Udo de Haes HA, Jolliet O, Finnveden G, Hauschild M, Krewitt W, Mu¨ller-Wenk R (1999):
Best available practice regarding impact categories and category indicators for life
cycle impact assessment, Part 1. Int J Life Cycle Assess 4: 66–74.
van de Meent, D, Huijbregts MAJ (2005): Calculating life-cycle assessment effect factors
from potentially affected fraction-based ecotoxicological response functions. Environ
Toxicol Chem 24: 1573-1578.
van Zelm R, Huijbregts MAJ, van de Meent D (2009): USES-LCA 2.0-a global nested multi-
media fate, exposure, and effects model. Int J Life Cycle Assess 14: 282-284.
118
5. Critical Load Analysis in Hazard Assessment of Metals Using a Unit
World Model©
5.1 Abstract
A Unit World approach has been used extensively to rank chemicals for their hazard and to
understand differences in chemical behaviour. Whereas the fate and effects of an organic
chemical in a Unit World analysis vary systematically according to the fraction of organic
carbon chosen to characterize the system, metals can change in their hazard ranking
according to aqueous chemistry, notably pH and DOC. We present a Unit World Model
(UWM) for metals that loosely couples the Biotic Ligand Model (BLM) to a geochemical
speciation model and then to a TRANsport-SPECiation (TRANSPEC) model. The UWM is
applied to estimate critical load (CL) of cationic metals Cd, Cu, Ni, Pb, Zn using three lake
chemistries that vary in trophic status, pH and other parameters. The model results indicated
a four orders-of-magnitude difference in particle-to-total dissolved partitioning (Kd) that, in
turn, is translated to minimal differences in fate due to the short water residence time used.
However, a 300-fold difference was calculated in Cu toxicity amongst the three chemistries
and three aquatic organisms, which was greatest amongst the metals. Although the highest
fraction of free metal ion in relation to total metal (i.e., bioavailability) was calculated for the
mesotrophic system, the CL was greater for oligotrophic due to the amelioration of toxicity
by competing cations. Thus, the water chemistry has a major impact on CL through effects
on aquatic toxicity. Hazard ranking was in the order of Cd, Cu and Zn based on toxicity to
Fathead minnow and did not change with the chemistries.
© Contents of this chapter have been adopted from the manuscript that is in press in Environmental Toxicology
& Chemistry:
Gandhi, N., Bhavsar, S.P., and Diamond, M.L. Critical load analysis in hazard assessment of metals using a
Unit World Model. Environmental Toxicology & Chemistry In press.
I was primarily responsible for the data analysis, model applications, analyis of model results, and writing of
this manuscript.
119
5.2 Introduction
Finding a scientifically sound and tractable approach to evaluate ecotoxicological impacts of
metals released to the environment requires a modification in use of the criteria of persistence
(P), bioaccumulation (B) and toxicity (T) as applied to assess hazard of organic chemicals in
several jurisdictions (e.g., Adams and Chapman 2005). Harvey et al. (2007) recommended
calculating a Critical Load (CL) for metals that relates a chemical input to an effects
concentration is a more informative criterion than persistence which is infinite for metals. A
CL is estimated by starting from a sensitive toxicity effect endpoint to calculate the
corresponding emission rate of a chemical to a defined system. This estimate of critical
emission rate can then be compared with an estimate of the actual emission rate for potential
hazard or risk analysis. We refer to this as a “reverse” use of the model rather than the usual
“forward” mode in which an emission is specified in order to calculate a resultant
concentration.
The “defined system” or “Unit System Model” (UWM) is employed to estimate average
environmental fate using a uniform, evaluative environmental construct for the purpose of
chemical evaluation and/or hazard ranking, in contrast to models that address site-specific
fate and potential toxicity. The UWM was first proposed by Neely and Mackay (1982),
based on the idea of a “slice of the earth” that represents all major environmental components
relevant to chemical fate. Their Unit World included air, water, sediment and soil in volumes
representative of these media in the environment. The “average” behaviour estimated from a
UWM may never be measured day-to-day, but would be apparent if many measurements
were averaged over many years. UWMs have had a long and successful history of use for
non-polar semi-volatile organic compounds (SVOCs), the behaviour of which can be
reasonably described by standard physical-chemical properties (e.g., vapour pressure,
solubility, KOW, KOA) and environmental characteristics that are specified in the UWM.
Using a UWM for ionizing organics and metals is more complicated than for non-polar
SVOCs because the former exist as multiple interconverting species in an aqueous solution,
e.g., neutral and ionized forms of ionizing organics and chemistry-specific metal species.
However, because of the enormous value of the UWM, the Apeldoorn (Aboussouan et al.
120
2004) and Pellston 2003 (Adams and Chapman 2005) workshops, which were held to discuss
approaches to improve the science behind metal hazard and risk assessment, recommended
adopting the UWM approach and using it to calculate CLs (Aboussouan et al. 2004, Adams
and Chapman 2005).
In keeping with the original intention of a UWM, appropriate parameter values must be
selected that are representative of commonly found environmental conditions. This choice is
informed by the model’s sensitivity to each parameter, such as water residence time and
particle transport rates, and more importantly for metals, ambient chemistry, such as pH,
presence of major ions and composition of suspended sediments or soil solids that can
significantly affect their speciation and partitioning in the environment. For organic
chemicals, this analysis of sensitivity to environmental chemistry is relatively simple since
their relative hazard ranking will not change as a function of the fraction of organic carbon
(foc) which is the key variable affecting the fate and bioavailability of these chemicals. Thus,
choosing a “representative” ambient chemistry is a critical issue when developing a UWM
for metals. The importance of ambient chemistry on estimating the fate and effects of metals
was a focal point of the Clearwater Consensus (Diamond et al. 2010).
Harvey et al. (2007) first developed the UWM that is applicable to both metals and organic
substances and would allow for comparison of the hazards posed by both classes of
substances. However, their model used a fixed value of metal particle-to-total dissolved
partition coefficients (Kd) from the literature which eliminates the transparency and
potentially introduces biases when using a UWM to compare metal CLs and CLs of metals
with organics. Bhavsar et al. (2004a, 2008a) presented a loosely coupled metal TRANsport
and SPECiation model TRANSPEC that allowed the calculation and exploration of metal
species-specific complexation and speciation depending on the choice of ambient aqueous
chemistry values. Gandhi et al. (2010) used this model formulation as the basis of a method
to calculate metal hazard or risk. They added the parallel consideration of water chemistry
on metal complexation and speciation for fate calculations as for toxicity assessment, where
the later was based on the Biotic Ligand Model (BLM). The main point here was to consider
the free metal ion for the toxicity assessment using similar speciation calculations as for the
fate assessment. Gandhi et al. (2011) applied this model formulation to the 24 ecoregions of
121
Canada and found that water residence time, hardness, dissolved organic carbon (DOC) and
pH could change the absolute values of metal ecotoxicity (expressed through the method of
Comparative Toxicity Potential in Life Cycle Impact Assessment), as well as the relative
ranking of metals.
Farley et al. (2011) developed TICKET-UWM (the Tableau Input Coupled Kinetic
Equilibrium Transport-Unit World Model) for the similar purpose of screening the potential
environmental risks associated with the release of metals to lakes. Whereas the model of
Gandhi et al. (2010, 2011) used the geochemical model WHAM to estimate water column
complexation and speciation using literature-derived values of carbon and sulphur and their
model neglected speciation and complexation in sediment, TICKET-UWM included a fully
implicit, one-step solution that explicitly considered the water column and underlying
sediment in addition to organic carbon and sulphur cycling. They also considered the
dissolution kinetics for metal powders, massives and other solid forms. Similarly to Gandhi
et al. (2010, 2011), they found that CLs varied significantly as a function of water residence
time and water hardness, and also metal dissolution kinetic rates.
The goal of this paper was three fold. First, we aimed to develop and present a modelling
structure to analyze freshwater fate and effects of metals within the UWM framework that
can be used for organics or metals. The method calculates CL (or Critical Concentrations,
CC) for a “Unit Lake” that builds on the TRANSPEC model. Our second goal was to
examine the effect of freshwater lake chemistry on the linkage between loadings and
potential ecotoxicity and thus CL of the cationic metals Cd, Cu, Ni, Pb and Zn. This analysis
of the sensitivity of the load-effect relationship to lake chemistry is a necessary step towards
developing “standard” conditions in a UWM. We explored the sensitivity of metals to water
chemistries of three Canadian lakes that vary in trophic status, pH, and other chemistry
parameters in both “forward” and “reverse” directions of the model. We used the full suite
of measured chemistry to provide congruence among water and sediment chemistry
parameters and particle transport rates because it is incongruent to uncouple these values, i.e.,
a model parameterization must consistently represent a eutrophic or hard water or acidic
system. This congruence is analogous to the use of consistent physical-chemical properties
for organic compounds (Diamond et al. 2010, Beyer et al. 2000). Finally, we examined our
122
results to determine whether lake chemistry was more decisive for the analysis of fate or
toxicological effects in the systems that we defined. This analysis is intended to shed light
on those components of the analysis requiring greater efforts and/or consensus decisions that
must be made to bring a UWM approach for metals closer to implementation.
5.3 Methods
Defining a Unit World environment is a critical step in developing a unit world model.
Generally, the surface aquatic systems are classified according to limnology which includes
physical, chemical and biological aspects of the systems. In this study, I explored chemical
and to some extent, biological aspects with respect to metal fate.
5.3.1 Modelling Approach
The UWM is constructed such that it retains simplicity in modeling the main processes and
minimizes data requirements, yet includes sufficient complexity regarding lake and
chemistry processes to simulate average conditions in a unit lake. We start by considering a
metal in the bulk aqueous phase to consist of total dissolved and particulate phases where the
total dissolved (also referred to as soluble in the literature) to consist of the truly dissolved
and colloidal phases. For metals, the toxicologically relevant species is usually the free metal
ion which is one species among truly dissolved fraction (Campbell 1995).
The model presented here connects three separate models: (1) a coupled metal fate-speciation
model, TRANSPEC, that uses the output from a geochemical model in a multi-species fate
model, (2) a geochemical speciation model, Windermere Humic Adsorption Model
(WHAM), that estimates metal distribution and speciation at equilibrium, and (3) an
ecotoxicity model, Biotic Ligand Model (BLM), that estimates the metal concentration that
has the potential to cause an adverse effect (e.g., acute LC50 or alternative regulatory effects
level that is used in a ranking analysis) to an aquatic organism in a given environmental
chemistry. Below we describe each model in details.
123
5.3.1.1 Fate Model - TRANSPEC
Bhavsar et al. (2004a, 2004b) developed a coupled metal fate-TRANsport and SPECiation
model (TRANSPEC) that explicitly considers effects of varying chemistry on metal fate.
The fate component is based on the QWASI (Quantitative Water Air Sediment Interaction)
fugacity/aquivalence concept that considers the multi-species formulation for multiple,
interconverting metal species (Diamond et al. 1992, Diamond and Mackay 1989). Bhavsar et
al. (2004a) describes the model in detail.
The “Unit Lake” consists of three well-mixed compartments: a single water column and
surficial (0-5 cm) and deeper (5-20 cm) layers of sediment. The deeper sediment layer is
underlain by buried sediment. The addition of the deeper sediment layer lengthens the
residence time of chemical in the system, which is consistent with descriptions of sediment
mixing processes and the slow response time of sediments to changes in loadings
(Thibodeaux et al. 2001). We assume that the surficial sediment layer is oxic and the deeper
layer is anoxic. The model also considers three phases within each compartment: truly
dissolved, colloidal and particulate. The colloidal phase is operationally defined as particles
ranging from 1 nm to 0.2 µm diameters that are retained by an ultrafilter membrane and is
effectively Dissolved Organic Matter (DOM). Colloids were added to the model because of
the high affinity of some metals (e.g., Cu) for DOM (Christensen et al. 1999, Rozan and
Benoit 1999) which can dramatically alter bioavailability (e.g., Winner 1985, Ma et al. 1999,
De Schamphelaere and Janssen 2004) and, to some extent, fate (Bhavsar et al. 2008b). Metal
is assumed to be instantaneously distributed among the phases according to ambient
chemistry. We extended the assumption of instantaneous equilibrium to resuspended bottom
sediment that adopts the chemistry of the surface water and similarly for sediment particles
mixed between the oxic and anoxic sediment layers.
Metal in the three phases can enter the lake through direct discharge or stream inflow,
calculated as the product of measured water flow and total metal concentrations, and by wet
and dry atmospheric deposition of aerosol-bound metal calculated using measured total metal
concentrations in aerosols, rain rate and a dry deposition velocity. Truly dissolved and
colloidal-bound metal exchange between the water column and pore water by bi-directional
124
diffusion which is calculated using an average mass transfer coefficient (MTC) 1.0 x 10-5
and
7.0 x 10-7
m/h for the dissolved and colloidal species, respectively (Valsaraj et al. 1993,
1996). The exchange of particle-bound metal between the water column and surficial
sediment occurs by particle deposition and resuspension, and between sediment layers by
benthic mixing and burial. Export from the lake and burial to deep sediments ultimately
removes metal from the system. The fate model uses species fractions calculated by the
speciation model to estimate the species-specific Kd, Z values, and aquivalence fractions (see
Bhavsar et al. 2004a). The model can be run in steady-state and dynamic modes. The fate
model is written in Visual Basic and run on a WindowsTM
(Microsoft Corporation, Redmond,
WA, USA) based personal computer.
5.3.1.2 Speciation Model – WHAM
The Windermere Humic-Aqueous Model (WHAM 6.0, Tipping 1998) is an equilibrium
based metal speciation/complexation model comprised of the Humic Ion-Binding Model VI
and an inorganic speciation code for aqueous solutions. We selected WHAM over other
geochemical speciation models (e.g., MINEQL+, Visual MINTEQ) in order to calculate
consistent estimates of speciation in fate and toxicity because BLM is also based on WHAM.
The model calculates metal distribution among total and particulate phases, expressed as the
distribution coefficient Kd (L/kg). One advantage of WHAM is its sophisticated treatment of
metal binding to humic and fulvic acids in both particulate and colloidal phases. In addition,
it can also estimate metal adsorption to oxides of Fe and Mn. However, the applicability of
WHAM 6.0 is limited for redox related reactions due to its inability to track changes in the
thermodynamic distribution of redox coupled species that could occur in sediment. In such
cases, semi-empirical relationships based on simultaneously-extracted metals and acid
volatile sulphide (SEM/AVS) for anoxic pore waters in deeper sediments can be used (Di
Toro et al. 2001). However, such calculations demand significant data that are generally not
available. In absence of data, we used sediment Kd values that were measured for the
systems modelled in this application. However, we recognize that an alternative approach to
address variations in sediment chemistry in response to water chemistry and other
limnological conditions is necessary for calculating Kd for anoxic sediments such that the
125
method is consistent with estimating water chemistry-specific Kd within UWM framework.
We neglect metal speciation with reduced sulphur in oxygenated waters, which is consistent
with BLM (Bianchini and Bowles 2002). The assumption of chemical equilibrium between
water and air allows for the use of the partial pressure of CO2 when calculating the
dissolution of CO2 into carbonate species. WHAM and other geochemical models assume
that metal species are at chemical equilibrium, i.e., the model does not consider kinetically
controlled reactions such as microbial methylation as done by Gandhi et al. (2007). We
believe that this is a reasonable approach since few kinetic rates are known and that including
such complexity is not in line with other simplifications made in the UWM.
5.3.1.3 Ecotoxicity Model – BLM
The Biotic Ligand Model (BLM) calculates the concentration of bioavailable concentration
of metal that binds to the biotic ligand (BL) in relation to competitive binding with other
ligands (Di Toro et al. 2001, De Schamphelaere and Janssen 2004). The calculations of free
metal ion activity and its equilibrium speciation with other competing inorganic ligands are
performed within WHAM 6.0. The gill surface interaction model (GSIM) in BLM calculates
accumulation of metal at BL by considering the competition among free metal ions and other
cationic ligands (e.g., H+, Na
+, Ca
+2, Mg
+2) to bind at BL sites. The toxicity data used in the
BLM to calculate values of LC50 are specific for each metal and biotic species, and pertain to
acute responses only. Currently, BLMs are publicly available to estimate toxicity of Ag, Cd,
Cu, and Zn for five aquatic organisms (Fathead minnow, Rainbow trout, and the zooplankton
Daphnia magna, Daphnia pulex and Ceriodaphnia dubia). BLMs are being developed to
estimate toxicities of other cationic metals (e.g., Co, Pb), for chronic exposure to aquatic
organisms and to terrestrial biota. Here, we used Hydroqual BLM (version 2.1.2) that was
freely available at the time of running our model calculations
(http://www.hydroqual.com/wr_blm.html).
126
5.3.2 Model Parameterization
We examined the effect of lake chemistry by using a full suite of the water and sediment
chemistry parameters from three aquatic systems, Ross, Kelly and Tantaré Lakes, that differ
in pH, trophic status, and metal loading history. We used the full suite of parameters (e.g.,
concentrations of major cations, anions) because it is unrealistic to vary, for example, pH or
nutrient levels independently of other parameters since they are interrelated. These lakes are
described in detail by Bhavsar et al. (2008b). Below we briefly describe the history and
characteristics of each lake.
5.3.2.1 Ross Lake
Ross lake is a shallow lake located in Flin Flon, Manitoba, Canada (54º46’N, 101º52’W).
Ross Lake has high concentrations of Cu and Zn in water and sediments due to discharges of
treated effluents from Hudson Bay Mining and Smelting Inc., a base metal mine and copper
smelter/zinc refinery, in operation since 1930. In addition, the lake also received raw and
processed sewage input from the 1930s to early 1950s. As a result, the sediments have an
organic carbon content of 10-12% (Evans 2000). The lake consists of two basins, north and
south. We focussed on the north basin due to the availability of data (Bhavsar et al.
2004a,b). .
5.3.2.2 Kelley Lake
Kelley Lake (Sudbury, Ontario, Canada; 46º27’N, 81º04’W) lies immediately downstream of
Sudbury and the Vale INCO (formerly International Nickel Company or CVRD INCO)
mining, milling and smelting complex. The lake has been receiving treated effluents for over
100 years from several mines and smelters, a municipal sewage treatment plant as well as
atmospheric deposition from stack emissions. The lake is divided into east and west basins
of which we focussed on the east basin.
127
5.3.2.3 Lake Tantaré
Lake Tantaré (Quebec, Canada; 47º04’N, 71º32’W) is located in an ecological reserve on the
Precambrian Shield situated about 40 km north of Quebec City. Due to its remote location
and the absence of habitants within its watershed, metal inputs are from natural rock
weathering and long-range atmospheric transport. The lake has four basins: we considered
the most downstream basin which is the deepest and most well studied (Alfaro-De La Torre
and Tessier 2002).
Table 5.1: Parameters values selected for the Unit Lake in the metal fate calculations using
TRANSPEC, in comparison to the values measured for the Ross Lake (MB, Canada), Kelly
Lake (ON, Canada), and Lake Tantaré (QC, Canada). Data for Ross Lake were obtained
from HBMS (unpublished data); for Kelly Lake from field study and Lock (unpublished
data); and for Lake Tantaré from Alfaro-De la Torre and Tessier (2002) and Alfaro-De la
Torre (unpublished data).
Unit Lake Ross Lake Kelly Lake Lake Tantaré
(north basin) (east basin) (west basin)
Water flow m3/day 100000 53000 210000 13900
Water surface area m2 500000 575000 2000000 146000
Mean water depth m 8 2.2 9 7.4
Active sediment area m2 400000 460000 1200000 73000
Surficial sediments
depth
m 0.05 0.05 0.05 0.04
Lower sediments depth m 0.2 0.2 0.2 0.08
Porosity of S-Sed % 95 95 95 98
Porosity of L-Sed % 85 85 80 94
128
Table 5.2: System-specific chemistry parameters for Lakes Ross, Kelley and Tantaré used in
WHAM for speciation calculations to assess the effects of chemistry on fate and toxicity
using the Unit World Model. Data for Ross Lake (MB, Canada) were obtained from HBMS
(unpublished data); for Kelly Lake (ON, Canada) from field study and Lock (unpublished
data); and for Lake Tantaré (QC, Canada) from Alfaro-De la Torre and Tessier (2002) and
Alfaro-De la Torre (unpublished data). Background metal concentrations of 0.1 µg/L for Cd,
1 µg/L for Cu, Ni and Pb, and 10 µg/L for Zn were used in WHAM calculations.
Ross Kelley Tantaré
Trophic Status Eutrophic Mesotrophic Oligotrophic
pH 7.8 7.3 5.3
TSS mg/L 11 7.5 4
DOC mg/L 8.5 5 3
Humic-Fulvic % 20-80 15-85 10-90
Fe M 5.77E-06 1.35E-05 5.72E-07
Mg M 5.21E-04 1.85E-03 8.67E-06
Na M 3.90E-03 5.00E-03 8.00E-04
K M 3.74E-04 6.39E-04 1.00E-04
Ca M 1.11E-02 6.36E-03 2.09E-05
Cl M 9.56E-03 3.37E-04 2.61E-05
NO3 M 1.00E-03 1.53E-04 5.52E-06
SO4 M 7.75E-03 7.98E-04 3.61E-05
PO4 M 1.00E-04 3.34E-06 3.19E-05
Transport rates
Net Deposition g/m2day 2.5 1.5 0.2
Bioturbation-Mixing g/m2day 0.1 0.1 0.05
Burial g/m2day 2 1 0.1
TSS = total suspended solid, DOC = dissolved organic carbon
129
Ross, Kelley and Tantaré water chemistries have trophic status indices (TSI, Carlson 1977)
of 53-55, 42-45 and 26 which classify them as eutrophic, mesotrophic and oligotrophic
systems, respectively (Kratzer and Brezonik 1981). Lake Tantaré has also been classified as
a soft water lake due to its low Ca+2
water concentration (Hare and Tessier 1996).
We assumed one set of typical limnological characteristics of a hypothetical lake (Table 5.1)
located on the Canadian Shield (Moser et al. 1998). In addition to water chemistry
parameters, particle transport rates were taken from each of the three study lakes since these
rates are a function of lake trophic status and as such differ among lakes (Table 5.2). We set
the sediment burial at 0.1 to 2 g/m2/day of sediment solids.
To compare fate results for metals, the model was run with a constant water inflow rate and
therefore a constant water residence time. We assumed the water residence time of the lake
at 40 days, which is relatively short but not uncommon. Since background concentrations of
metals vary by orders of magnitude and metal speciation and distribution vary non-linearly
with concentration, we assumed 0.1 µg/L for Cd, 1 µg/L for Cu, Ni and Pb, and 10 µg/L for
Zn as the background concentrations of these metals. For the analysis of metal behaviour in
the Unit Lake setting, these concentrations result in metal loadings of 10, 100, 100, 100, and
1000 g/d for Cd, Cu, Ni, Pb and Zn, respectively. The atmospheric concentration was set at
0.1 ng/m3 for all metals, which contributes negligibly to total metal loading.
The CC calculations (where CC, in this analysis, is linearly related to CL) are done in three
stages that start from the “critical water concentration” (W-LC50) that would have potential to
cause adverse effects to an aquatic organism and ends with the concentration entering the
system (I-LC50) that will result in the “toxic” concentration. Here, the “W” refers to the
water column and “I” refers to the inflowing water to the system. The first stage used BLM
to estimate W-LC50. BLM expresses W-LC50 in terms of total dissolved metal. Second, we
used WHAM to calculate total metal concentration comprised of particulate and total
dissolved phases where the latter is connected to BLM. Finally, we used TRANSPEC to
back-calculate the corresponding total metal concentration entering the lake, or a “critical
inflow concentration” (I-LC50) that would result in the total dissolved concentration in the
water column equivalent to W-LC50. Thus, the final I-LC50 quantifies a corresponding
130
concentration that could be added to this system that will result in a metal concentration
related to a toxicological effect or no effect, depending on the endpoint selected in BLM
assessment, on aquatic species. Note that this I-LC50 was equivalent to a load for comparison
and ranking among metals since the water inflow rate is constant for the unit lake.
5.4 Results and Discussion
5.4.1 Model Evaluation
With a UWM, one must be confident that the underlying model faithfully describes chemical
fate. Bhavsar et al. (2004a, 2004b, 2008b) evaluated TRANSPEC for Cu and Zn in Ross
Lake, Cu, Ni, Pb and Zn in Kelley Lake, and Cd, Pb and Zn in Lake Tantaré. All modeled
water concentrations were within the range of measured concentrations or within two-fold of
the measured values. Since measured species-specific concentrations were not available for
the lakes, the evaluation was limited to comparing measured and modeled total (total
dissolved plus particulate phases) water concentrations that are product of speciation and fate
estimates. The results of WHAM and BLM were not explicitly evaluated due to the lack of
data.
For the CL analysis, we first analyzed the ‘average’ behaviour of Cd, Cu, Ni, Pb and Zn
within the UWM framework. For this exercise, we ran TRANSPEC in forward mode by
introducing a unit load of metal that would be representative of the background
concentration. Below we evaluate speciation/complexation and fate/transport estimates from
this exercise. Next, we calculate the toxicity and CLs for Cd, Cu and Zn. Our analysis of
CLs excludes Ni and Pb due to inability of the version of BLM we used.
5.4.2 Speciation/Complexation Results
In terms of fate, we are most concerned with metal distribution amongst total dissolved and
particulate phases since this distribution distinguishes metal subject to advective outflow
versus sedimentation. Model estimated that >90% of the metals were in the total dissolved
phase, except for Cu in eutrophic (~85%) and Pb in all systems (Figure 5.1a). Within the
131
total dissolved phase, over 75% of Cd and Ni were in the truly dissolved phase in the
eutrophic and mesotrophic systems, but only 60-75% in the oligotrophic system as the
remaining of the metal was bound to colloids. About 85-90% of the total Cu was in the
colloidal phase in all three systems. This is consistent with Cu’s high affinity for DOM
(Rozan and Benoit 1999, Borg et al. 1989). It is puzzling to us that WHAM estimated
relatively higher fractions of Cd, Ni and Cu in colloidal phase for the oligotrophic system
despite the lowest concentrations of DOC among three systems. A possible explanation of
this higher metal binding could be the presence of very low pH (of 5.3) and higher fraction of
fulvic acid which provides stronger and more binding sites than humic acids.
Pb and particularly Zn were more sensitive to lake chemistry than Cd, Ni and Cu.
Approximately 75-80% of Pb was colloidal and only 20-25% was in the particulate phase in
the eutrophic and mesotrophic lakes, but 98% of Pb was in the particulate phase in the
oligotrophic lake. The high percentage of colloidal-bound Pb in the eutrophic and
mesotrophic lakes is not surprising as Pb also has high affinity to DOM (e.g., Lamelas and
Slaveykova 2008, Weng et al. 2002). However, it should be noted that WHAM, with its
default database, typically overestimates complexation of Cu and Pb to DOM (e.g.,
Christensen et al. 1999). For Zn, phase distribution among the truly dissolved, colloidal and
particulate phases was 40-75% (highest in mesotrophic), 20-50% (highest in eutrophic) and
5-10%, respectively. Considering that only 5-15% of all metals was in the particulate phase
with the exception of Pb in the oligotrophic system, metal fate was driven by water flow
(which is constant among lakes) rather than particle movement (which varies among lakes).
Estimated values of Kd for the five metals ranged over four orders of magnitude for the three
systems and up to three orders for metals within a system (Figure 5.1b). Values of Kd were
generally in the order of oligotrophic > mesotrophic > eutrophic. Within a single lake
chemistry, values of Kd decrease from Pb > Cu > Zn > Cd > Ni. This order of metals for
LogKd values differs somewhat from those for suspended matter in the U.S.EPA database
(LogKd median; range; n): Pb (5.6; 3.4-6.5; 48) > Zn (5.1; 3.5-6.9; 75) > Cd (4.7; 2.8-6.3; 67)
= Cu (4.7; 3.1-6.1; 70) > Ni (4.6; 3.5-5.7; 30) (http://www.epa.gov/athens/publications/
reports/Ambrose 600 R 05 074 Partition Coefficients.pdf). Estimated values of Kd in the
132
water column for more soluble metals, such as Cd, Ni and Zn, were similar for all three
trophic states.
Next, we discuss the results of the aqueous phase speciation since this affects BLM results.
Within the truly dissolved phase, >95% of all metals were predicted to occur as the free
metal ion in the oligotrophic system (Figure 5.1c). However, in the eutrophic and
mesotrophic systems the percentage of free metal ion differed by up to a factor of three
among metals. An example of this three-fold difference was Pb for which the pH values of
7.3 and 7.8 of the mesotrophic and eutrophic systems yielded 50% and 15% of the truly
dissolved phase as free metal ion and 40% and 70% as carbonate, respectively. Similarly to
Pb, 40% of Cu as free metal ion in the mesotrophic system was more than double that of the
15% in eutrophic system. Again, the difference was made up by the higher fraction of Cu
carbonate and bicarbonate in the higher pH system. Overall, the greatest difference due to
water chemistry among the percentages of free metal ion with respect to total metal was a
factor of two.
The percentage of free metal ion in relation to other truly dissolved species is a key factor in
the magnitude of the CL and CC. This percentage was extremely low for Cu and Pb in all
three systems for which the metals were either predominantly in colloidal or particulate
phase (Figure 5.1d). In comparison, up to 70-85% of total Cd, Ni and Zn were estimated to
be the free metal ion in the mesotrophic system. Interestingly, the mesotrophic system was
estimated to have the highest free ion concentrations of Cd, Ni and Zn, rather than the
oligotrophic system which had the lowest pH, DOC and total suspended solids (TSS). This
is because although in the oligotrophic system the free metal ion fraction was the highest
within the truly dissolved phase, the fraction of truly dissolved form in the total metal was
lower compared to that in the mesotrophic system. Again we believe this could be the result
of the highest metal binding to DOC in oligotrophic system for reasons discussed earlier.
133
1E+0
1E+2
1E+4
1E+6
1E+8
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Kd
ps (
L/k
g)
a
b
0%
20%
40%
60%
80%
100%
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Dissolved Colloidal Particulate
1E+0
1E+2
1E+4
1E+6
1E+8
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Kd
ps (
L/k
g)
a
b
0%
20%
40%
60%
80%
100%
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Dissolved Colloidal Particulate
Free ion %
in Total Metal
0
20
40
60
80
100
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
a
b
0%
20%
40%
60%
80%
100%
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Me+2 MeOH+ Me(OH)2 MeHCO3+ MeCO3 MeSO4 Other
Free ion %
in Total Metal
0
20
40
60
80
100
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
a
b
0%
20%
40%
60%
80%
100%
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Me+2 MeOH+ Me(OH)2 MeHCO3+ MeCO3 MeSO4 Other
c
d
a
b
Figure 5.1: WHAM estimated (a) phase distribution, (b) partition coefficients, Kd, between
particulate and total dissolved phases, (c) metal speciation in the total dissolved phase, and
(d) percentage of metal in the free ion form relative to the total metal (sum of total dissolved
and particulate phases) for five metals (Cd, Cu, Ni, Pb, and Zn) and for the selected
chemistries of eutrophic, mesotrophic and oligotrophic systems in the UWM analysis.
134
5.4.3 Fate-Transport Results
Normalized for differences in unit emissions, estimated total metal concentrations in the
water column were lowest for mesotrophic and highest in oligotrophic systems for the unit
loading, except for Zn for which the lowest concentration was estimated for the eutrophic
system (Figure 5.2). This is in contrast to findings for organic chemicals, for which the
lowest concentrations are estimated in eutrophic systems where sorption to particles followed
by removal via sedimentation results in greater loss of chemical from the water column
(Larsson et al. 1998). Total concentrations of Cd, Ni and Zn, which were predominantly in
the total dissolved phase, were similar in the eutrophic and oligotrophic systems. However,
for predominantly particle-bound Pb as well as colloidal-bound Cu, the concentrations were
about 10-15% less in the mesotrophic than in the oligotrophic system (Figure 5.2). For Pb
and Cu, differences in concentrations between two trophic systems ranged from about 4 –
15%; the corresponding range for Cd, Ni and Zn was 1 – 7% (Figure 5.2). Overall, water
column concentrations (after normalizing for the magnitude of the unit input) ranked Pb < Cu
< Zn < Ni ≈ Cd in all three systems.
Metal deposition rates, which ranged from <1 to 22% of the loadings, were lowest in the
oligotrophic system due to the low TSS concentration and hence net particle deposition
(Figure 5.2). In addition, the low pH reduced the amount of metal that was in the particulate
phase. About 11% of Cu and 22% of Pb loadings in the mesotrophic were subject to particle
deposition. Metal deposition rates in the mesotrophic system were similar for Cd, Cu and Ni,
~2 times higher for Pb and ~2 times lower for Zn than those estimated for the eutrophic
despite higher net particle deposition in the eutrophic system. These differences in metal
deposition rates were similar to the relative differences in metal partitioning between two
trophic systems but were magnified due to differences in net particle sedimentation rates.
For all three systems, the diffusive flux was insignificant for all metals: it accounted for ~1%
of inputs to the water column for Cd and Pb in the mesotrophic system (Figure 5.2).
135
We next explored the effect of water chemistry and particle movement on metal fate. Recall
that for all simulations we held the water residence time constant at 40 days. We summarize
metal fate in terms of metal residence time in the water column: a more particle-bound metal
in a more eutrophic lake with the higher TSS will have a lower water residence time due to
higher sedimentation and burial rates than vice versa. Conversely, more soluble metals for
which both truly dissolved and colloidal phases are not subject to settling will be subject to
water export with the maximum residence time of 40 days.
The results of this analysis are consistent with those from the analysis of speciation and
phase distribution. The more soluble metals Cd, Ni and Zn had residence times in the water
column of 37-39 days (Figure 5.3). In contrast, the residence times of Cu and Pb ranged
from 30-38 days. The residence time of all metals in the oligotrophic system approached that
of the water because of the low TSS concentration (4 mg/L) which limited the sedimentation
of all metals, including Pb which was predominantly (~98%) present in the particulate phase.
The shortest residence time was for Pb in the mesotrophic system because of the combination
of the highest percentage in the particulate phase and a medium concentration of TSS. Thus,
the greatest effect of water chemistry on fate, via phase distribution and particle movement,
was ~15% in the unit lake system. We should note however, that the short water residence
time of 40 days minimized the effect of water chemistry on fate which would be accentuated
in large lakes with longer water residence times and in the extreme case oceans (e.g., Gandhi
et al. 2011).
136
Bulk Water Conc
0.007
-0.14
0.002
0.3
0.4
0.1
10
10
10
9.7
9.7
9.9
96
96
99
Cd(a)0.037
0.037
0.037
Bulk Water Conc
<0.001
-0.003
-0.004
10.8
11.2
0.9
100
100
100
89.2
88.8
99.1
881
877
979
Cu(b)0.037
0.037
0.037
Bulk Water Conc
0.08
0.06
0.07
1.4
1.7
0.4
100
100
100
98.6
98.3
99.6
974
971
990
Ni(c)0.037
0.037
0.037
Bulk Water Conc
<0.01
-0.9
-0.04
13
22
9
100
100
100
87
78
91
861
772
900
Pb(d)0.037
0.037
0.037
Bulk Water Conc
-0.4
0.6
0.4
72
50
4.6
1000
1000
1000
928
950
995
9170
9380
9870
Zn(e)0.037
0.037
0.037
Wa
ter
Co
lum
nS
urf
icia
l
Se
dim
en
t
Air
Lo
wer
Se
dim
en
t 0.3
0.3
0.1
10..8
11.2
0.9
1.4
1.7
0.4
13
22
9
72
50
5
Bulk Water Conc
0.007
-0.14
0.002
0.3
0.4
0.1
10
10
10
9.7
9.7
9.9
96
96
99
Cd(a)0.037
0.037
0.037
Bulk Water Conc
<0.001
-0.003
-0.004
10.8
11.2
0.9
100
100
100
89.2
88.8
99.1
881
877
979
Cu(b)0.037
0.037
0.037
Bulk Water Conc
0.08
0.06
0.07
1.4
1.7
0.4
100
100
100
98.6
98.3
99.6
974
971
990
Ni(c)0.037
0.037
0.037
Bulk Water Conc
<0.01
-0.9
-0.04
13
22
9
100
100
100
87
78
91
861
772
900
Pb(d)0.037
0.037
0.037
Bulk Water Conc
-0.4
0.6
0.4
72
50
4.6
1000
1000
1000
928
950
995
9170
9380
9870
Zn(e)0.037
0.037
0.037
Wa
ter
Co
lum
nS
urf
icia
l
Se
dim
en
t
Air
Lo
wer
Se
dim
en
t 0.3
0.3
0.1
10..8
11.2
0.9
1.4
1.7
0.4
13
22
9
72
50
5
Figure 5.2: Estimated fate and transport of total metals in shallow unit world lakes with physical properties described in Table 5.1
and chemistries of the eutrophic, mesotrophic and oligotrophic systems (Table 5.2). Transport rates and concentrations are in g/day
and ng/L, respectively. Unit loadings were considered with the concentrations being close to background values for each metal.
137
Figure 5.3: Comparison of metal residence time (days) estimated using the Unit Lake set up
in TRANSPEC model.
5.4.4 Aquatic Ecotoxicity
We estimated values of LC50 for Cd, Cu, and Zn in the water column of each lake (W- LC50)
based on BLM for five aquatic organisms (Fathead minnow, Rainbow trout, and the
zooplankton Daphnia magna, Daphnia pulex and Ceriodaphnia dubia). We could not
calculate W-LC50 for Ni and Pb due to the limited capacity of this version of BLM. The
predicted values of W-LC50 represented total dissolved concentration of metal in the water
column that would result in an adverse effect on exposed organisms.
The values of W-LC50 were lowest (highest relative toxicity) for the oligotrophic and highest
(lowest relative toxicity) for the eutrophic water chemistries (Figure 5.4). This is despite the
0
10
20
30
40
Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn Cd Cu Ni Pb Zn
Eutrophic Mesotrophic Oligotrophic
Me
tal
res
ide
nc
e t
ime
(d
ay
s)
138
highest fraction of free metal ion as a function of total metal was in the mesotrophic system
where toxicity was ameliorated by competing cations. Daphnids were more sensitive to Cu
followed by Cd and Zn, whereas fish were more sensitive to Cd followed by Cu and Zn. The
values of W-LC50 ranged ~25-fold for Cd, ~100-fold for Cu, and ~10-fold for Zn between the
most and least sensitive organisms for a given water chemistry. The values of W-LC50 for
the most sensitive organism varied by ~10-fold for Cd, ~300-fold for Cu and ~3-fold for Zn
among three water chemistries. Zn showed the least variability in W-LC50 values ranging 3-
4 folds between the eutrophic and oligotrophic systems for all organisms. Thus, both the
choice of sensitive organism and water chemistry are important in the Unit World context.
Oligotrophic versus eutrophic system values of W-LC50 for Cu, Cd and Zn for the Fathead
minnow, the only organism for which BLMs were available for the three metals, varied 10-,
15- and 3-fold, respectively (Figure 5.4). The greatest sensitivity amongst water chemistries
and organisms was the 300-fold difference in values of W-LC50 for daphnids versus fish
exposed to Cu.
The order in values of W-LC50 of Zn > Cu > Cd was consistent within each system chemistry
but the relative magnitude between rankings were different between two system chemistries.
For example, W-LC50 of Cd for the eutrophic system was higher (less toxicity) than that for
Cu (greater toxicity) in the eutrophic and mesotrophic systems. In general, model results
supported the well described phenomenon of increased toxicity of metals in low pH,
oligotrophic systems in which a higher percentage of total metal exists as the free ion in
comparison to higher pH, more nutrient-rich systems.
139
Figure 5.4: “Water critical concentration (W-LC50)” and “Inflow critical concentration (I-
LC50)” (mg/L) for (a) Cd, (b) Cu, and (c) Zn estimated for three levels of aquatic organisms
using BLM and the characteristics of eutrophic (Ross Lake), mesotrophic (Kelly Lake) and
oligotrophic (Lake Tantaré) systems.
5.4.5 Critical Load
Values of I-LC50 were higher than the values for W-LC50 for each metal and water chemistry
by the amount of metal lost to burial, and followed the same trend as for the values of W-
0.0
0.5
1.0
1.5
Fathead Minnow Rainbow Trout Daphnia magna Daphnia pulex Ceriodaphnia
0
2
4
6
8
10
12
0.0
1.0
2.0
3.0
Eutrophic
Mesotrophic
Oligotrophic
NA NA
NA
NA NA
Fathead Minnow
Rainbow Trout
Daphnia magna
Daphnia pulex
Ceriodaphniadubia
Cd
Cu
Zn
W-L
C50
(mg
/L)
0.0
0.5
1.0
1.5
Fathead Minnow Rainbow Trout Daphnia magna Daphnia pulex Ceriodaphnia
0
2
4
6
8
10
12
0.0
1.0
2.0
3.0
Eutrophic
Mesotrophic
Oligotrophic
NA NA
NA
NA NA
Fathead Minnow
Rainbow Trout
Daphnia magna
Daphnia pulex
Ceriodaphniadubia
Cd
Cu
Zn
W-L
C50
(mg
/L)
0.001
0.01
0.1
1
10
100
0.001
0.01
0.1
1
10
100
Eutrophic
Mesotrophic
Oligotrophic
0.001
0.01
0.1
1
10
100
Cd
Cu
Zn
Fathead Minnow
Rainbow Trout
Daphnia magna
Daphnia pulex
Ceriodaphniadubia
I-L
C50
(mg
/L)
0.001
0.01
0.1
1
10
100
0.001
0.01
0.1
1
10
100
Eutrophic
Mesotrophic
Oligotrophic
0.001
0.01
0.1
1
10
100
Cd
Cu
Zn
Fathead Minnow
Rainbow Trout
Daphnia magna
Daphnia pulex
Ceriodaphniadubia
0.001
0.01
0.1
1
10
100
0.001
0.01
0.1
1
10
100
Eutrophic
Mesotrophic
Oligotrophic
0.001
0.01
0.1
1
10
100
Cd
Cu
Zn
Fathead Minnow
Rainbow Trout
Daphnia magna
Daphnia pulex
Ceriodaphniadubia
I-L
C50
(mg
/L)
140
LC50 as discussed above (Figure 5.4). We calculated the ratio between I-LC50 and the W-
LC50 (Figure 5.5), which incorporates two phenomena. First, for the more soluble metals, I-
LC50 (which includes total dissolved + particulate phases) should almost equal W-LC50 (total
dissolved phase only) since minimal metal would be in the particulate phase. Thus, we
expected the greatest departure from a ratio of 1 for Cu since it had higher percentages (10-
15%) of total metal in the particulate phase in comparison to Cd and Zn (Figure 5.1).
Second, the ratio incorporates the removal of metal from the water column due to net
deposition, which was greatest in the eutrophic system with the highest sedimentation rate.
Net deposition was greatest for Cu with 1% (oligotrophic) to 11% (eutrophic) of total
loadings lost to the sediment. In contrast, only 1-3% of Cd and 1-6% of Zn were lost to the
sediments of the lakes. Again, we note that these small differences in fate were due to the
short water residence time chosen since long water residence times can produce similar
importance of fate as ecotoxicity in the final outcome of the analysis (Gandhi et al. 2011).
Figure 5.5: Ratios of I-LC50 (mg/L) to W-LC50 (mg/L) for Cd, Cu and Zn that would protect
Fathead minnow in the selected aquatic systems.
0
0.5
1
1.5
2
Cd Cu Zn
Eutrophic Mesotrophic Oligotrophic
141
The ratios of I-LC50 to W-LC50 were ~1.1 for Cd and Zn (Figure 5.5), indicating that for
these systems, phase distributions and fate had minimal influence over the critical load. In
other words, the similarity of I-LC50 to W-LC50 indicated that most metal was in total
dissolved form and minimal metal was lost due to sedimentation. The ratio departed most
from 1 for Cu (1.33 in the eutrophic versus 1.1 in oligotrophic systems, respectively) which
was consistent with the greatest fraction being in the particulate phase and greatest sediment
retention of this metal in the eutrophic system (Figure 5.5). Whereas the ratio of I-LC50 to
W-LC50 differed by only 5-20% suggesting the minimal effect of lake chemistry on fate,
values of I-LC50 between eutrophic and oligotrophic systems varied by up to a factor of 20
for Fathead minnow and 300 for the Daphnid species due to differences in toxicity. In other
words, the greatest effect of lake chemistry, from 3- to 300-fold, was for chemical speciation
in relation to toxicity, not chemical speciation in relation to fate.
pH
I-LC 50 mg/L
In-lake LC50 mg/L Trout
Fathead Minnow
D. magna
Figure 5.6: The effect of pH on values of in-lake and inflow LC50 (W- and I-LC50) for Zn
for the three receptors considered in BLM, Rainbow trout, Fathead minnow and D. magna.
Here water chemistry parameters are representative of eutrophic system, Ross Lake.
142
pH was among the most important factors influencing speciation, fate and toxicity. In Figure
5.6, we summarize the effect of pH on values of I- and W-LC50 for Zn. The most sensitive
receptor was Rainbow trout, followed by the Fathead minnow and the least sensitive D.
magna. For all biotic species, low pH resulted in the greatest toxicity and the similarity of I-
and W-LC50 (most metal loadings remained in the water column). At values of pH greater
than ~8, toxicity diminished and less metal remained in the water column, resulting in the
greatest difference between I- and W-LC50 and higher values for CL.
5.5 Conclusions
Metals, for which persistence and bioaccumulation are not useful criteria for ranking hazard,
are better assessed using Critical Loads (CLs) in which a Unit World Model approach is
imbedded. We present a framework for estimating CLs according to freshwater ecotoxicity
that is consistent for organics compounds and metals. The framework consists of three
loosely coupled models – metal speciation/complexation (WHAM), fate (TRANSPEC) and
ecotoxicity (BLM). Since all these are a function of water chemistry, we explored the
implications of the choice of water chemistry on CL, as illustrated by application of the
method to Cd, Cu, Ni, Pb and Zn in three water chemistries, for up to five aquatic species.
The results indicated that water chemistry, including hardness, pH and DOC, influenced
metal speciation and phase distribution, including the percentage of total metal as free metal
ion, by up to a factor of two. Despite differences of up to four orders of magnitude in Kd, the
influence of water chemistry on fate was only 15% in our defined system with a short water
residence time of 40 days. CLs were most sensitive to differences in acute toxicity that
varied as a function of water chemistry by 3 to 300 times depending on the metal and biotic
species. CLs were lowest (greatest hazard) in the oligotrophic water chemistry and highest
(least hazard) in the eutrophic water chemistry. The variation in CLs on water chemistry was
greatest for eutrophic systems with high pH where a small change in pH could translate into a
large change in metal phase distribution and toxicity. Conversely, variation in CLs was least
in oligotrophic systems at low pH due to the dominance of total dissolved metal species.
143
To estimate the hazard of metals in freshwaters we recommend taking a Critical Load
approach, which includes a Unit World Model for fate and effects calculations. However, a
thorough exploration of water chemistry and a choice of one or more water chemistries, as
well as aquatic species which vary in sensitivities to metal toxicity, are necessary before such
a framework can be implemented.
5.6 References
Apeldoorn. Aboussouan L, Saft R, Schönnenbeck M, Hauschild M, Delbeke K, Struijs J,
Russell A, Udo de Haes H, Atherton J, Van Tilborg W, Karman C, Korenromp R,
Sap G, Baukloh A, Dubreuil A, Adams W, Heijungs R, Jolliet O, De Koning A,
Chapman P, Ligthart T, Van de Meent D, Kuyper J, Van der Loos R, Eikelboom R,
Verdonck F. 2004. Declaration of Apeldoorn on LCIA of non-ferro metals. SETAC
Globe 5:46-47. Available at: http://www.leidenuniv.nl/cml/ssp/projects/declaration of
apeldoorn.pdf.
Adams WJ, Chapman PM. 2005. Assessing the hazard of metals and inorganic metal
substances in aquatic and terrestrial systems: Proceedings of a SETAC Pellston
workshop held in 2003. SETAC, Pensacola, FL.
Alfaro-De La Torre C, Tessier A. 2002. Cadmium deposition and mobility in the sediments
of an acidic oligotrophic lake. Geochimica et Cosmochimica Acta 66:3549-3562.
Beyer A, Mackay D, Matthies M, Wania F, Webster E. 2000. Assessing long-range transport
potential of persistent organic pollutants. Environ Sci Technol 34:699-703.
Bhavsar SP, Diamond ML, Evans LJ, Gandhi N, Nilsen J, Antunes P. 2004a. Development
of a coupled metal speciation-fate model for surface aquatic systems. Environ Toxicol
Chem 23:1376-1385.
Bhavsar SP, Diamond ML, Gandhi N, Nilsen J. 2004b. Dynamic coupled metal transport-
speciation model: Application to assess a zinc-contaminated lake. Environ Toxicol
Chem 23:2410-2420.
144
Bhavsar SP, Gandhi N, Diamond ML. 2008a. Extension of coupled multispecies metal
transport and speciation (TRANSPEC) model to soil. Chemosphere 70:914-924.
Bhavsar SP, Gandhi N, Diamond ML, Lock AS, Spiers G, De la Torre MCA. 2008b. Effects
of estimates from different geochemical models on metal fate predicted by coupled
speciation-fate models. Environ Toxicol Chem 27:1020-1030.
Bianchini A, Bowles KC. 2002. Metal sulfides in oxygenated aquatic systems: implications
for the biotic ligand model. Comparative Biochemistry and Physiology 133:51-
64(14).
Borg H, Andersson P, Johansson K. 1989. Influence of acidification on metal fluxes in
Swedish forest lakes. Science of the Total Environment 87-8:241-253.
Campbell PGC. 1995. Interactions between trace metals and aquatic organisms: A critique of
the free-ion activity model. In Tessier A, Turner DR, eds, Metal Speciation and
Bioavailability in Aquatic Systems. John Wiley, New York, pp 45-102.
Carlson RE. 1977. Trophic state index for lakes. Limnology and Oceanography 22:361-369.
Christensen JB, Botma JJ, Christensen TH. 1999. Complexation of Cu and Pb by DOC in
polluted groundwater: A comparison of experimental data and predictions by
computer speciation models (WHAM and MINTEQA2). Water Research 33:3231-
3238.
De Schamphelaere KAC, Janssen CR. 2004. Development and field validation of a biotic
ligand model predicting chronic copper toxicity to Daphnia magna. Environ Toxicol
Chem 23:1365-1375.
Diamond ML, Mackay D, Welbourn PM. 1992. Models of multi-media partitioning of multi-
species chemicals: the fugacity/aquivalence approach. Chemosphere 25:1907-1921.
Diamond ML, Gandhi N, Adams WJ, Atherton J, Bhavsar SP, Bulle C, Campbell PGC,
Dubreuil A, Fairbrother A, Farley K, Green A, Guinee J, Hauschild MZ, Huijbregts
MAJ, Humbert S, Jensen KS, Jolliet O, Margni M, McGeer JC, Peijnenburg W,
145
Rosenbaum R, van de Meent D, Vijver MG. 2010. The Clearwater consensus: the
estimation of metal hazard in fresh water. International Journal of Life Cycle
Assessment 15:143-147.
Di Toro DM, Allen HE, Bergman HL, Meyer JS, Paquin PR, Santore RC. 2001. Biotic ligand
model of the acute toxicity of metals. 1. Technical basis. Environ Toxicol Chem
20:2383-2396.
Evans LJ. 2000. Fractionation and aqueous speciation of zinc in a lake polluted by mining
activities, Flin Flong, Canada. Water, Air, and Soil Pollution 122:299-316.
Farley KJ, Carbonaro RF, Fanelli CJ, Costanzo R, Rader KJ, Di Toro DM. 2011. TICKET-
UWM: A coupled kinetic, equilibrium and transport screening model for metals in
lakes. Environ Toxicol Chem DOI: 10.1002/etc.1518.
Gandhi N, Bhavsar SP, Diamond ML, Kuwabara JS. 2007. Development of a mercury
speciation, fate, and biotic uptake (BIOTRANSPEC) model: Application to Lahontan
reservoir (Nevada, USA). Environ Toxicol Chem 26:2260-2273.
Gandhi N, Diamond ML, van de Meent D, Huijbregts MAJ, Peijnenburg W, Guinee J. 2010.
New method for calculating comparative toxicity potential of cationic metals in
freshwater: Application to copper, nickel, and zinc. Environ Sci Technol 44:5195-
5201.
Gandhi N, Huijbregts MAJ, van de Meent D, Peijnenburg W, Guinee J, Diamond ML. 2011.
Implications of geographic variability on Comparative Toxicity Potentials of Cu, Ni
and Zn in freshwaters of Canadian ecoregions. Chemosphere 82:268-277.
Hare L, Tessier A. 1996. Predicting animal cadmium concentrations in lakes. Nature
380:430-432.
Harvey C, Mackay D, Webster E. 2007. Can the unit world model concept be applied to
hazard assessment of both organic chemicals and metal ions? Environ Toxicol Chem
26:2129-2142.
146
Kratzer CR, Brezonik PL. 1981. A Carlson-type trophic state index for nitrogen in Florida
lakes. Water Resources Bulletin 17:713-715.
Lamelas C, Slaveykova VI. 2008. Pb uptake by the freshwater alga Chlorella kesslerii in the
presence of dissolved organic matter of variable composition. Environmental
Chemistry 5:366-372.
Larsson P, Okla L, Cronberg G. 1998. Turnover of polychlorinated biphenyls in an
oligotrophic and an eutrophic lake in relation to internal lake processes and
atmospheric fallout. Canadian Journal of Fisheries and Aquatic Sciences 55:1926-
1937.
Ma H, Kim SD, Cha DK, Allen HE. 1999. Effects of kinetics of complexation by humic acid
on toxicity of copper to Ceriodaphnia dubia. Environ Toxicol Chem 18:828–832.
Mackay D, Diamond M. 1989. Application of the QWASI (quantitative water air sediment
interaction) fugacity model to the dynamics of organic and inorganic chemicals in
lakes. Chemosphere 18:1343-1365.
Moser KA, Smol JP, Lean DRS, MacDonald GM. 1998. Physical and chemical limnology of
northern boreal lakes, Wood Buffalo National Park, northern Alberta and the
Northwest Territories, Canada. Hydrobiologia 377:25-43.
Neely WB, Mackay D eds. 1982. Modelling the Fate of Chemicals in the Aquatic
Environment. Science, Ann Arbor.
Rozan TF, Benoit G. 1999. Geochemical factors controlling free Cu ion concentrations in
river water. Geochim Cosmochim Acta 63:3311-3319.
Thibodeaux LJ, Valsaraj KT, Reible DD. 2001. Bioturbation driven transport of hydrophobic
organic contaminants from bed sediment. Environmental Engineering Science
18:215-223.
Tipping E. 1998. Humic ion-binding model VI: An improved description of the interactions
of protons and metal ions with humic substances. Aquat Geochem 4:3-48.
147
Valsaraj KT, Verma S, Sojitra I, Reible DD, Thibodeaux LJ. 1996. Diffusive transport of
organic colloids from sediment beds. Journal of Environmental Engineering 122:722-
729.
Valsaraj KT, Thoma GJ, Porter CL, Reible DD, Thibodeaux LJ. 1993. Transport of dissolved
organic carbon-derived natural colloids from bed sediments to overlying water:
Laboratory simulations. Water Science and Technology 28:139-147.
Weng LP, Temminghoff EJM, Lofts S, Tipping E, Van Riemsdijk WH. 2002. Complexation
with dissolved organic matter and solubility control of heavy metals in a sandy soil.
Environ Sci Technol 36:4804-4810.
Winner RE. 1985. Bioaccumulation and toxicity of copper as affected by interactions
between humic acid and water hardness. Water Research 19:449–455.
148
6. Conclusions and Recommendations
6.1 Scientific Significance of my Research
Numerous organizations, from non-governmental to intergovernmental, as well as industry
and industry consortia, are involved with environmental protection through chemical
adjudication and ranking. In Canada chemical adjudication is the purvey of the Canadian
Environmental Protection Act (CEPA, 1999) and the relatively new Toxics Reduction Act in
Ontario. The basic tenants of first tier screening in the adjudication process are “PBT”–
persistence, bioaccumulation and toxicity. The P & B are principally based on scientific
knowledge and experience with non-polar organic compounds. Persistence is used as a
surrogate for exposure because exposure is most likely to come from long-lived chemicals.
Bioaccumulation is necessary to account for those chemicals that may have low
concentrations in air, water, etc., but that could accumulate in biota. Again, the
bioaccumulation criterion is based on “classic” knowledge of the bioaccumulation of non-
polar chemicals in fish.
Hazard and/or risk assessment allow for second and third tier screening within chemical
adjudication. Life Cycle Assessment (LCA) also includes an examination of chemical
hazard through the toxicity impact category within Life Cycle Impact Assessment (LCIA).
All these tools include assessments of chemical fate and toxicity arising from a release of
chemical into a model environment, based on methods originally developed for non-polar
organic chemicals. However, the goal of all screening methods is their applicability to a
wide range of chemicals, including metals, polar and ionizing organic compounds and
polymers.
Critics have justifiably pointed out that application of the “PBT” criteria and the evaluative
assessment of chemical hazard and/or risk by means of linked fate and toxicity assessments,
result in the biased adjudication of metals (e.g., Apeldoorn 2004, Adams and Chapman 2005,
Diamond et al. 2010). The criticisms pertain to both the fate and toxicity components of the
assessment and were discussed in the Introduction to this thesis and the chapters herein. The
149
goal of this thesis was to address several of the limitations of current approaches to assess the
hazard/risk/impact of metals.
My thesis presents a new generic framework for metal hazard assessment and LCIA that is
scientifically robust and addresses the geochemical attributes and behaviours of metals and
metal compounds relevant to the unbiased assessment of their ecological hazards/risks and
LCIA. Specifically, I presented the new metal hazard assessment framework by incorporated
chemical speciation and bioavailability in the calculations of freshwater toxicity or
comparative toxicity potential (CTP) in the context of LCIA (Chapter 2; Gandhi et al. 2010).
The framework was illustratively applied to the well studied metals Cu, Ni and Zn to
calculate their hazard potentials using the representative environmental chemistries of 12
European freshwater systems. Next, I extended the model utility by applying it to a regional
model of 24 Canadian ecoregions (Chapter 3; Gandhi et al. 2011a). This model incorporated
geographically differentiated systems that vary in ambient chemistry, metal background
concentrations and morphological properties like water residence time of the system in order
to assess the effects of these parameters on freshwater ecotoxicity potentials of metals. The
model’s sensitivity to various chemistry and fate parameters was also explored. A detailed
examination of the effect of using revised CTPs was presented in two metal LCIA case
studies previously studied by Gloria et al. (2006), that of copper water pipe and zinc roof
gutter system (Chapter 4; Gandhi et al. 2011b). The results of these case studies
demonstrated the practical implications of using the revised metal CTPs for assessing
freshwater ecotoxicity impacts. The final major contribution of my thesis was the application
of this new framework to calculate a chemical’s Critical Load, which has been advocated as
an improved, unbiased alternative to the P&B screens for chemical assessment. A Critical
Load approach is the reverse application of the framework presented in Chapter 2. The
critical load approach starts from a concentration in a freshwater environment that would be
toxicologically protective, and works backwards to the related chemical emission rate.
Chemicals could then be ranked according to their Critical Load that would incorporate fate
and toxicity considerations. Within the Critical Load framework, as well as conventional
“forward” hazard, risk and LCIA calculations, the idea of using a “Unit World Model
(UWM)” has been advocated to establish model’s generality and utility in the regulatory
arena (Chapter 5; Gandhi et al. 2011c). As with hazard, risk and LCIA assessments,
150
questions arise about how to parameterize such a model so that metals receive an unbiased
assessment and so that the tool is easily used. “Ease of use” requires the model to have a set
of default parameter values. Chapter 5 explores the question of default parameter values
with respect to the choice of water chemistry with the Critical Load framework.
The research presented here has lead to “next steps” that should be taken to allow the new
framework to be broadly applied to metals. The first point comes from the outcome of
Chapters 3 and 5, that fate and toxicity of metals in a freshwater environment, and hence
hazard (expressed as CTP) depend on the chemistry of that water. It is important to
emphasize that different chemistries can yield not only different estimates of absolute hazard,
but also can change hazard rankings among metals and for metals among organic
compounds. Thus, the obvious question is “what freshwater chemistry should be used for
standard assessments?” If we choose one chemistry, then which one and how will that bias
metal and overall chemical rankings? If we choose several chemistries in a form of
freshwater archetypes, how many and which should be chosen?
In the context of LCIA, I propose three distinct ways of characterizing freshwater systems
based on: (1) geographic distinctions; (2) abundance of system chemistry according to its
contribution towards global freshwater volumes; and (3) generic classes of freshwater
systems regardless of their frequency in nature. Under a geographic classification scheme,
freshwater archetypes can be proposed either based on the political boundaries (e.g., Canada
versus Denmark as in GLOBOX model by Sleeswijk 2006) or global ecoregions based on
common landscape characteristics (e.g., eastern versus prairie ecoregions of Canada as in
ChemCAN model by Webster 2004, Gandhi et al. 2011a). This scheme can work if
sufficient water chemistry data are available for the regions and if there is sufficient
chemistry homogeneity within regions for such a scheme to be sensible. Providing
archetypes based on the abundance of each water chemistry again depends on the availability
of data. Adopting such an approach would require the compilation of a comprehensive set of
data which is not trivial. Undoubtedly gaps would exist for poorly studied regions and this
pertains as well to geographic classification scheme. The third approach is to propose
generic archetypes that are representative of the range of water chemistries found globally.
The archetypes would include circumneutral, mesotrophic waters (medium pH, DOC,
151
harness), acidic (low pH, low alkalinity), alkaline (high pH), eutrophic (high DOM, medium
to high pH), and oligotrophic or pristine systems (low DOC). A complete discussion of this
issue is forthcoming from the analysis I completed for several cationic metals in 100 distinct
freshwater chemistries comprising large rivers and lakes world-wide.
The second major issue arising from this thesis is the extension of the new framework to
metals beyond those for which BLMs have been developed. Chronic BLMs are currently
available for metals like Cu, Ni, and Zn with models nearing completion for Al, Pb and Co.
Is it possible to use the Free Ion activity Model (FIAM; Campbell 1995) to replace BLMs for
other metals? To explore this question, I calculated and compared ecotoxicity effect factors
(EFs) for several distinct freshwater-types distributed globally. The results from the two
models, BLM and FIAM, are within two-fold for Cu, Ni and Zn which is probably
reasonable for the purpose of hazard ranking and LCIA. The analysis also showed greater
differences arising from the choice of freshwater chemistry than the choice of BLM versus
FIAM. Thus, with the use of FIAM, the new framework can be extended to those metals for
which geochemical calculations are available.
In summary, the new framework addresses issues that have been identified to bias the
ranking of metals in chemical hazard and risk assessments. The target audience for this
framework is intended to be regulators, environmental managers and specifically risk
assessors, as well as model developers and stakeholders of the metal industry. The new
framework can be used in conjunction with guidance developed by the various jurisdictions
and regions for use in site-specific risk assessments, criteria derivation, ranking,
categorization and other regulatory activities.
6.2 Major Findings
The following are general conclusions that emerge from the work presented in this thesis:
� A new metal modelling framework has been developed and presented as a general,
loosely coupled speciation-fate-toxicity model for calculating comparative toxicity
potentials (CTPs) or relative hazards of metals.
152
� The new framework explicitly incorporates the concept of bioavailability that can be
estimated based on metal speciation/complexation by assuming that metals’ toxic
effects are related to the truly dissolved free metal ion, not the total dissolved phase.
This is an important modification over the previous approach that assessed chemical
hazard ranking assuming that all dissolved species of a chemical are equally
bioavailable. The new approach takes a step towards distinguishing metal speciation
among three phases: dissolved, colloidal and particulate phases and by considering
metal complexation with a colloidally-bound fraction that is not bioavailable.
� The framework is generic by being consistent with the current hazard assessment
method of assessing organic chemicals, but with attention paid to the geochemical
properties of most cationic metals. The model is an improvement over past efforts by
explicitly addressing the effect of system-specific chemistry on metal speciation-
complexation, and then using this information in bioavailability, fate and ecotoxicity
calculations.
� The setup of new framework has been illustratively demonstrated using (1) USEtoxTM
(Hauschild et al. 2008; Rosenbaum et al. 2008) for environmental fate, (2)
Winderemere Humic Aqueous Model (WHAM 6.0; Tipping 1998) for metal
partitioning and speciation in aquatic systems, and (3) Biotic Ligand Model (BLM;
Di Toro et al. 2001) for the calculation of average toxicity of metals.
� The new method is applicable to a range of aquatic systems and requires systems-
specific data on ambient chemistry and transport parameters. The relative ranking of
hazard (CTPs) can change depending on the choice of water type.
� The sensitivity of metal speciation-complexation calculations and hence estimates of
fate and toxicity to the choice of ambient chemistry is greatest for metals, notably Cu
and Pb, that have a high affinity for organic matter and shift speciation according to
pH and water hardness.
153
� The greatest contributor to variability in metal ecotoxicity (as shown in CTPs) was
metal bioavailability (as represented by Bioavailability Factor; BF), followed by the
toxicity (as indicated by the Effect Factor; EF) and fate (Fate Factor; FF).
� The revised CTPs for freshwater metal ecotoxicity estimated using of new framework
are up to three orders of magnitude lower than the values previously used in metal
LCIA (e.g., Huijbregts et al. 2000) and thus can dramatically decrease the hazard of
metals relative to the ranking of organic compounds.
� Metal fate in Canadian ecoregions results in differences in the absolute values of
metal CTPs within 20 times. The results of this analysis supported the previous
findings that (a) water chemistry can change the absolute hazard for metals; (b) both
fate and ecotoxicity assessments may contribute equally to CTPs; and (c) metal fate is
sensitive to water residence time in the system.
� Application of the new framework to the case studies of copper pipe and zinc roof
gutter systems assessed using published inventory data showed that the choice of
model and most importantly the inclusion of the bioavailability factor significantly
changes the overall freshwater ecotoxicity score (ΣCTP x emissions) and the
contribution of metals to this score.
� The new modelling framework can be used as the basis for calculating Critical Loads
using a “Unit World Model” that is suitable for a wide range of chemicals while
accounting for metal- and system-specific chemistry.
� Unlike organic compounds, the choice of water chemistry is important and hence
decisions must be made on which water chemistry and environmental characteristics
should be used for a screening levels model, such as that used to calculate
“consensus” CTPs for metals in a generic LCIA. Similarly a “consensus” evaluative
environment must be chosen to assess critical load of metals for protection of aquatic
ecotoxicity whereas site-specific risk assessment of metals must incorporate system-
specific differences in morphological and ambient chemistry parameters.
154
� Screening level models need to connect the parameter values used in the fate
calculation (FF) with those used to calculate speciation/complexation calculation (to
calculate metal phase partition coefficient Kd, BF and EF). Both total suspended solid
(TSS) concentration and sedimentation rate in an aquatic system depend on its
watershed characteristics, primary productivity and hence water chemistry. The value
of TSS for the freshwater compartment specified in the fate model must be linked to a
corresponding net sedimentation rate used in the model.
6.3 Lessons Learned
The model framework and its applications presented in my thesis are primarily based on the
following loosely coupled models: (1) a equilibrium based geochemical model, WHAM 6.0
(Tipping 1998) that calculates metal speciation and bioavailability; (2) USEtoxTM
(Hauschild
et al. 1998, Rosenbaum et al. 2008) that predicts multi-media fate; and (3) BLM (Di Toro et
al. 2001) which provides estimation of metal freshwater ecotoxicity. The following is a
critique of these models derived from their extensive use.
Geochemical models are extremely useful for estimating metal speciation and complexation,
particularly since methods available for measuring metal species are limited and few
researchers are able to make reliable measurements. As an advancement of the well used
models such as MINEQL+ (Schecher and McAvoy 1992) and MINTEQA2 (Allison et al.
1990, Allison and Perdue 1994), WHAM was introduced by Tipping (1998) to improve
estimates of metal binding with humic substances. WHAM is particularly suitable for
calculations in those circumstances where the metal speciation is dominated by organic
matter, however WHAM does not include some of the features of the other models notably
incorporation of metal precipitation and redox sensitive reactions. WHAM has been widely
embraced by researchers involved with metal hazard and risk assessment.
WHAM, as with other geochemical models, is based on the fundamental principles of the
simultaneous consideration of all reaction equilibria, quantified with measured metal binding
constants. Whereas the mathematics of geochemical models are well tested, the weakness of
all the models ultimately comes down to the assumptions upon which the models are based
155
and the reliability of the measured binding constants (Bryan et al. 2002, Zhang 2004),
especially those for the metal adsorption to humic and fulvic acids (Bryan et al. 2002,
Unsworth et al. 2005). Some of the key issues that must be considered while using WHAM
for metal hazard framework are the following.
1. WHAM is developed for 14 metals (Al, Ba, Be, Co, Cd, Cu, Cs, Fe, Hg, Mn, Ni, Pb, Sr,
and Zn). MINEQL+ is able to estimate speciation for a more comprehensive set of metals
(Schecher and McAvoy 1992), for example Ag, As, Cr, and Cs to name a few.
2. The conditional binding constants in WHAM are statistically developed based on limited
titrations performed for a specific range of solution chemistries and major assumptions
involved in characterizing humic material (i.e., fraction of humic and fulvic acids; their first
and second order dissociation constants; and their strong and weak proton binding
capacities). Application of this model outside the range of calibration chemistry and
violations of major assumptions may result in erroneous estimates of metal
speciation/bioavailability.
3. The evaluations of WHAM predictions are limited. Studies have shown that WHAM can
overestimate complexation of Cu and Pb with DOM and thus underestimate their
bioavailability (e.g., Christensen et al. 1999), whereas for other soluble metals like Ni and
Zn, the model underestimates metal complexation with DOM and overestimates free metal
ion fractions in natural waters (e.g., Warnken et al. 2009). WHAM predictions are also less
satisfactory at predicting the pH dependence of metal binding with DOM. For example
Unsworth et al. (2005) reported that the reasonable fits to the Cd measurements could only be
obtained from WHAM 6 when the effective binding constant LogK-MA was changed from 1.6
to 1.5, the value of DeltaLK(1) from 2.8 to 1.0 to minimize the dependence on pH, and the
value of DeltaLK(2) from 1.48 to 1.0 to decrease the strength of the strong bidentate and
tridentate binding sites. In general, WHAM predictions are improved for systems with high
ligand concentrations (Tipping 1998, Yapici et al. 2008).
4. Metal precipitation and other redox sensitive processes are neglected in WHAM or
modelled primitively in other geochemical models due to either a lack of saturation indices or
due to computational limitations in tracking the precipitated metals. Mineral adsorption
156
processes in WHAM are poorly characterized and incorporate a limited number of isotherms
to assess metal adsorptions on solid surfaces such as Fe and Mn oxides. This may lead to
underestimation of overall metal partitioning and thus fate in aquatic systems. Precipitation
reactions are of great importance in considering the fate of metals in sediments and soil
because these reactions remove metals from a system. The later is a key point, since in
comparison to organic compounds, metals have infinite physical persistence. Therefore, by
not adequately capturing these processes, modellers err on the side of biasing the longevity
and exposure of metals in a system relative to organic compounds.
5. The main assumption upon which geochemical models are based is that metal speciation
and complexation are at chemical equilibrium. In reality, metal speciation is not always at
equilibrium. The abundance of As(III) and As(V) in surface waters is a case in point.
Another example is the methylation of mercury, which could significantly alter overall metal
fate and toxicity. Thus, a weakness of geochemical models is the lack of incorporating
kinetically controlled metal dissolution and microbially mediated reactions to which metals
are subject to.
6. WHAM and other geochemical models have limited or no capacity to estimate metal
speciation in high ionic strength solutions, e.g., marine water, soil and sediment pore water.
Although WHAM has been applied to soil systems (e.g., Thakali et al. 2006), it provides
satisfactory estimates only for metals and soil types for which soil organic matter is the
principal binding phase.
The differences among equilibrium processes modelled in WHAM and other geochemical
models (e.g., MINEQL+, MINTEQ) result in significant variability in metal speciation and
bioavailability estimates as shown by Bhavsar et al. (2008). These differences could
significantly change fate and toxicity estimates of these metals and hence their hazard
ranking between metals and among other chemicals. Thus, a next step to improve
speciation/complexation estimates would be adding a sub-model of kinetically controlled
reactions such as metal dissolution (e.g., Skeaff et al. 2000) and/or microbial reactions (e.g.,
Gandhi et al. 2007). Recently, Farley et al. (2011) introduced a next generation of metal
speciation, transport and toxicity model (TICKET-UWM) that explicitly incorporates
157
selected metal precipitation reactions with hydroxides, carbonates and sulfides, and
dissolution kinetics for metal powders, massives and other solid forms.
BLM is an outgrowth of geochemical modelling and specifically WHAM. The use of BLM is
justified and supported by its relative simplicity and theoretical basis. However, many of the
limitations discussed above apply to BLM. Recall that BLM links a concentration in a
medium with a concentration on fish gill (or an equivalent biotic ligand site) that would
cause an adverse effect or toxicity. In addition to the criticisms raised for WHAM, the
following limitations must be recognized regarding the use of BLM in a metal hazard
assessment.
1. The model provides a sound geochemical treatment of metals but is based on a semi-
empirical approach that fails to account for physiological processes after metals reach the gill
binding sites, i.e., it does not incorporate mechanisms of metal detoxification and tolerance
for metals in organisms. This introduces a bias in toxicity assessment among metals with
different physiological uptake and toxicological dynamics. This limitation is especially
important when using the model on a site-specific basis. In the context of hazard assessment,
a bias is introduced between the assessment of metals and organics since biota can adapt to
metal but not xenobiotic exposure.
2. A major criticism of the reliance on BLM is the extensive investment of time and cost to
carry out toxicity tests for each metal and organism, necessary to quantify model parameters.
These test parameters are required to distinguish the effects of individual chemistry
parameters, such as pH, DOC, and major cations on the competition of metal binding with
fish gill. A potential solution to this problem could be the use of in-vitro and in-silico testing
methods, but many of these results can be difficult to interpret and extrapolate. Perhaps more
important to note here is that model parameters are mathematically fitted to toxicity
measurements in order to calibrate the model for each metal and organism under
investigation, and therefore lacks rigour in its mechanistic aspect on a physiological side.
This is further evident by the fact that model parameters are calibrated, often without a valid
scientific justification, to fit the experimental results within an undefined reasonable range
(~2 to 10X of an observation). Since the method of quantifying model parameters depends
158
on many critical assumptions (because the model is fitted to experimental data within a range
of pH, DOC and hardness), these conditional binding constants have limitations to their use
which are often overlooked in practice.
3. Further, the toxicity test data (e.g., EC50 or LC50) are extrapolated across species and
organism classes without confirmation of the extrapolations. This problem is common to any
hazard and risk assessment application.
4. Finally, BLM fails to provide estimates of body burden which can then be transferred to
organisms at the next level of the food chain. Therefore, this hazard assessment fails to take
a step further for human exposure assessment or it neglects exposure through the food chain.
This is important in light of several studies that suggest that diet could become a significant
source of exposure to ecological receptors and humans (e.g., Croteau and Luoma 2008,
Szebedinszky et al. 2001).
Another important aspect of toxicity estimation presented in this thesis is the use of
experimental data for generating a representative measure of toxicity at the ecosystem level.
This is generally accomplished via the use of approaches like the Species Sensitivity
Distribution (SSD; Traas et al. 2001). Within this area of research, major lessons learned
include the following:
1. SSD is a strictly statistical approach that may fail to include natural ecosystem interactions
and dependence among species/organisms, e.g., keystone species, predator-prey
relationships.
2. The method uses results from standard test species (e.g., Daphnia magna, rainbow trout)
to generate water chemistry-specific SSDs. Although the framework incorporates varying
ambient chemistry for speciation, fate and toxicity calculations, it uses the same ecosystem
structure for all water-types. Thus, the composition of an ecosystem and presence of various
organism classes in different water chemistries are overlooked. The importance of this issue
vis-a-vis comparative hazard ranking and LCIA purposes has yet to be considered but this
could introduce a significant bias for site-specific risk assessments since many organisms
included in SSD may not be relevant to the system being studied.
159
3. Finally, SSDs do not incorporate variability in responses of organisms due to factors like
adaptation and tolerance, which introduces bias for metals.
Although my thesis focused on aspects of chemistry in this metal hazard assessment, there is
room for improving physical processes modeled in the multi-media fate calculations that
were performed using USEtoxTM
. The following major limitations were identified
specifically for USEtoxTM
but also pertain to other fate models available for chemical hazard
assessment.
1. Sediment and soil fate processes do not incorporate slow, kinetically limited processes like
sediment digenesis, leaching or weathering that impact the overall residence time of
chemicals in a system. Although this may be a “metal specific” issue, neither USEtoxTM
nor
any other multi-media fate model such as ChemCAN (Webster et al. 2004), EUSES-LCA
(Hujbregts et al. 2000), considers the 2-phase sorption and partially irreversible desorption of
organics, which is a similar bias to that of metals. However, these improvements would
require considerable data that are currently not available nor could be easily measured.
Further, adding these details would make fate models more complex for the purpose of
screening and hazard assessment.
2. The sediment compartment is not explicitly modeled in USEtoxTM
and sediment-water
exchange processes are fixed using an empirical approach for the freshwater compartment
that is parameterized as 2.5 metres deep. Although metal diffusion rates are minimal in
comparison to other fate processes (Bhavsar et al. 2004, 2008), sediment mixing,
resuspension and degradation rates could be substantially different between metals and
organic chemicals and can change the overall fate. This limitation leads to two problems.
First, the model fails to explicitly assess chemical contamination for the sediment
compartment and hence exposure and toxicity to benthic organisms. Second, the empirical
approach may inaccurately assess chemical feedback from sediment-to-water since non-polar
organic chemical degrade in sediment whereas metals do not. This may not be a problem for
deeper systems where sediment-to-water exchange has less impact in terms of freshwater
ecotoxicity, but it could become significant for shallow systems in which the water column
160
and sediment are more closely coupled. The issue of when sediment should be explicitly
included merits further discussion in terms of hazard assessment.
3. When implementing the inclusion of a set of default parameters that allows the user to
choose among several default freshwater archetypes, the model architecture needs to be able
to link freshwater chemistry with fate parameters. For example, archetypal differences
between TSS concentrations currently do not translate to differences in net sedimentation rate
whereas in reality these are connected. The use of inconsistent values of DOM and foc also
leads to differences in an evaluation platform on which both organics and metals assessed.
6.4 Recommendations for Future Work
The research described in this thesis has contributed to the development of a generic
framework and a common platform for describing chemical fate and toxicity in screening
level and more detailed chemical hazard and/or risk assessments, and LCIA. Whereas the
thesis has advanced this new framework, many aspects of the method and its application
within a regulatory context require further research.
The following areas require further research for model application specifically to aquatic
systems and in general to other environmental media including air, sediment, soil and human
exposure:
� Form of metals released into the Environment – Unlike organic chemicals, metals are
neither created nor destroyed by biological or chemical processes. However, metal
speciation/complexation processes can transform metals from one species (e.g.,
valence states) or a complex (with inorganic ligands such as chloride and sulphate) to
another, as well as conversion between inorganic and organic metal forms. Metals
are often released to the environment in various particle sizes, from small particles to
large masses. The type or form of a metal, often characterized by the chemical
species, compound, matrix, and particle size, influences the metal’s bioaccessibility,
bioavailability, fate, and effects. For example, certain forms of metals such as free
metal ion (e.g., Cu+2
) or ionic forms of hydroxyl complexes (e.g., CuOH+) are used
161
for exposure assessments based on their competitive binding to specific biological
sites of action (e.g., fish gill surfaces) for evaluating exposure and effects. The
toxicokinetics and toxicodynamics of metals depend on the metal, the form of the
metal or metal compound, and the organism’s ability to regulate and/or store the
metal. The form of the metal, in turn, is influenced by inherent biogeochemical
characteristics of a metal and environmental properties, such as pH, particle size,
moisture, redox potential, organic matter, cation exchange capacity (CEC), and acid-
volatile sulfides (AVS). However, currently environmental emission data, and
specifically in life cycle inventory data in the context of LCIA, do not specify the
forms of metal emitted to the environment. For the reasons listed above, it is
recommended that such datasets should specify the forms of metal emitted into the
environment. This would enable further improvement of the assessment of total
bioavailability and fate of emitted metals.
� Kinetic modelling of Metal Speciation/Complexation – The method proposed here
assumes that metals listed in an emission inventory are available for instantaneous
distribution in an evaluative environment described by an equilibrium geochemical
model that distributes a metal into its various forms and complexes based on the
geochemical properties and the ambient chemistry of the environment. This
assumption may not be valid in several cases since, except for emissions of soluble
metal salts, most particulate forms of metals emitted undergo a slow dissolution
process. Therefore, it is recommended that speciation and fate calculations
incorporate estimates of media-specific dissolution of the emitted metal species
reported in an inventory database. More specifically, a consensually derived method
of how this should be done is needed, which includes a discussion of time horizons.
� Metal Mixtures – All environmental media have naturally occurring mixtures of
metals and metals are often introduced into the environment as mixtures. The
presence of metal mixtures in the environment has the following implications for
hazard assessment and LCIA: (a) interactions among metals within organisms may
occur when they compete for binding locations on specific enzymes or receptors
during the processes of absorption, excretion, or sequestration, or at the target site, (b)
162
some metals act additively when they are present together in an exposure medium,
many metals act independently of each other, whereas some metals are antagonistic
or synergistic in their toxicological mode of action. Such interactions are important
aspects for assessing overall exposure and toxicity to biota and humans, and (c) the
presence and amount of other metals are important when conducting and interpreting
laboratory tests. A scientifically sound and robust method of estimating potential
impacts of metal mixtures must be incorporated in the multimedia analysis of metal
exposure and effects.
� Essentiality and Tolerance – Several metals are essential micronutrients for
microorganisms, plants, animals, and humans. Nutritional deficits can cause adverse
effects such as increases the vulnerability of organisms to other stressors, including
those associated with other metals. Exposure to elevated concentrations of essential
metals can also result in adverse effects if they overwhelm an organism’s homeostatic
mechanisms. Hence, essentiality should be viewed as part of the overall dose-
response relationship for metals that are micronutrients, recognizing that the dose-
response relationship is metal- and species (biotic) -specific. For a given population,
‘‘reference doses’’ designed to protect from toxicity of excess should not be set
below doses identified as essential. Essential doses are typically life-stage and gender
specific.
� Chemistry Aspects of Metal Background Concentrations – Metals are naturally
occurring constituents in the environment and vary in concentrations across
geographic regions. Since metal speciation-complexation estimates depend in a non-
linear fashion on the metal concentration in an aquatic environment, metal
background concentrations can significantly change the bioaccessibility and
bioavailability of metals. Should hazard, risk and/or LCIA be used on a
geographically specific basis, then metal background concentrations need to be
considered. A consensually derived map of metal background concentrations or
characterization of global ecoregions based on these differences would facilitate this
analysis.
163
� Physiological Aspects of Metal Background Concentrations – As mentioned above,
metals are naturally occurring constituents in the environment and vary
geographically. Humans, other animals, and plants have evolved in the presence of
metals and are adapted to various levels of metals. Many animals and plants exhibit
geographic distributions that reflect adaptation and/ or tolerance to certain metals.
The question that adaptation and tolerance raise is “should, and if so how, could
adaptation and tolerance be included in hazard and risk assessments?”
� Ecosystem Characteristics – The new model framework presented here assumes that
the same ecosystem structure pertains to all water-types. For example, the same
aquatic species were used to calculate effect factors using BLM regardless of water
chemistry. However, particular trophic levels and/or taxonomic groups may not be
present within the water-type of interest. For example, Forbes and Calow (2002)
recognised that the ecosystem structure strongly depends on local environmental
conditions, such as aquatic chemistry and tolerance developed by organisms to
continuous natural exposure over a long period of time. The use of “generic” trophic
levels and/or taxonomic groups in a toxicity assessment within a metal hazard, risk or
LCIA assessment should be further investigated.
� Exposure Routes & Tissue Body Burden Approach – Certain metal compounds are
known to bioaccumulate and exert toxicity in tissues other than fish gills. The new
framework falls short of incorporating other exposure routes such as diet.
Kinetically-based bioaccumulation models (e.g., DYNBAM; Goulet et al. 2007) have
been shown to accurately describe bioaccumulation resulting from different exposure
routes for various metals and aquatic organisms. Models such as this should be
considered as alternatives to the strict use of BLM or FIAM in order to broaden the
assumption that toxicity occurs only at the gill site.
� Atmospheric Emissions of Metals – Metals are often released to atmosphere via stack
emissions and a major fraction of metals is often bound to particulate matter (PM)
although a few metals and metal compounds may exist as vapours or in gaseous form
(e.g., mercury). Limited understanding and thus tools are available regarding
164
atmospheric metal speciation. This area requires further investigation, particularly
given the importance of metal exposure via inhalation of fine, respirable particles.
� Framework Extension to Other Environmental Media – Since the new framework is
generic, it should be readily adapted to estimate bioavailability, fate and toxicity of
metals in other environmental media such as sediment and soil. However,
equilibrium geochemical models are currently limited in their ability to estimate
speciation-complexation under low redox conditions as often occurs in sediment and
in soil environments with high solids content. As such, it is necessary to rely on
empirical estimates of metal distribution among phases in sediment such as the use of
AVS-SEM (e.g., Di Toro et al. 2001). For soil, empirical models (e.g., Sauvé et al.
2000) can be used to estimate this distribution if the soil lies within the set of
conditions for which these statistical relationships were developed. The development
of speciation models for soils is an active area of research and when the models are
available, they should be used to extend emission-fate-toxicity models.
� Framework Extension to Other Metal Forms and Chemical Classes – Finally, we
now know that the information developed on the fate and effects of one form of metal
may not be applicable to other forms. Similarly, this is true for metal salts versus
organometallic forms. Therefore, it is recommended that research be conducted into
the application of this framework to these and other forms of chemicals.
6.5 References
Adams W.J. and Chapman, P.M. (2005) Assessing the hazard of metals and inorganic metal
substances in aquatic and terrestrial systems: summary of a SETAC Pellston
workshop. Pensacola, FL. USA.
Allison J.D., Brown D.S., Novo-Gradac K.J. (1990) MINTEQA2/PRODEFA2, A
geochemical assessment model for environmental model for environmental systems
(EPA/600/3-91-021), Athens, GA: Environmental Protection Agency.
165
Allison J.D., Perdue E.M. (1994) Modeling metal-humic interactions with MINTEQA2. In:
Senesi, N., Miano, T.M. (Eds.), Humic Substances in the Global Environment.
Elsevier, Amsterdam, pp. 927-942.
Apeldoorn. (2004) Declaration of Apeldoorn on LCIA of non-ferro metals. Results of a
workshop by a group of LCA specialists, held in Apeldoorn, NL, April 15th, 2004.
Bhavsar S.P., Diamond M.L., Evans L.J., Gandhi N., Nilsen J., and Antunes P. (2004)
Development of a coupled metal speciation-fate model for surface aquatic systems.
Environmental Toxicology and Chemistry 23(6):1376-1385.
Bhavsar S.P., Gandhi N., Diamond M.L., Lock A.S., Spiers G. and Alfaro De La Torre M.C.
(2008) Effects of estimates from geochemical models on metal fate predicted by
coupled speciation-fate models. Environmental Toxicology and Chemistry 23:2410-
2420.
Bryan S.E., Tipping E., Hamilton-Taylor J. (2002) Comparison of measured and modelled
copper binding by natural organic matter in freshwaters. Comparative Biochemistry
and Physiology Part C 133: 37-49.
Campbell P.G.C. (1995) Interactions between trace metals and aquatic organisms: A critique
of the free-ion activity model. In Tessier A, Turner DR, eds, Metal Speciation and
Bioavailability in Aquatic Systems. John Wiley, New York, pp 45-102.
Christensen J.B., Botma J.J., Christensen T.H. (1999) Complexation of Cu and Pb by DOC in
polluted groundwater: A comparison of experimental data and predictions by
computer speciation models (WHAM and MINTEQA2). Water Research 33:3231-
3238.
Croteau M.N., Luoma S.N. (2008) A biodynamic understanding of dietborne metal uptake by
a freshwater invertebrate. Environmental Science and Technology 42: 1801-1806
Diamond M.L., Gandhi N., Adams W.J., Atherton J., Bhavsar S.P., Bulle C., Campbell
P.G.C., Dubreuil A., Fairbrother A., Farley K., Green A., Guinee J., Hauschild
M.Z., Huijbregts M.A.J., Humbert S., Jensen K.S., Jolliet O., Margni M., McGeer
166
J.C., Peijnenburg W., Rosenbaum R., van de Meent D., Vijver M.G. (2010) The
Clearwater consensus: the estimation of metal hazard in fresh water. International
Journal of Life Cycle Assessment 15:143-147.
Di Toro D.M, Allen H.E., Bergman H., Meyer J.S., Paquin P.R., and Santore C.S. (2001)
Biotic ligand model of the acute toxicity of metals. 1. Technical basis.
Environmental Toxicology and Chemistry 20:2383-2396.
Farley K.J., Carbonaro R.F., Fanelli C.J., Costanzo R., Rader K.J., Di Toro D.M. (2011)
TICKET-UWM: A coupled kinetic, equilibrium and transport screening model for
metals in lakes. Environmental Toxicology and Chemistry DOI: 10.1002/etc.1518.
Forbes V.E., Calow P. (2002): Species sensitivity distributions revisited: A critical appraisal.
Human and Ecological Risk Assessment 8: 473-492.
Gandhi N., Bhavsar S.P., Diamond M.L., Kuwabara J.S., Marvin-DiPasquale M. (2007)
Development of mercury speciation, fate and bioaccumulation (BIOTRANSPEC)
model: application to the Lahontan Reservoir. Environmental Toxicology and
Chemistry 26:2260-2273.
Gandhi N., Diamond M.L., van de Meent D., Huijbregts M.A.J., Peijnenburg W., Guinee J.
(2010) New method for calculating comparative toxicity potential of cationic metals
in freshwater: Application to copper, nickel, and zinc. Environmental Science and
Technology 44:5195-5201.
Gandhi N., Huijbregts M.A.J., van de Meent D., Peijnenburg W., Guinee J., Diamond M.L.
(2011a) Implications of geographic variability on Comparative Toxicity Potentials
of Cu, Ni and Zn in freshwaters of Canadian ecoregions. Chemosphere 82:268-277.
Gandhi N., Diamond M.L., Huijbregts M.A.J., Guinee J., Peijnenburg W., van de Meent D.
(2011b) Implications of considering metal bioavailability in estimates of freshwater
ecotoxicity: Examination of two case studies. International Journal of Life Cycle
Assessment In press.
167
Gandhi, N., Bhavsar, S.P., Diamond, M.L. (2011c) Critical load analysis in hazard
assessment of metals using a unit lake model. Environmental Toxicology and
Chemistry In press.
Gloria T.P., Russell A.J., Atherton J., Baker S.R., Cook M. (2006) Ecological toxicity
methods and metals: An examination of two case studies. International Journal of
Life Cycle Assessment 11 (1), 26-33.
Goulet R.R., Krack S., Doyle P.J., Hare L., Vigneault B., McGeer J.C. (2007) Dynamic
multi-pathway modeling of Cd bioaccumulation in Daphnia magna using
waterborne and diet borne exposures. Aquatic Toxicology 81: 117-125.
Hauschild M.Z., Huijbregts M., Jolliet O., Macleod M., Margni M., van de Meent D.,
Rosenbaum R.K., and McKone T.E. (2008) Building a model based on scientific
consensus for Life Cycle Impact Assessment of chemicals: the search for harmony
and parsimony. Environmental Science and Technology 42:7032–7037.
Huijbregts M.A.J., Thissen U., Guniee J.B., Jager T., Kalfe D., van de Meent D., Ragas
A.M.J., Wegener Sleeswijk A., and Reijnders L. (2000) Priority assessment of toxic
substances in life cycle assessment, I: Calculation of toxicity potentials for 181
substances with the nested multi-media fate, exposure and effects model USES-
LCA. Chemosphere 41: 541-573.
McGeer J.C., Brix K.V., DeForest D.K., Brigham S.I., Skeaff J.M., Adams W.J., and Green
A. (2003) Bioconcentration factor for the hazard identification of metals in the
aquatic environment: a flawed criterion? Environmental Toxicology and Chemistry
22: 1017-1037.
Rosenbaum R.K., Bachmann T.M., Swirsky Gold L., Huijbregts M.A.J., Jolliet O., Juraske
R., Koehler A., Larsen H.F., MacLeod M., Margni M., McKone T.E., Payet J.,
Schumacher M., van de Meent D., Hauschild M.Z. (2008) USEtox—the
UNEP/SETAC toxicity model: recommended characterisation factors for human
toxicity and freshwater ecotoxicity in life cycle impact assessment. International
Journal of Life Cycle Assessment 13:532–546.
168
Sauvé S., Hendershot W., Allen H.E. (2000). Solid-solution partitioning of metals in
contaminated soils: dependence on pH, total metal burden, and organic matter.
Environmental Science and Technology 34: 1125-1131.
Schecher W.D., McAvoy D.C. (1992) MINEQL+ - A software environment for chemical-
equilibrium modeling. Computational Environmental Urban Systems 16: 65-76.
Skeaff J., Delbeke K., Van Assche F., Conard B. (2000) A critical surface area concept for
acute hazard classification of relatively insoluble metal-containing powders in aquatic
environments. Environmental Toxicology and Chemistry 19: 1681-1691.
Sleeswijk, A.W. (2006) GLOBOX – A spatially differentiated multimedia fate and exposure
model. International Journal of Life Cycle Assessment 11: 141-141.
Szebedinszky C., McGeer J.C., McDonald D.G., Wood C.M. (2001) Effects of chronic Cd
exposure via the diet or water on internal organ-specific distribution and subsequent
gill Cd uptake kinetics in juvenile rainbow trout (Oncorhynchus mykiss).
Environmental Toxicology and Chemistry 20: 597-607.
Thakali S., Allen H.E., Di Toro D.M., Ponizovsky A.A., Rooney C.P., Zhao F.J., McGrath
S.P. (2006) A Terrestrial Biotic Ligand Model. 1. Development and application to
Cu and Ni toxicities to barley root elongation in soils. Environmental Science and
Technology 40: 7085-7093.
Tipping E. (1998) Humic ion-binding model VI: An improved description of the interactions
of protons and metal ions with humic substances. Aquatic Geochemistry 4:3-48.
Traas T.P., van de Meent D., Posthuma L., Hamers T.H.M., Kater B.J., de Zwart D.,
Aldenberg T. (2001) The Potentially Affected Fraction as a measure of ecological
risk. In Posthuma L., Suter G.W. II, Traas T.P., eds, Species Sensitivity
Distributions in Ecotoxicology. CRC Press.
Unsworth E.R., Zhang H., Davison W. (2005) Use of diffusive gradients in thin films to
measure cadmium speciation in solutions with synthetic and natural ligands:
169
Comparison with model predictions. Environmental Science and Technology 39:
624-630.
Warnken K.W., Lawlor A.J., Lofts S., Tipping E., Davison W., Zhang H. (2009) In situ
speciation measurements of trace metals in headwater streams. Environmental
Science and Technology 43: 7230-7236.
Webster E., Mackay D., Di Guardo A., Kane D., and Woodfine D. (2004) Regional
differences in chemical fate model outcome. Chemosphere 55:1361-1376.
Yapici T., Fasfous I.I., Murimboh J., Chakrabarti C.L. (2008) Investigation of DGT as a
metal speciation technique for municipal wastes and aqueous mine effluents.
Analytica Chimica Acta 622: 70-76.
Zhang, H. (2004) In-situ speciation of Ni and Zn in freshwaters: Comparison between DGT
measurements and speciation models. Environmental Science and Technology 38:
1421-1427.
170
APPENDIX – A
1. The Clearwater Consensus: the Estimation of Metal Hazard in
Freshwater©
1.1 Abstract
Background, Aim, and Scope. Task Force 3 of the UNEP/SETAC Life Cycle Initiative has
been working towards developing scientifically sound methods for quantifying impacts of
substances released into the environment. The Clearwater Consensus follows from the
Lausanne (Jolliet et al. 2006) and Apeldoorn (2004) statements by recommending an
approach to and identifying further research for quantifying comparative toxicity potentials
(CTPs) for ecotoxicological impacts to freshwater receptors from nonferrous metals. The
Clearwater Consensus describes stages and considerations for calculating CTPs that address
inconsistencies in assumptions and approaches for organic substances and nonferrous metals
by focusing on quantifying the bioavailable fraction of a substance.
Methods. A group of specialists in Life Cycle Assessment, Life Cycle Impact Assessment,
metal chemistry, and ecotoxicology met to review advances in research on which to base a
consensus on recommended methods to calculate CTPs for metals.
Conclusions and Recommendations. Consensus was reached on introducing a
bioavailability factor (BF) into calculating CTPs where the BF quantifies the fraction of total
dissolved chemical that is truly dissolved, assuming that the latter is equivalent to the
© Contents of this chapter have been adopted from the publication in the International Journal of Life Cycle
Assessment:
Diamond ML, Gandhi N, et al. (2010) The Clearwater consensus: the estimation of metal hazard in fresh water.
Int J Life Cycle Assess 15: 143-147.
A link to the published paper can be found at
https://openaccess.leidenuniv.nl/bitstream/1887/14574/2/CB_2010_Diamond_the_clearwater_consensus.pdf
I was primarily responsible for the data collection, illustrative model applications, analyis of model results, and
sensitivity of model parameters to illustrate major aspects of metal speciation and facilitate discussions during
the workshop. I also helped in oraganizing the workshop in Clearwater, Florida.
171
bioavailable fraction. This approach necessitates calculating the effects factor, based on a
HC50EC50, according to the bioavailable fraction of chemical. The Consensus recommended
deriving the BF using a geochemical model, specifically WHAM VI. Consensus was also
reached on the need to incorporate into fate calculations the speciation, size fractions, and
dissolution rates of metal complexes for the fate factor calculation. Consideration was given
to the characteristics of the evaluative environment defined by the multimedia model, which
is necessary because of the dependence of metal bioavailability on water chemistry.
Keywords. Comparative toxicity potentials; Freshwater ecotoxicity; Life cycle impact
assessment; Metal bioavailability; Nonferrous metals.
1.2 Background, Aim and Scope
A group of specialists in Life Cycle Assessment (LCA), Life Cycle Impact Assessment
(LCIA), and metal chemistry and ecotoxicity from academia, industry, and government met
in Clearwater, Florida, USA from November 14 to 15, 2008. The meeting was co-sponsored
by UNEP/SETAC LCA and International Council on Mining and Metals and had
representation from the UNEP/SETAC Life Cycle Initiative through its LCIA Toxic Impacts
Task Force. The goal of the meeting was to recommend a method for developing ecological
comparative toxicity potentials (CTPs) for metal substances that would be consistent with the
current multimedia based practice of setting ecological CTPs for organic substances within
the context of LCIA. The group considered only freshwater ecotoxicity. However, the
principles expressed in this Consensus may also be relevant for expressing the
ecotoxicological hazard of metals in other environmental media such as coastal waters and
terrestrial systems.
The meeting started from the conclusions expressed in the Lausanne review workshop
(Jolliet et al. 2006) and the Apeldoorn Declaration (Apeldoorn 2004), which among others
stated the need to consider metal-specific properties, speciation, and bioavailability when
assessing chemical hazard of metal emissions. This need derives from the intent of LCA to
compare products and processes using a unified framework and specifically in LCIA, to
compare the hazard of all chemicals on a common scale. Following in the spirit of the
172
Apeldoorn and Lausanne statements, the group agreed that differences between organics and
nonferrous metals with respect to the bioavailable form of the chemical result in an
inconsistent assessment of hazard. The group reached consensus on changes to current
practices used to estimate metal hazard that will bring consistency between methods used to
assess and estimate the hazard of organic compounds and metals.
This document is based on the following definitions and assumptions:
The bioavailable fraction of chemical: “[...] the fraction of the total amount of a chemical
present in a specific environmental compartment that, within a given time span, is either
available or can be made available for uptake by (micro)organisms from either the direct
surrounding of the organism [...]” (Peijnenburg and Jager 2003).
Figure A.1: Fractions of total chemical. For metals, the truly dissolved fraction, which is
assumed to be bioavailable, is within the total dissolved fraction. In turn, the fraction of free
metal ion (e.g., Me+2
) is within the truly dissolved fraction.
Total Dissolved In water
Colloidally
bound
Truly Dissolved
Total chemical (mg/kg)
Particulate Irreversibly
sorbed
Reversibly
sorbed
Me+2
173
Fractions comprising total chemical: (Figure 1)
○ Total Chemical = Total Dissolved + Particulate
○ Total Dissolved (or Soluble) = Colloidal + Truly Dissolved.
1.3 Conclusions and Recommendations
Agreement was reached on the following points:
• Metals in Life Cycle Inventory (LCI) – LCI must account for the species and particle
sizes of metals released into the environment. In order to assign appropriate physical–
chemical properties in the LCIA phase, so too must the exact species of metal
released be known. In addition, for alloys, particle size is relevant because it controls
dissolution rates and fate.
• Metal emissions – Current practice assumes that chemicals listed in LCI are available
for distribution in an evaluative environment described by a multimedia fate model.
Since, except for emissions of soluble metal salts, most particulate forms of metals
emitted undergo a slow dissolution process, this assumption is not valid in most
cases. We recommend that fate calculations incorporate estimates of dissolution of
the emitted species of metal reported in an LCI. How this should be done,
particularly which time horizon should be considered for the dissolution process and
what influence metal mineralization has on long-term bioavailability, are topics for
further research.
• Bioavailability in comparative toxicity potentials (CTPs) – Currently, CTPs express
the relative hazard of a chemical as the product of a fate factor (FF) and an effects
factor (EF): CTP = FF × EF. CTPs have been developed for the total chemical
emitted into the environment (reported by the LCI in “elementary” form). The FF is
calculated in terms of total chemical, whereas, the EF is calculated for the total
dissolved fraction, which is comprised of a colloidal fraction and the fraction of truly
dissolved chemical, which is assumed to be bioavailable (see Figure A.1). Current
practice distinguishes between total dissolved (which is often assumed to be truly
174
dissolved) and particulate forms using a particle-to-dissolved partition or distribution
coefficient.1 This approach assumes that chemical uptake by aquatic organisms is
directly from water and does not address dietary uptake.
• Bioavailability Factor (BF): definition – The BF explicitly expresses the relationship
between total dissolved and bioavailable chemical where the latter is assumed to be
truly dissolved. For metals, BF expresses the truly dissolved (not soluble) fraction of
metal. For organics, current practice typically assumes that the total dissolved
fraction, including colloidally bound chemical, is bioavailable despite evidence to the
contrary (e.g., Haitzer et al. 1998).
• Bioavailability Factor (BF) – The bioavailability factor makes the correction
between the total chemical and the truly dissolved fraction that is bioavailable, which
for metals can be based on a Biotic Ligand Model (BLM). We recommend that CTP
be calculated in terms of the bioavailable fraction of chemical, which for organics and
metals, is the truly dissolved fraction and does not include colloidally bound
chemical, i.e., CTP = FF × BF × EF.
• Bioavailability Factor (BF): calculation – A geochemical speciation and
complexation model should be used to calculate the BF as the truly dissolved fraction
of metal in solution based on inputs of water chemistry (e.g., pH, DOC, total
suspended solids or TSS, concentrations of major cations and anions). This
geochemical model must be able to consider the binding of metals to natural DOC.
Presently, WHAM VI is the most commonly known and used model in this category
(Centre for Hydrology and Ecology. Windermere Humic Aqueous Model (WHAM).
Natural Environmental Research Council, NERC, Windermere, UK. 2001). CTPs of
metals that use BFs for which a robust geochemical calculation is not available
should be identified as interim. Further, we recommend that the use of Quantitative
1 For metals, the particle-dissolved distribution coefficient (Kd) is derived empirically or by using a
geochemical model. For organic substances, the organic carbon–water partition coefficient is
calculated, often based on the substance’s octanol – water partition coefficient, KOW.
175
Ion Character-Activity Relationships (QICAR) be explored to obtain BFs for metals
lacking a robust method of calculation (Ownby and Newman 2003).
• Fate Factor (FF) – FFs for freshwaters may be calculated using the SETAC/UNEP
Task Force 3 Consensus Model USEtox (Hauschild et al. 2008, Rosenbaum et al.
2008). The group acknowledges that the consensus model uses a simplified
formulation of net sedimentation that presently does not account for sediment-to-
water diffusive release of soluble metal resulting from post-diagenetic fate processes.
• Effect Factor (EF): metal speciation – EFs for the freshwater ecotoxicity of cations
should be calculated based on the metal’s truly dissolved fraction, assuming that the
free metal ion, which is a fraction within the truly dissolved fraction, is responsible
for toxicity. The free metal ion activity should be calculated using a geochemical
model. The use of the free metal ion activity is reasonable because of the
correspondence between effect concentrations (e.g., EC50) obtained using BLM,
which incorporates a geochemical model, and estimates of the Free Metal Ion
Activity (FIAM).
• Effect Factor (EF): toxicity benchmark – EFs should be calculated based on the
HC50EC50, the geometric mean value of EC50s for chronic ecotoxicity tests for multiple
freshwater biotic species. The HC50EC50 is equivalent to the HC50 obtained from a
species sensitivity distribution or SSD when the statistical distribution of the SSD is
log-normal. In the absence of at least three values of chronic EC50s, the HC50 can be
calculated using acute EC50s based on the correspondence between acute and chronic
test results. A factor incorporating typical acute-to-chronic ratios should be included
in this case.
• Archetypes for freshwaters: the default – The relative value of a metal’s CTP
depends on ambient chemistry. For freshwaters this effect is most important for
bioavailability and toxicity and to a lesser extent, fate. LCA practitioners often do not
have information on the location of emission and will therefore require a default
value with its corresponding variability range. Metal CTPs should be calculated for
one default chemistry (water, pH, DOC, TSS, and concentrations of major cations and
176
anions) chosen to reflect the “central tendency” of European Union (EU) archetypes
and their frequency in emission locations. The European archetypes are well
characterized and used within chemical risk assessment. We acknowledge that this
central tendency does not reflect the central tendency of freshwaters worldwide (see
below). The values of all chemistry parameters should be taken from one archetype
determined to be the central tendency, rather than each chemistry parameter
independently taken as the central tendency over all archetypes. The variability of a
metal’s CTP due to the choice of water archetype should be assessed by giving a CTP
for this default archetype and the extreme maximum and minimum CTP values
obtained for EU archetypes.
• Archetypes for freshwaters: options – CTPs should be calculated for several
freshwater archetypes that relate to the frequency of the occurrence of these
freshwater chemistries and their relevance in terms of proximity to emissions
expressed in the LCI data. Currently, EU water archetypes are available. Future
efforts should be directed towards gathering data to characterize global water
archetypes.2
• Metal concentrations used to calculate CTPs – Metal complexation and speciation
and hence, the BF and CTP, vary nonlinearly with background metal concentrations.
Background concentrations are highly variable among each metal and geographically
at local to global scales (Reinman and Garett 2005). We recommend that as a start,
the default archetype and each of the EU archetypes contain background
concentrations for each metal. Additional research is recommended to evaluate
appropriate metal background concentrations to calculate CTPs.
• Use of internally consistent parameter values – Calculation of each component of
the CTP must use consistent parameter values. For example, a consistent value of
TSS must be used in USEtox to calculate the FF and in the geochemical model to
calculate Kd and BF. Further consideration may be given to the effect of the
2 Determining the relationship between the EU water archetypes and their proximity to emissions
(i.e., current LCI databases) is the responsibility of the UNEP/SETAC Task Force 3.
177
archetype-specific value of TSS on the net sedimentation parameter value used in
USEtox. However, this consideration should recognize the relative insensitivity of
CTPs to fate parameter values, where CTPs are most sensitive to BF. Another
example of the need for consistency is the aerosol settling rate that depends on
particle size of emitted metal. The parameters of the default water archetype should
be used for determining the CTPs for organic substances to provide for a consistent
ranking of CTPs for all substances.
The recommendations contained herein will be implemented under the auspices of the
UNEP/SETAC Task Force 3 to obtain CTPs for several common cationic metals.
1.4 References
Apeldoorn 2004. Ligthart T, Aboussouan L, van de Meent D, Schönnenbeck M, Hauschild
M, Delbeke K, Struijs J, Russell A, Udo de Haes H, Atherton J, van Tilborg W,
Karman Ch, Korenromp R, Sap G, Baukloh A, Dubreuil A, Adams W, Heijungs R,
Jolliet O, de Koning A, Chapman P, Verdonck F, van der Loos R, Eikelboom R, and
Kuyper J (2004) Declaration of Apeldoorn on LCIA of Non-Ferrous Metals.
Workshop by a group of LCA specialists, held in Apeldoorn, NL, April 15th, 2004.
Abstract by Sonnemann G (2004) Int J Life Cycle Assess 9(5): 334. Full text
available at http://www.leidenuniv.nl/cml/ssp/projects/declaration_of_apeldoorn.pdf
Haitzer M, Höss S, Traunspurger W, Steinberg C (1998) Effects of dissolved organic matter
(DOM) on the bioconcentration of organic chemicals in aquatic organisms – a review.
Chemosphere 37: 1335-1362
Hauschild MZ, Huijbregts M, Jolliet O, Macleod M, Margni M, van de Meent D, Rosenbaum
RK, McKone TE (2008) Building a model based on scientific consensus for Life
Cycle Impact Assessment of chemicals: the search for harmony and parsimony.
Environ Sci Technol 42: 7032-7037
178
Jolliet O, Rosenbaum R, Chapman PM, McKone T, Margni M, Scheringer M, van Straalen
N, Wania F (2006) Establishing a framework for Life Cycle Toxicity Assessment -
Findings of the Lausanne Review Workshop. Int J Life Cycle Assess 11: 209-212
Ownby DR, Newman MC (2003) Advances in quantitative ion character-activity
relationships (QICARs): Using metal-ligand binding characteristics to predict metal
toxicity. QSAR Combinatorial Sci 22(2): 241-246
Peijnenburg W, Jager T (2003) Monitoring approaches to assess bioaccessibility and
bioavailability of metals: matrix issues. Ecotoxicol Environ Safe 56: 63-77
Reinman C, Garett RG (2005) Geochemical background – concept and reality. Sci Tot
Environ 350(1-3): 12-27
Rosenbaum RK, Bachmann TM, Swirsky Gold L, Huijbregts MAJ, Jolliet O, Juraske R,
Koehler A, Larsen HF, MacLeod M, Margni M, McKone TE, Payet J, Schuhmacher
M, van de Meent D, Hauschild MZ (2008) USEtox—the UNEP-SETAC toxicity
model: recommended characterisation factors for human toxicity and freshwater
ecotoxicity in life cycle impact assessment. Int J Life Cycle Assess 13: 532-546
Walker JD, Enache M, Dearden JC (2003) Quantitative Cationic-activity relationships for
predicting toxicity of metals. Environ Toxicol Chem 22(8): 1916-1935