improvements in hazard and life cycle impact …...that made this work very exciting. i am also...

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


3.5.1 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


4.4.1 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


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


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.


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.


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












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


workshop; SETAC: Pensacola, FL, 2005.

Bhavsar, S. P.; Diamond, M. L.; Evans, L. J.; Gandhi, N.; Nilsen, J.; Antunes, P.


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


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


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.


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




model. LogKd (suspended particles –

water partitioning coefficient) and total

dissolved fractions of the metals are

listed in Table 4.4.


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





model. LogKd (suspended particles –

water partitioning coefficient) and total

dissolved fractions of the metals are

listed in Table 4.4.




mean of toxicity data which are listed

for each metal in Table 4.4.

Not Applicable

USEtox(new

method)

CTPMe = FF × EF × BF


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



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







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.


4.4.1 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


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


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


USEtox (new)

∑ Water ∑ Air


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


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.


5.4.1 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%



Dissolved Colloidal Particulate

1E+0

1E+2

1E+4

1E+6

1E+8



Kd

ps (

L/k

g)

a

b

0%

20%

40%

60%

80%

100%



Dissolved Colloidal Particulate

Free ion %

in Total Metal

0

20

40

60

80

100



a

b

0%

20%

40%

60%

80%

100%



Me+2 MeOH+ Me(OH)2 MeHCO3+ MeCO3 MeSO4 Other

Free ion %

in Total Metal

0

20

40

60

80

100



a

b

0%

20%

40%

60%

80%

100%



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



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


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


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


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


Bhavsar S.P., Diamond M.L., Evans L.J., Gandhi N., Nilsen J., and Antunes P. (2004)



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


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

improvements in hazard and life cycle impact …...that made this work very exciting. i am also...

Documents