Page 1
Natural capital and distributive justice; a
multidisciplinary, multi-scalar assessment Glenn Althor
Bachelor of Environmental Management
A thesis submitted for the degree of Doctor of Philosophy at
The University of Queensland in 2019
Faculty of Science
Page 2
2
Abstract
People depend on natural capital goods and services for their well-being. However, throughout
history the distribution of the benefits and burdens of natural capital exploitation has not been equal
across people and groups. These unequal distributions can result in matching unequal distributions
of human capability and quality of life. Within this thesis, I engage with distributive justice theories
which allow for an examination of the fairness of distributive patterns and apply these insights to
real world natural capital exploitation cases across diverse geographic scales and locations, using a
range of methodological approaches. I use empirical research methods to describe distributions and
use theoretical frameworks of distributive fairness to evaluate them.
To do this, I sought to meaningfully engage with describing and evaluating the distribution of
natural capital in a way that offers new links between theoretical perspectives and empirical
applications; a critical synthesis for advancing knowledge on distributive justice. I therefore have
pursued an overarching and primary research question: Is it possible to re-frame distributive
justice theories in ways that are more amenable to (a diversity of) empirical research
methods? To answer this research question, the thesis first presents new principles for examining
natural capital distributive justice, developed through theoretical argumentation and presented in
Chapter 1:
P1: Social and economic inequalities, resulting from the utilisation of natural capital, are to
be of the greatest benefit of the least advantaged members of society.
P2: Where harm is accepted as a tolerable by-product of natural capital use for the
betterment of society, exposure levels of harm ought to be equal among all persons.
P3: Where burdens upon any individual or exclusive group are the by-product of the use of
natural capital for the betterment of another individual or exclusive group, burdens are to be
accompanied by (at least) commensurate benefits.
These principles provide a new theoretical contribution to the study of natural capital distributive
justice in that they explain foundational principles from theories of justice in a form suitable for
empirical interrogation. This potential of the new principles, then, is examined in the remainder of
the thesis through multi-scalar cases.
First, in Chapter 2, a systematic review of the natural capital distributive justice is presented in
order to understand the breadth of literature of relevance to this thesis. I find this literature is rapidly
growing, and somewhat diverse, but is dominated by particular geographies, methods,
demographics, livelihood indicators, and topics. Overall, evidence indicates that natural capital
Page 3
3
distributive issues are inequitable, with over half of the literature yielding this result, but with many
unclear determinations and only 2% of studies finding equitable distributions.
Next, in Chapter 3, I statistically analysed the global distributions of benefits and burdens resulting
from global greenhouse gas emissions. In this analysis I find an unjust inequity, which is globally
pervasive, in which nations that have the greatest responsibility for climate change are also those
that are the least vulnerable.
Then, shifting scale from global to national, in Chapter 4 I present statistical modelling that
examines Australia-wide the distributive relationships between industrial pollution and
socioeconomic indicators. Through this detailed empirical analysis, I find little evidence of unfair
distributions, a finding which does not align with existing research. This points to the importance of
policy in shaping the equitability of outcomes and methodological design in how we establish
evidence.
Again, shifting scale from national to local scale, in Chapter 5 I present a qualitative study of the
lived experiences of Cambodian subsistence fisher-people in the context of rapid anthropogenic
environmental change. This study offers a rich and in-depth complement to the prior quantitative
analyses that unpacks the ways in which natural capital distributive inequities affect some of the
world’s most vulnerable people. I find that these communities have little capacity to respond to this
change and are suffering unfair burdens as a result of economic activities from which they derive no
benefit.
Taken together, these studies demonstrate the multiple ways in which natural capital distributive
justice can be examined across scales and with a range of methodological approaches. The thesis
also contributes new understandings about pressing real-world distributive justice issues. In Chapter
6 I offer a synthesis of these findings, returning again to my principles for study of natural capital
distributive justice. This thesis highlights the importance of continued scholarship on natural capital
distributive justice, and I conclude that the four studies show that there are indeed ways in which
theory and methods can be integrated to answer questions about distributive justice in real world
settings. However, my research shows that there is no singular answer to the question of justice in
natural capital distribution; outcomes are context dependent. Critically, I also conclude that natural
capital distributive justice can be evaluated transparently and rigorously through empirical research
supported by my distributive justice principles.
Page 4
4
Declaration by author
This thesis is composed of my original work, and contains no material previously published or
written by another person except where due reference has been made in the text. I have clearly
stated the contribution by others to jointly-authored works that I have included in my thesis.
I have clearly stated the contribution of others to my thesis as a whole, including statistical
assistance, survey design, data analysis, significant technical procedures, professional editorial
advice, financial support and any other original research work used or reported in my thesis. The
content of my thesis is the result of work I have carried out since the commencement of my higher
degree by research candidature and does not include a substantial part of work that has been
submitted to qualify for the award of any other degree or diploma in any university or other tertiary
institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for
another award.
I acknowledge that an electronic copy of my thesis must be lodged with the University Library and,
subject to the policy and procedures of The University of Queensland, the thesis be made available
for research and study in accordance with the Copyright Act 1968 unless a period of embargo has
been approved by the Dean of the Graduate School.
I acknowledge that copyright of all material contained in my thesis resides with the copyright
holder(s) of that material. Where appropriate I have obtained copyright permission from the
copyright holder to reproduce material in this thesis and have sought permission from co-authors for
any jointly authored works included in the thesis.
Page 5
5
Publications included in this thesis
Althor G and Witt B. (2019) A quantitative systematic review of distributive environmental justice
literature: a rich history and the need for an enterprising future. Journal of Environmental Studies
and Sciences online.
– incorporated as Chapter 2.
Contributor Statement of contribution
Althor G (Candidate) Conception and design (100%)
Analysis and interpretation (90%)
Drafting and production (90%)
Witt B Conception and design (0%)
Analysis and interpretation (10%)
Drafting and production (10%)
Althor G, Watson JEM, Fuller RA. (2016) Global mismatch between greenhouse gas emissions and
the burden of climate change. Scientific Reports 6.
– incorporated as Chapter 3.
Contributor Statement of contribution
Althor G (Candidate) Conception and design (80%)
Analysis and interpretation (90%)
Drafting and production (80%)
Watson JEM Conception and design (10%)
Analysis and interpretation (2 %)
Drafting and production (10%)
Fuller RA Conception and design (10%)
Analysis and interpretation (5%)
Drafting and production (10%)
Althor G, Mahood S, Witt B, Colvin RM, Watson JEM. 2018. Large-scale environmental
degradation results in inequitable impacts to communities: a case study from the floating villages of
Cambodia. Ambio 47 (7).
- incorporated as Chapter 5.
Contributor Statement of contribution
Page 6
6
Althor G (Candidate) Conception and design (80%)
Analysis and interpretation (90%)
Drafting and production (80%)
Colvin R Conception and design (5%)
Analysis and interpretation (5%)
Drafting and production (5%)
Witt B Conception and design (5%)
Analysis and interpretation (5%)
Drafting and production (5%)
Mahood S Conception and design (5%)
Analysis and interpretation (0%)
Drafting and production (5%)
Watson J Conception and design (5%)
Analysis and interpretation (0%)
Drafting and production (5%)
Submitted manuscripts included in this thesis
Althor G, Witt B, Colvin RM. (forthcoming) The distribution of industrial pollution across socio-
economic groups in Australia; a ten-year, a fine scale distributive justice analysis. Target Journal:
Environmental pollution
- incorporated as Chapter 4.
Contributor Statement of contribution
Althor G (Candidate) Conception and design (100%)
Analysis and interpretation (100%)
Drafting and production (90%)
Witt B Conception and design (0%)
Analysis and interpretation (0%)
Drafting and production (10%)
Other publications during candidature
Althor G, McKinnon M, Cheng SH, Klein C, Watson J. (2016) Does the social equitability of
community and incentive based conservation interventions in non-OECD countries, affect human
well-being? A systematic review protocol. Environmental Evidence 5 (1):26.
Page 7
7
Brown CJ, Althor G, Halpern BS, et al. (2018) Trade-offs in triple-bottom-line outcomes when
recovering fisheries. Fish and Fisheries 19: 107-116.
Di Marco M, Chapman S, Althor G, et al. (2017) Changing trends and persisting biases in three
decades of conservation science. Global Ecology and Conservation 10: 32-42.
Lewis SC, Perkins-Kirkpatrick SE, Althor G, et al. (2019) Assessing contributions of major
emitters' Paris-era decisions to future temperature extremes. Geophysical Research Letters 46:7.
McIntosh EJ, Chapman S, Kearney SG, et al. (2018) Absence of evidence for the conservation
outcomes of systematic conservation planning around the globe: a systematic map. Environmental
Evidence 7: 22.
Contributions by others to the thesis
No contributions by others.
Statement of parts of the thesis submitted to qualify for the award of
another degree
“No works submitted towards another degree have been included in this thesis”.
Research Involving Human or Animal Subjects
Chapter 5 involved human subjects. Ethics approval granted by Dr Paul Dargusch at the University
of Queensland, GPEM Ethics Officer. Approval number: GPEM number 20150701
Page 8
8
Acknowledgements
I acknowledge the Traditional Owners of the continent now called Australia. In particular, I
acknowledge that I now live and work on land of the Ngunnawal people – the Traditional Owners
of the area now known as Canberra, and I also acknowledge the Traditional Owners of the land on
which the University of Queensland is situated, the Turrbal and Jagera people. I acknowledge that
this land was invaded and colonised by my European ancestors, and that I live in privilege as a
result of stolen land and resources, historic and ongoing racism, and inhumane structural violence. I
acknowledge that the strength and resilience of Aboriginal and Torres Strait Islander people is
astounding, and that Australians like me need to do more in supporting Aboriginal and Torres Strait
Islander people to freely achieve the functionings and capabilities they deem valuable.
Interpersonally, first and foremost I want to give thanks and praise to my partner, best friend,
mentor, teacher, and all-round infinite-patience-machine, Bec Colvin. Without her guidance,
kindness, and companionship this thesis simply would not exist. There’s not a word, thought, idea,
or statistic in this thesis that hasn’t been touched by Bec in some manner. You’re the best!
I also want to thank my Mum, Anille ‘Nelly’ Ellis, who only got to see her eldest son’s first thirteen
years. Even though Fortune only allowed us to spend a shockingly short time together, my Mum
instilled in me an ethic of kindness and empathy for others, a profound love of animals and the
natural world, and a powerful devotion to scholarship and knowledge. Every single day the wisdom,
concern, and care that Mum imparted on me motivates me to work towards supporting the world’s
least represented and most vulnerable people in their fight for the lives and capabilities we all
deserve.
A million and more thanks to my wonderful friend, mentor, and supervisor, Bradd Witt. Bradd’s
disciplined patience, kindness, and support have pushed me over the line many times when I was
facing seemingly insurmountable barriers and thinking this this thesis was an impossible task.
My thanks to my ‘Outlaw’ family – the Colvins and the Hoogleys. All of whom have supported me
with a grace and kindness that I will never be able to repay.
Many thanks to the wonderful folks who gave me the opportunity to discover my love of tertiary
teaching, as a tutor, demonstrator, and lecturer in their courses. Karen McNamara, Lorrae Van
Page 9
9
Kerkhoff, Bruce Doran, Matthew Brookhouse, and Sophie Lewis. Thanks also to all my amazing
friends and family who are sadly too numerous to name here, I hope you know who you are and
how much I appreciate your support!
I also want to thank all the wonderful people who in big and little ways have made this thesis a
reality with kindness and support, and those who I’ve also had the pleasure to work and write with.
It’s been an absolute privilege to work with some of the most amazing minds from around the
world, including: James Watson my co-supervisor; and the folks at WCS Cambodia for their
amazing support during and beyond my fieldwork (particularly Sarah, Simon, Visal, and Ross).
Also including my co-authors (in no particular order, you’re all amazing) Richard Fuller, Madeleine
Mckinnon, Samantha Cheng, Carissa Klein, Simon Mahood, Moreno DiMarco, Sarah Chapman,
Stephen Kearney, Joseph Maina, Oscar Venter, Christopher Brown, Ben Halpern, Sayed Iftekhar,
Elizabeth Pride, Steve Schlizzi, Becky Twohey, Hugh Possingham, Emma McIntosh, Brooke
Williams, Jessica Thorn, Sophie Lewis, Sarah Perkins-Kirkpatrick, Andrew King, Luke Kemp, and
of course all the wonderful academics at ANU and UNSW who I’m currently developing new
research horizons with! Finally, I want to thank all of the folks who have provided other valuable
support and help along the way, particularly Karen McNamara, Ruth Garside, Phil Kokic, Grace
Chou, Genna and the world’s best finance team - SEES finance, and all the amazing and
hardworking UQ librarians.
I wish there was more I could do to show my appreciation for everyone beyond a simple thank you,
but I promise that I ‘pay it forward’, and use your kindness to guide me whenever I have
opportunities to support others.
Page 10
10
Financial support
Chapter 5 was part funded by an Australian Research Council Discovery Grant (DP140100733) to
JW. Additional funding during candidature by The Australian Postgraduate Award scholarship to
GA (PN: 3018293). Funding for translation work provided by WCS, Cambodia.
Keywords
Distributive justice, natural resource management, natural capital, human wellbeing, environmental
management, social justice.
Australian and New Zealand Standard Research Classifications
(ANZSRC)
ANZSRC code: 220104 Human Rights and Justice Issues 34%
ANZSRC code: 050205 Environmental Management 33%
ANZSRC code: 160403 Social and Cultural Geography 33%
Fields of Research (FoR) Classification
FoR code: 0502 Environmental Science and Management 34%
FoR code: 1604 Human Geography 33%
FoR code: 2201 Human Rights and Justice Issues 33%
Page 11
11
Table of Contents
Abstract ................................................................................................................................................ 2
Declaration by author ........................................................................................................................... 4
Publications included in this thesis ...................................................................................................... 5
Submitted manuscripts included in this thesis ..................................................................................... 6
Other publications during candidature ................................................................................................. 6
Contributions by others to the thesis .................................................................................................... 7
Statement of parts of the thesis submitted to qualify for the award of another degree ........................ 7
Research Involving Human or Animal Subjects .................................................................................. 7
Acknowledgements .............................................................................................................................. 8
Financial support ................................................................................................................................ 10
Keywords ........................................................................................................................................... 10
Australian and New Zealand Standard Research Classifications (ANZSRC) ................................... 10
Fields of Research (FoR) Classification ............................................................................................ 10
Table of Contents ............................................................................................................................... 11
Chapter 1 Thesis overview ................................................................................................................. 16
Foreword ........................................................................................................................................ 16
Introduction .................................................................................................................................... 18
Natural capital and distributive justice........................................................................................... 19
Distributive justice ..................................................................................................................... 21
Thesis aim and research questions ................................................................................................. 25
Caveats ........................................................................................................................................... 29
Thesis structure .............................................................................................................................. 30
Chapter 2 A quantitative systematic review of distributive environmental justice literature; a rich
history, and the need for an enterprising future ................................................................................. 31
Overview ........................................................................................................................................ 31
Abstract .......................................................................................................................................... 33
Keywords ....................................................................................................................................... 33
Introduction .................................................................................................................................... 33
Methods .......................................................................................................................................... 35
Systematic quantitative literature review ................................................................................... 35
Search string development ......................................................................................................... 35
Database creation, inclusion criteria, and article screening ....................................................... 36
Page 12
12
Typologies and article categorization ........................................................................................ 37
Data analysis and figure generation ........................................................................................... 37
Results ............................................................................................................................................ 38
The scope of EDJ article types, methods and data ..................................................................... 38
The geographic scope of EDJ articles ........................................................................................ 38
The thematic scope of EDJ articles ............................................................................................ 42
Demographics ............................................................................................................................ 43
Human well-being outcomes ..................................................................................................... 43
Natural resource exploitation issues .......................................................................................... 43
Natural hazards .......................................................................................................................... 44
Equitability of outcomes ............................................................................................................ 45
Relationships between data ........................................................................................................ 46
Discussion ...................................................................................................................................... 48
Study Limitations ....................................................................................................................... 51
Conclusion ..................................................................................................................................... 52
Chapter 3 Global mismatch between greenhouse gas emissions and the burden of climate change . 53
Overview ........................................................................................................................................ 53
Abstract .......................................................................................................................................... 55
Introduction .................................................................................................................................... 55
Methods .......................................................................................................................................... 57
Results ............................................................................................................................................ 59
Discussion ...................................................................................................................................... 62
Conclusion ..................................................................................................................................... 63
Author contributions ...................................................................................................................... 64
Competing Financial Interests statement ....................................................................................... 64
Chapter 4 The distribution of industrial pollution across socio-economic groups in Australia; a ten-
year, a fine scale distributive justice analysis .................................................................................... 65
Overview ........................................................................................................................................ 65
Abstract .......................................................................................................................................... 66
Introduction .................................................................................................................................... 66
1.1 A testable principle for distributive justice .......................................................................... 69
Methods .......................................................................................................................................... 69
2.1 Spatial data ........................................................................................................................... 70
2.2 Industrial pollutant emissions data ....................................................................................... 71
Page 13
13
2.3 Total Toxicity Potential ....................................................................................................... 71
2.4 Socio-demographic Advantage and Disadvantage .............................................................. 72
Results ............................................................................................................................................ 74
3.1 Summary statistics ............................................................................................................... 76
3.2 Proportions of variance explained and statistical significance ............................................ 77
Discussion ...................................................................................................................................... 78
4.1 Study limitations .................................................................................................................. 80
Conclusions .................................................................................................................................... 81
Acknowledgements ........................................................................................................................ 81
Author Contributions ..................................................................................................................... 82
Chapter 5 Large-scale environmental degradation results in inequitable impacts to already
impoverished communities: A case study from the floating villages of Cambodia .......................... 83
Overview ........................................................................................................................................ 83
Abstract .......................................................................................................................................... 85
Keywords ....................................................................................................................................... 85
Introduction .................................................................................................................................... 85
The biophysical dimensions of the Tonle Sap Great Lake ........................................................ 88
Environmental Protection within the Tonle Sap ........................................................................ 90
Threats to the ecosystem and fishery ......................................................................................... 91
Methods .......................................................................................................................................... 92
Prek Toal Biosphere Reserve and surrounding communities .................................................... 92
Data Collection .......................................................................................................................... 94
Data Analysis ............................................................................................................................. 95
Results ............................................................................................................................................ 96
Fishing is critical to local livelihoods ........................................................................................ 96
People believe the fishery is highly vulnerable to environmental change ................................. 97
People believe the fishery is undergoing rapid declines ............................................................ 99
Discussion ...................................................................................................................................... 99
Conclusion ................................................................................................................................... 102
Chapter 6 Thesis summary and conclusion...................................................................................... 104
Modifying distributive justice principles ..................................................................................... 106
Contribution of each research chapter ......................................................................................... 107
Chapter 2: A quantitative systematic review of distributive environmental justice literature; a
rich history, and the need for an enterprising future ................................................................ 107
Page 14
14
Chapter 3: Global mismatch between greenhouse gas emissions and the burden of climate
change ...................................................................................................................................... 109
Chapter 4: The distribution of industrial pollution across socio-economic groups in Australia; a
ten-year, a fine scale distributive justice analysis .................................................................... 110
Chapter 5: Large-scale environmental degradation results in inequitable impacts to already
impoverished communities: A case study from the floating villages of Cambodia ................ 111
Advantages and limitations of the distributive justice approach, and future research ................. 112
Advantages and limitations ...................................................................................................... 112
Future research ......................................................................................................................... 114
Concluding remarks ................................................................................................................. 114
Bibliography..................................................................................................................................... 116
Supplementary materials .................................................................................................................. 133
Chapter 2 Supplementary materials ............................................................................................. 133
Additional file 1 - List of key EDJ studies............................................................................... 133
Additional file 2 - Search string development log ................................................................... 135
Additional file 3 – Variable typologies .................................................................................... 141
Additional file 3 – Data tables ................................................................................................. 150
Additional file 4 Complete dataset ........................................................................................... 157
Chapter 3 Supplementary materials ............................................................................................. 158
S1 Overview ............................................................................................................................. 158
S2 Summary of Lorenz curve .................................................................................................. 158
S3 A summary of per capita GHG emissions and national vulnerability ................................ 158
Supplementary Figure S3 ......................................................................................................... 159
Supplementary table S4 ........................................................................................................... 161
Chapter 4 Supplementary materials ............................................................................................. 172
Supplementary file 1 ................................................................................................................ 172
Supplementary file 2 ................................................................................................................ 195
Supplementary file 3 ................................................................................................................ 196
Chapter 5 Supplementary materials ............................................................................................. 199
Supplementary material S1 ...................................................................................................... 199
Supplementary material S2 ...................................................................................................... 203
Supplementary material S3 ...................................................................................................... 204
Supplementary material S4 ...................................................................................................... 205
Page 16
16
Chapter 1 Thesis overview
Foreword
First and foremost, I want to acknowledge and thank you, the reader of this work. I very much
appreciate your time spent reading this thesis.
Next, I need to discuss my philosophical positionality as a researcher. As scientists, I believe that it
is critical that we understand and acknowledge the moral and philosophical ideas and ideals that we
bring into our work. These (consciously and unconsciously) pre-conceived ideas shape our research
choices (topic, theoretical framing etc.), and are ultimately the lens through which we interpret our
data and findings – I firmly believe that pretending otherwise isn’t helpful for social or scientific
progress, nor is it somehow more objective. As such, I will clearly state how/why I came to
undertake this work, and how this has shaped my interpretation of how my studies model reality.
However, I want to push back on this interpretation (or perhaps straw person of my own making?)
and re-enforce my earlier claim that I (and other, highly accomplished people e.g. Moon et al.
(2019)) firmly believe that as scientists we can be blinded by the notion of pure objectivity, and
forget the moral positing, and inherent biases of the person(s) behind the research – ourselves. This
is eloquently stated by the authors above:
“The philosophical position of the researcher frames their theoretical perspective (i.e., the ideas,
concepts, and assumptions the researcher brings to their research), influencing the kinds of
questions they ask and how they seek to answer them. These elements inform which methodologies
will best suit the philosophy, how theory and the desired research outcome/s are integrated, and the
rationale for the chosen methods. While philosophy might not always appear to drive research, it
will always implicitly underpin the choices made.” (Moon et al., 2019).
It is imperative that we reflect on, and openly state and acknowledge our philosophical
positionality, when and where possible/practicable. This of course requires a deliberate degree of
self-reflection, which I argue is a critical scientific skill. A wonderfully executed example of this
can be seen in Carney (2016), which openly discusses the author’s positionality as an activist,
scholar, and woman of colour in a “Personal Reflexive Statement”. I strongly believe that this is a
standard all researchers should aspire to. In the absence of such transparency we risk making it easy
Page 17
17
for others to misinterpret our work, which can in turn lead to erroneous conclusions, critiques, and
perhaps most importantly can feed into the crisis of scientific replication.
My research positionality is mostly as an extremely concerned outsider to the issues I have
investigated. I have the privilege to live in a very wealthy, and relatively fair and equitable country,
and consider it imperative that I use this privilege to work toward a fairer, and more equitable
world. In so far as I understand these terms, I consider myself an intersectional feminist, a
humanitarian, a strong believer in the purposefulness of a universal moral framework for the
equality of rights and fair treatment of all people, regardless of any form of socially constructed or
biologically determined difference, and I am a staunch advocate for improving the lives and
capabilities of disadvantaged people and communities. The driving force behind this thesis is my
belief that historic and ongoing injustices committed by any group against some ‘other’
(particularly in cases of socially constructed and re-enforced identity) are abominable, only serve to
further the nihilistic whims of a few ‘elites’ while compromising the needs of the majority, and
ultimately serve to hinder the just progress of our pretty wonderful species. I don’t think it’s a
stretch of the imagination to claim that in our conflict riven, and nuclear armed world, which is
facing potential ecological collapse, that the true enemies of keeping humanity extant (and happy,
and healthy) are parochialism, and socially constructed division.
By my layperson’s reading, the collapse of the Roman Empire was driven by the kind of internal
division, parochialism, elitism, environmental change, and racism that we’re seeing rise in
occurrence across the world in contemporary times. While I believe that we need to be wary about
holding up a mirror to history as if it is a one-to-one model of the contemporary world, I do think
we can gain worthy insights from our shared history, and the fate of the Romans paints a grim
picture of where unchecked elitism and social division lead.
Page 18
18
Introduction
It is widely accepted that every living person has intrinsic value (Bradley, 2006). This should make
weighing the value of one person’s life and well-being against another’s impossible. Yet, this is
how our systems of natural resource management and use most often manifest. Throughout human
history, the natural environment has been, and remains, the primary source of material that we
utilise, extract, and exploit, in order to progress our societies, and ultimately improve our collective
capabilities. Be it the ancient (though by no means extinct) practice of hunting wildlife for
sustenance, or the modern utilisation of rare and highly refined minerals in complex technologies,
people have and will always depend on natural capital for personal, social, and economic progress.
However, throughout history, including in our contemporary world, these resources have not been
equally accessible or distributed among individual people, or among our global, multiscalar
communities – neither across time nor space (Gerlagh and Keyzer, 2001). Given that people depend
on natural capital for well-being, these unequal distributions can result in matching unequal
distributions of human capability and well-being. As people, groups, and nations do not exist in a
vacuum and are instead highly interactive, these unequal distributions can result in situations where
the comparatively excessive lot of one group, when contrasted against others, can be reasoned to be
unfair and immoral (Sen, 1995; Sen, 2009).
Within this thesis I empirically identify and describe present day cases of these kinds of divisions
through a distributive justice lens, in the context of natural resource use, with a broad aim of asking
if our national and global institutions are producing fair outcomes. I have not approached this work
with an a priori position that strict equality (i.e. equal distribution of resources regardless of need or
contribution) is either a sufficient or tenable position, neither philosophically nor practically.
Rather, within this thesis I have adopted the position argued by Rawls (1971), and strengthened by
Scanlon (2018) whereby distributive injustices can only be called such if there is sufficient evidence
or moral reasoning to do so. That is to say, inequalities (economic or otherwise) are only unfair, if
they cause harm, particularly if such harm is experienced by any given society’s most vulnerable
people (e.g. economically poor, or ethnically marginalised) (Scanlon, 2018). Furthermore, such
inequalities can be said to be fair (and even desirable), if they result in an improvement in the well-
being of people, again, particularly if the benefit is experienced by the most vulnerable (Crick,
1987; Rawls, 1971).
Page 19
19
I aim to add to the social justice scholarship by combining theoretical distributive justice
frameworks with the tools of empirical science, in the context of environmental problem solving.
Within this framing, and to meet this aim, I have developed and authored several empirical,
descriptive, and evaluative studies (as per definitions in Robeyns (2017)), which make up the body
of this thesis. Each is descriptive in that I have designed them to provide empirical accounts of
phenomena, and evaluative as I have applied normative (largely fairness) values to evaluate my
empirical descriptions. Explicitly, I have described the distributive patterns of natural resource use
across several cases and evaluated them through a lens of distributive justice.
In this thesis, I present four publications, a systematic literature review and three cases. Each is self-
contained and includes an account of its specific background and importance, the methods I used,
my findings, and my conclusions. As such, this thesis consists of six distinct chapters – two
ancillaries (this introduction and an overall conclusion), and four publications – and is also
supported by several technical and method appendices. First, I introduce the key concepts and
theoretical frameworks that I have used, modified, and developed throughout my doctorate, and the
research questions underpinning this body of work. In following chapters, I present the four
empirical studies, organised in ‘spatial’ order, presenting the most geographically broad studies
first, followed by the smaller scale studies. Finally, I summarise each research chapter’s
contribution to the overall thesis, the advantages and limitations of the distributive justice approach,
and present proposals for future research directions and my concluding remarks.
Natural capital and distributive justice
Much of human history, including current times, can be characterised by the exploitation, trade
politics, and wars of natural capital. These purely social phenomena are generally an artefact of
natural resource scarcity and their naturally and socially unequal distributions. This is exemplified
by humanity’s defining historical moments, such as the momentous, exploitation colonialism of
Europe throughout the 15th to 19th centuries, which resulted in unprecedented levels of imperial
growth, trade, wealth creation, transnational politics, and widespread and interrelated horrors such
as genocide, cultural and sovereign extinction, human slavery, and dispossession – all of which
were largely driven by one group’s desire for another’s natural capital (Findlay and O’Rourke,
2012; Gilmartin, 2009; Piketty, 2014).
Page 20
20
In our modern world, covetous nationalism is still very much a primary driver of trade, politics, and
war. However, the disastrous levels of human suffering experienced during the globally scaled wars
of the 20th century, and the substantial though less visceral grief caused by ‘softer’ conflicts such as
the Cold War, have led to a relatively more cooperative world (Copeland, 1996). Rather, it is now
relatively more common for international trade and foreign politics to further national interest,
rather than spilled blood and spent munitions. While the modern world is far from utopian, it is
undeniable that in the last century there has been critically important and meaningful progress
beyond simplistic, uncooperative national parochialism. While national self-interest is far from a
spectre of the past, today the majority of the world’s nations and their leaders subscribe (though to
widely varying degrees) to moral frameworks that are relatively much more universal in
recognising social justice, such as the rights of individuals, groups, and nations beyond their own
interests (e.g. quite literally, as signatories of international treaties such as the Universal Declaration
of Human Rights (UN General Assembly, 1948), and more recently the Vienna Declaration and
Programme of Action (OHCHR, 2019)). Such progress has not happened by accident. Rather, in
many areas of the world, the larger part of the last century has often been (and still is) marked by
the extraordinary efforts of political activists, civil rights movements/leaders, philosophers,
academics, politicians, and every-day-people, arguing and fighting for the recognition of their own
rights, or the rights of other people (Davis, 1981; Freischlag and Faria, 2018; Carney, 2016).
However, while the relatively recent progress of social justice and equality has been incredible in its
scale, it has hardly been perfect, nor is it by any means complete, or is a path to global
egalitarianism certain (not even an egalitarianism of so called ‘basic needs’). With even a cursory
glance at the state of the world any observer can see that it is typified by inequality and social
hierarchies that result in deep injustices and unfair distributions of power, wealth, and natural
capital (Piketty, 2014). Globally, income inequality is rising alarmingly, resulting in the repugnant
and absurd situation where some individuals have more wealth than entire nations (Bagchi and
Svejnar, 2015), while hundreds of millions of people are materially deprived of the basic needs
required to realise their full potential (Sumner, 2016; Sen, 1999). Armed conflict is still associated
with natural capital exploitation (Ayres, 2012), and the inequitable nature of greenhouse gas
emissions and the resulting climatic effects see industrialised economies achieving historically high
profit levels, while entire small island nations are sinking under water (Willcox, 2016). And so on
and so forth.
Page 21
21
As such, it is clear that much work remains to be done if human societies are to address
recognizably harmful forms of inequality. Within this thesis I will use the context of natural capital
distributions to demonstrate how an initial stage of this process is to use empirical research methods
to describe distributions and use theoretical frameworks of distributive fairness to evaluate them. In
doing so, I hope to (humbly) contribute alongside the traditions, research, and extraordinary efforts
of political activists, civil rights movements/leaders, philosophers, academics, politicians, and
every-day-people, arguing and fighting for the recognition of their or others equal rights to a live
they deem worth living (Davis, 1981; Freischlag and Faria, 2018; Carney, 2016).
Distributive justice
It is important to note that an exhaustive discussion of the history and philosophies of justice and
fairness are beyond the scope of my thesis, which instead focuses on the distributive justice theories
that I have utilised within this thesis, which I believe (as per the forward) best suit my analytical
approach and support my broader aims of using research to further the equality of rights and fair
treatment of all people. As such, the core theoretical framework that I use within this thesis, is a
modified form of Rawls’ difference principle - arguably the most influential and impactful theory of
distributive justice (Allingham, 2014; Olsaretti, 2018). Positive economics alone cannot, without
the guidance of normative principles, recommend which policies, structures, or institutions to
pursue. As such there has been meaningful theoretical progress over the last century, for examining
and analysing the fairness of natural resource distributions, and how such analysis can be used to
guide social decision making such as policy.
Distributive justice theories allow for an examination of these frameworks and their resulting
distributive patterns (Lamont and Favor, 2017; Allingham, 2014). Through distributive justice
approaches, we are able to measure and normatively evaluate how the social, environmental, and
economic constructs and contexts of a society (e.g. laws, policy, natural resource access, taxes,
welfare) result in a diverse array of benefit and burden distributions. Furthermore, we can assess
how these distributions differ across social and demographic groups, and (also normatively)
evaluate if these distributive structures affect people’s capabilities and by extension well-being
(Sen, 1995). Additionally, and most importantly for this thesis, distributive justice frameworks
provide clear ethical frameworks through which we can make normative claims of what are just or
fair distributions (Allingham, 2014). The application of distributive justice theories (among others,
e.g. Gross (2008)) provides researchers, governments, policy makers, and activists an outline of
Page 22
22
moral guidance for how social, environmental, and economic frameworks ought to be shaped, in
order to actually change distributions in a fair manner1.
Due to an ever-increasing utilisation of natural capital globally, there are likewise an ever-
increasing number of environmental justice issues surfacing. Some of these issues can be highly
complex, and require an in-depth examination of associated procedural, institutional, or social
justice. However, for this thesis I have elected to examine cases where understanding each person
(or group’s) distributive due ought to be fairly easy to describe and analyse. This is not to say that
more complex issues are more or less important, however I argue (and will use this work to support
the notion) that, particularly in environmental issues, there are cases where examining, describing
burden and benefit distributions among groups, and reaching conclusions about how fair they are,
can be a relatively simple process. I further argue that understanding and undertaking these (again,
relatively) simple analyses can be a time and resource efficient method for establishing a
distributive justice baseline or ‘snapshot’ of an issue. Such a baseline is intended to be useful for
interested parties (e.g. researchers, policy makers, NGOs, or global intergovernmental
organizations) to understand the landscape of EDJ issues and prioritise actions to better understand
or address them. As such, I have deliberately selected cases of natural resource use where I can use
relatively simplistic forms of distributive justice to understand their fairness. This is not to say I
have selected unimportant issues; I have still elected to examine cases which are critical for the
well-being of groups of people. Furthermore, I have deliberately selected cases at widely diverse
scales, from a global issue between nations, to a local issue affecting a small group of rural villages.
Furthermore, distributive justice theories allow for the explicit justification of ethical choices. Often
the ethics or morality underlying decision making, or assessments of distributions are simply
implied (e.g. It is wrong that air pollution occurs more in places where particular racial groups live).
While this is potentially sufficient, by instead making fairness claims through a lens of distributive
justice we are able to argue in a clear, logical, and rational framework, which I believe strengthens
our position.
In this light, distributive justice is not simply a series of moral theories. Rather, they are best
thought of as a highly practical set of theoretical tools that we can use to describe, examine and
resolve real world problems (Lamont and Favor, 2017). I argue that within a universal moral
1 I must stress that distributive justice outlines guidance, as while it may be possible to do so, it is best that distributive
justice is not applied in a vacuum, and instead as part of a greater equitability process, which includes other critically
important forms of justice such as procedural and relational fairness (as per Gross 2009).
Page 23
23
framework of equality of rights and fair treatment of all, distributive justice theories are enormously
useful in describing and evaluating scientific data for use in real world dilemmas. Through the
application of these theoretical tools, I have been able to study physical and social phenomena
through a lens of the moral right to fairness in the distribution of benefits and burdens associated
with natural capital among different groups of people. Furthermore, I argue (and illustrate) that
distributive justice principles as applied to scientific enquiry can facilitate practical examination and
provides tools to shape fair policy pertaining to natural capital. It is important however to
acknowledge that distributive justice is neither perfect, nor infallible, or without critics (particularly
by libertarians (e.g. Nozick (1974); Hayek (1976)). However, these limitations are useful as they
highlight the normative nature of distributive justice claims and why we ought to avoid making
universal claims via distributive justice.
As such, and as argued elsewhere (Lamont and Favor, 2017), I contend that the most important
strength of distributive justice theory is its practical applicability, an idea which I will demonstrate
throughout. In particular, I have made extensive use of a modified form of John Rawls’ Difference
Principle (Rawls, 1971; Rawls, 1993; Allingham, 2014; Lamont and Favor, 2017; Rawls, 2001).
Specifically, I utilise and modify part (b) of Rawls’ second principle of justice, which (in full)
states:
“Social and economic inequalities are to satisfy two conditions: (a) They are to be attached to
positions and offices open to all under conditions of fair equality of opportunity; and (b), they are
to be to the greatest benefit of the least advantaged members of society”
Specifically, I have modified this principle to be appropriate to the examination of issues regarding
distributive justice and natural capital, and have removed condition (a) as my work does not
examine liberty or governance. As follows (and as utilised in Chapter 5):
P1: Social and economic inequalities, resulting from the utilisation of natural capital, are to be of
the greatest benefit of the least advantaged members of society.
Furthermore, I also utilise the simplest distributive justice principle - strict egalitarianism. Strict
egalitarianism states (Lamont and Favor, 2017):
Every person should have the same level of material goods (including burdens) and services.
Page 24
24
In practice, this means uniformity of access to resources for all people, regardless of need or
contribution. It is important to acknowledge the severe limitations of such a rigid and strict theory
of distribution. In a complex world, strict egalitarianism is both problematic and utopian, and as
such is rarely seriously considered in the empirical literature, and I would argue is impractical or
useable in a general manner for distributing (or re-distributing) resources. However, I argue that it
does have an important role, specific to the distribution of harm. If harm is caused by economic
activities (under the assumption that such activities are accepted/tolerated (by law) as they benefit
all of society, e.g. via taxation, economic growth) such as industrial production, any unequal
distribution of this harm, is unacceptable. As such, I have modified this principle as follows (and as
utilised in Chapter 4):
P2: Where harm is accepted as a tolerable by-product of natural capital use for the betterment of
society, exposure levels of harm ought to be equal among all persons.
Furthermore, these two principles may be combined in cases where some harm (or other type of
burden) is produced, for the specific benefit of a single person, or exclusive group, as follows (and
as utilised in Chapter 3):
P3: Where burdens upon any individual or exclusive group are the by-product of the use of natural
capital for the betterment of another individual or exclusive group, burdens are to be accompanied
by (at least) commensurate benefits.
It is critical that I acknowledge that this combined principle is potentially problematic if it were to
be misunderstood or misused, as it does not acknowledge differential power or agency in the choice
of exposure to burdens or benefits of one group due to the activities by another2. The purpose of this
principle is for purely pragmatic applications, for example examining cases where the power
dynamics between groups are realistically unable to be affected (as I will illustrate and discuss in
Chapter 5). As such, it is my intention that it should only ever be applied when assessing existing
distributions, and never for the purposes of policy or project design.
2 While agency is morally paramount, its examination is outside of the scope of this thesis, though I must again stress
that distributive justice approaches ought not be applied to real-world situations in a vacuum, rather their application
should take place alongside other analytical forms examining justice.
Page 25
25
As these principles make up the key theoretical foundation of this thesis, I will utilise them
throughout to justify the relevance of each research chapter and associated research questions, to the
overall thesis aims. For brevity they will be referred to henceforth as “modified distributive justice
principle(s) P1, P2, or P3”.
In summary, the main purpose of this thesis is to provide several cases (studies) that highlight how
distributive justice frameworks can be applied to empirical research in order to strengthen
normative arguments around fairness. The ever-increasing utilisation of natural capital globally is
giving rise to similarly increasing environmental justice issues. Many of these issues are highly
complex, and require an in-depth examination of associated procedural, institutional, and relational
fairness. However, for this thesis I have elected to examine cases where understanding each person
(or group’s) distributive due ought to be fairly easy to describe and analyse. This is not to say that
more complex issues are more or less important, however I argue that, particularly in environmental
issues, there are cases where examining, describing burden and benefit distributions among groups,
and reaching conclusions about how fair they are, can be a relatively simple process, and an
important first step to their resolution. By describing and evaluating cases through these relatively
simple distributive justice approaches we gain a time and resource efficient method for establishing
a fairness baseline or ‘snapshot’ of an issue. Such a baseline is likely to be highly useful for those
wanting to broadly understand the landscape of distributive justice issues, and to develop and
prioritise follow actions to better understand or address them. As such, I have deliberately selected
cases of natural resource use where I can use relatively simplistic forms of distributive justice to
understand their fairness. This is not to say I have selected unimportant issues; I have still elected to
examine cases which are critical for the well-being of groups of people. It is my hope that the
studies I present will both provide value in and of themselves, while forming part of a broader
narrative around the importance of distributive justice theories in empirical research.
Thesis aim and research questions
This thesis pursues the overarching research question of:
Is it possible to re-frame distributive justice theories in ways that are more amenable to (a diversity
of) empirical research methods?
In order to answer this question as systematically as a doctoral thesis permits, I have developed four
broad sub-questions. Each of these sub-questions specifically utilises one or more components of
Page 26
26
my modified distributive justice principles and is answered via four empirical examinations
(presented here as Chapters 2 to 5).
This is divided into the following sub-questions, which are variously utilised in each research
chapter:
1. Sub-question 1: Can distributive justice principles be re-framed and applied in a manner
that allows for the development and application of a systematic review of the natural
resources literature?
2. Sub-question 2: Can the benefit and burden distributions between people or groups,
resulting from natural resource exploitation, be empirically examined using modified
distributive justice principles?
3. Sub-question 3: Are distributive justice principles able to be modified to allow for a
quantitative empirical examination of the distribution of harms resulting from natural
resource exploitation?
4. Sub-question 4: Are distributive justice principles able to be modified to allow for a
qualitative empirical examination of the distribution of harms resulting from natural
resource exploitation?
As my modified principles have the potential for a diverse array of applications, they demanded a
diversity of empirical responses at this initial exploratory stage, to highlight their flexibility. As
such, the research chapters herein are diverse in their geographic scale, empirical methods, and
scale of outcomes. In doing so, it was my intention to highlight, indicate, and paint as broad a
picture of the potential diversity of empirical applications for my modified distributive justice
principles (as time and resources would allow).
Chapter specific research questions
The following is a breakdown of each of the research chapters’ research questions. For clarity and
due to each chapter being written to appear as a standalone peer-reviewed manuscript, I have opted
to delineate between thesis and study specific questions. As such, this section is structured by
specifying 1) the thesis research question(s), which explicitly link each to the overall thesis question
and the relevance to my modified distributive justice principles, 2) a list of which modified
principles I have applied, 3) the study specific research questions and sub-questions, which treat
each chapter as a self-contained study such as they might (or do in some cases) appear in a
Page 27
27
standalone peer-reviewed manuscript, and 4) a brief overview of how these questions are addressed
in the relevant chapter.
Chapter 2: A quantitative systematic review of distributive environmental justice literature; a
rich history, and the need for an enterprising future
Thesis sub-question 1: Can distributive justice principles be re-framed and applied in a manner
that allows for the development and application of a systematic review of the relevant literature?
Modified distributive justice principles utilised: P1, P2, and P3.
Study specific research questions: What is the depth and breadth of the distributive justice and
natural capital literature?
Study specific sub-questions:
1) Where are the research institutions of the authors of this research located;
2) What has been the geographic focus of the literature;
3) What has been the demographic, environmental, and human well-being scope of the literature;
4) What types of publications and study methods have been used; and
5) What are the gaps and areas for future focus in the literature.
These questions are answered in Chapter 2 by applying all three modified distributive justice
principles (P1, P2, and P3) as a framework to develop and undertake a quantitative systematic
review of the literature. This approach adheres to the method developed by Pickering et al. (2014),
and summarises the literature using a rigorous and robust, replicable review method.
Chapter 3: Global mismatch between greenhouse gas emissions and the burden of climate
change
Thesis sub-question 2: Can the benefit and burden distributions between people or groups,
resulting from natural resource exploitation, be empirically examined using modified distributive
justice principles?
Study specific research question: How are greenhouse gas emissions and the burdens of climate
change distributed globally?
Page 28
28
Study specific sub-questions:
1) What is the extent of global climate inequity; and
2) How is this expected to change in the near future.
In Chapter 3 these questions are answered in a study wherein I utilised my modified distributive
justice principles P1, and P3 to develop and undertake the first global level study to analyse the
distribution of burdens and benefits between nations resulting from greenhouse gas emissions
(GHGs).
Chapter 4: The distribution of industrial pollution across socio-economic groups in Australia;
a ten-year, a fine scale distributive justice analysis
Thesis sub-question 3: Are distributive justice principles able to be modified to allow for a
quantitative empirical examination of the distribution of harms resulting from natural resource
exploitation?
Study specific research question: Are there social inequities in the distribution of exposure to
harmful industrial pollution across Australia?
Study specific sub-questions:
1) Have these distributions changed over the last 10 years; and
2) If any, which social or economic groups are more likely to be exposed to industrial pollution.
These questions were developed and answered in Chapter 4 where I used my second modified
distributive justice principle (P2), to develop and undertake a study where I quantitatively modelled
the relationships between industrial pollution exposure and socio-demographic distributions across
the nation of Australia.
Chapter 5: Large-scale environmental degradation results in inequitable impacts to already
impoverished communities: A case study from the floating villages of Cambodia
Thesis sub-question 4: Thesis sub-question 3: Are distributive justice principles able to be modified
to allow for a qualitative empirical examination of the distribution of harms resulting from natural
resource exploitation?
Page 29
29
Study specific research question: How are the fishing communities of the Tonle Sap floating
villages experiencing major environmental changes?
Study specific sub-questions:
1) How are floating village communities experiencing a change in fisheries; and
2) How are floating village communities responding to these changes.
Finally, in Chapter 5 I show how I used my second modified distributive justice principle (P2) to
develop and answer these research questions through a qualitative method which captures the
harmful lived experiences of Cambodian subsistence fisher people in the context of an environment
which is rapidly changing due to interventions beyond their control.
Caveats
A note on terminology
The modified principles described above, and associated terminology, were developed over the
early stages of my PhD, but not completely before publications had been finalised and published.
As such, there are some cases of varying terminology applied throughout. To make the connection
between principle and study clear, I have included a short overview before each research chapter,
reiterated the conclusion, and described how each relates to the underlying theory and overall
research questions.
A note on GDP
Throughout I have made statements regarding GDP which may be read as problematic, such as in
Chapter 2 where I state: “…economic statistics (such as GDP) can be used as a coarse analogue for
human well-being…”. Such statements are not intended to be normative claims about GDP as a
good, worthwhile, non-problematic analogue for well-being, or that there are not far more
appropriate quantitative measures of human well-being such as the Human Development Index, or
Genuine Progress Indicator (as I qualify in the text following the above quote “…, [GDP] offer[s]
only limited insight into important dimensions such as capabilities” (Sen, 2009)). Rather their use is
intended to reflect the reality that economic growth reigns supreme in the minds and policies of the
majority of the world’s decision makers, such as national governments, as such GDP is the most
commonly used statistic (to assess well-being) by those with power (Ward et al., 2016). As such, I
wholly support the evidence based research that shows how GDP is problematic, but is also
appropriate as it speaks in the “language” of policy makers (Kenny et al., 2019; Sen, 2009; Ward et
Page 30
30
al., 2016). So, while I strongly believe that other quantitative metrics of well-being are important,
they are orthogonal to my purpose here.
Thesis structure
I have structured this thesis as follows:
1. Foreword material including this introductory chapter.
2. Four chapters based on the four distinct studies I have developed and undertaken as part of
this thesis. Each contains:
a. A brief overview discussing how they relate to the overall distributive justice
theories underpinning this thesis, the research questions they answer, and highlights.
b. The actual study, each presented as a peer-reviewed publication. Unless otherwise
stated they appear either exactly as published for currently published/under review
works, or how they are expected to appear for forthcoming works.
3. Concluding material.
4. A bibliography which aggregates all works cited throughout this thesis, including for each
individual study, and all ancillary texts.
5. Supplementary material for each chapter. Unless otherwise stated they appear either exactly
as published for currently published/under review works, or how they are expected to appear
for forthcoming works.
Page 31
31
Chapter 2 A quantitative systematic review of distributive
environmental justice literature; a rich history, and the need for an
enterprising future
Full citation: Althor G and Witt B. (2019) A quantitative systematic review of distributive
environmental justice literature: a rich history and the need for an enterprising future. Journal of
Environmental Studies and Sciences. Online.
Overview
This study is currently under review. It was designed to use the quantitative systematic review
method developed by Pickering et al. (2014). The overall aim of this study is to capture the depth
and breadth of literature which relates to natural capital and distributive justice.
Research question: Does existing literature on distributive justice have evident biases or limitations,
and what is the opportunity for use of empirical evidence to address these?
Sub-questions:
1) What is the depth and breadth of the distributive justice and natural capital literature?
2) Where are the research institutions of the authors of this research located;
3) What has been the geographic focus of the literature;
4) What has been the demographic, environmental, and human well-being scope of the literature;
5) What types of publications and study methods have been used; and
6) What are the gaps and areas for future focus in the literature.
The resulting ‘map’ of the literature was used to identify key knowledge gaps, which were further
used to develop and select specific topic areas for Chapters 3, 4, and 5. All of the theoretical
principles underlying this thesis P1, P2, and P3 (as described in Chapter 1) were used as a guide to
decide on the relevance of each individual study included in this systematic review.
Highlights:
• First systematic review to specifically seek to understand the depth and breadth of the
natural capital and distributive justice literature.
Page 32
32
• Rigorously applied systematic review method which reduced over 46,000 potential studies
down to a final list of 354.
• Provides a key resource for easily finding all literature under particular sub-topics and
identifying topical knowledge gaps in the literature.
Page 33
33
The following text appears as largely as published, with minor amendments based on thesis
examiner feedback.
Abstract
Environmental distributive justice contextually assesses social equities in relation to natural
resources. While there is a rich literature on environmental distributive justice there have been few
assessments quantifying the biases and scopes of this literature. We conduct a systematic review of
the literature. We find several key biases and summarize the breadth of subjects that have been
studied or discussed. We find a very distinct overlap between authorship nationality and study
location, which is concerning, as some of the world’s most polluted and inequitable societies are
least represented in authorship. Additionally, we find a dominance of quantitative studies. These
results are important for understanding both where future research efforts in this area could best be
directed, and how the literature could be enriched by diversified approaches. Improving
environmental justice studies is critical and important for many people across our global society,
which is increasingly shaped by widespread natural resource depletion.
Keywords
Distributive justice, natural resource management, environmental exploitation, burdens, benefits,
climate change, pollution.
Introduction
Throughout human history the use of natural resources has been the key means by which societies
have created social and economic progress. By exploiting the natural environment, societies are
able to make the places they live more habitable for humankind, and produce goods and services to
improve their well-being. This is (and has been) the case for all societies, from the earliest of
civilizations, whose existence was only possible due to the innovation of agriculture, to
contemporary consumption-based societies, which exploit natural capital globally for the
production of primary and secondary goods (Pearce and Turner, 1990).
While societies may benefit from exploiting natural capital, these processes rarely come without a
social cost. These costs take various forms and can result in a diverse range of human well-being
burdens. Examples of these issues are diverse (Brulle and Pellow, 2006), and can include a range of
issues from the loss of amenity and recreation due to water pollution in Australian lakes (Crase and
Gillespie, 2008), the ruin of local livelihoods due to deforestation in Pakistan (Shehzad et al., 2014),
to increases in cardiovascular disease due to exposure to air pollution in the USA (Brunekreef and
Page 34
34
Hoffmann, 2016). The key message from this literature is: societies exploit natural capital to
produce a diverse range of well-being benefits, however in doing so they also generate a diverse
range of social costs. While analysing and understanding this message is important for improving
human well-being, it does not elaborate on two critical and interlinked social issues: 1) How are the
costs and benefits, which result from the non-sustainable exploitation of natural capital, distributed
across social groups? and vitally 2) How fair are these distributions?
These questions are important as they highlight the potential imbalance between the benefits and
burdens of environmental exploitation among different social groups. Recognising and addressing
such imbalances is imperative, as they can lead to unjust outcomes for particular groups. Within the
broader social justice literature, issues of cost and benefit imbalances are commonly referred to as
issues of ‘distributive justice’ (Allingham, 2014). Understanding and describing distributive justice
in broad social issues has long been the preoccupation of many of the world’s preeminent political
philosophers (such as Rawls, Sen, Dworkin, and Nozick (as per Sen, 2009)). There has also been a
relatively recent increase in scientific research aiming to empirically quantify a diverse range of
distributive justice issues within and across societies (e.g. Essoka (2010); Rodriguez-Lara (2013)).
Furthermore, there has been a marked increase in research which specifically investigates the
distributive justice dimensions of the exploitation of natural capital (e.g. Dobbie and Green (2015);
Kyne and Bolin (2016)). Environmental distributive justice (EDJ) research is increasingly critical as
natural resource exploitation becomes ever more widespread, and as resultant benefits and burdens
progressively affect more people (Althor et al., 2016a). The practical benefit derived from EDJ
research is to highlight issues where particular populations may be unfairly burdened by
environmental exploitation. As such, EDJ studies can be used as a guide for decision makers. While
there have been significant and meaningful efforts to synthesize the geographic distribution of
environmental justice issues (e.g. The EJ Atlas (Temper et al., 2015)), and rigorously examined and
characterized in narrative style literature reviews (e.g. Brulle and Pellow (2006); Mohai et al.
(2009); Mohai and Saha (2015); Banzhaf et al. (2019)) the literature has not been systematically
quantified until now. As such this study significantly compliments existing works by providing a
unique, systematic quantification of the literature.
Here we utilise the systematic quantitative literature review method (Pickering et al., 2014), to
interrogate the peer-reviewed literature that regards distributive justice and the non-sustainable
exploitation of natural capital. This allows us to map the available literature both conceptually and
diagrammatically, and to answer several important questions: 1) Where are the research institutions
of the authors of this research located; 2) What has been the geographic focus of the literature; 3)
What has been the demographic, environmental, and human well-being scope of the literature; and
Page 35
35
4) What types of publications and study methods have been used. We use the data and analyses
generated by answering these questions to not only quantify the existing literature, but also to
identify potential research gaps and key trends. Our study provides insights into the nature of the
EDJ literature and ensure that future research in this area targets these gaps.
Methods
Systematic quantitative literature review
We undertook a systematic quantitative literature review using a well established method (see
Griffith University (2019) and Pickering (2014) for an overview of the method and a list of
publications utilising it). This method was developed in order to systematically search for and
categorize the literature on a particular academic topic. The purpose of this process is to quantify
the scope of the literature, and identify historic strengths or biases. Such information can be used to
identify potential knowledge gaps, and inform future research (CEE, 2013).
While this method is excellent for quantitatively examining the characteristics of any given topic
area, it is worth noting that it has limitations. Most importantly and by design, the method does not
wholly synthesize the literature, neither in a narrative nor analytical manner. As such theoretical
syntheses and meta-analyses often found in traditional systematic reviews are beyond the scope of
quantitative reviews.
Search string development
The purpose of our study was to find and quantify the peer-reviewed literature specifically relating
to EDJ. As such, we did not include articles which exclusively examined other aspects of
environmental justice (such as procedural justice), though multidimensional articles including EDJ
dimensions were retained.
To conduct our search do so we developed a Boolean style search string using the online academic
database Web of Knowledge (WOS) (Thomson Reuters, 2017). Google Scholar was deliberately
avoided, as it has been found to be inappropriate as a principal search system (Gusenbauer and
Haddaway, 2019). Further, due to time and resource constraints, we decisively chose to limit the
focus and targets of our searches (both for search string development, and final searches) to the
most relevant, and high-quality peer reviewed literature. Explicitly, we limited searches to peer-
reviewed articles written in the English language, within research areas relevant to EDJ, and
accessible from in three key academic literature databases, WOS, SCOPUS, and Proquest. These
databases were selected as they cover a large proportion of the academic literature.
Page 36
36
Our search string, and list of relevant research areas, were developed over an iterative five-step
process whereby we: 1) developed an initial search string based on key EDJ, distributive justice,
and environmental exploitation articles (Bournay, 2017; McKinnon et al., 2016; Shortt et al., 2012;
Heyward, 2007; Tornblom, 1977; Cook and Hegtvedt, 1983); 2) tested search string in WOS by
topic (denoted “TS” in WOS); 3) sorted results by relevance (a WOS parameter); 4) reviewed the
first 100 titles, screening for any EDJ terms missing from the initial search string; 5) updated the
search string with new terms, and repeated steps 2, 3 and 4, until the first 100 titles all included at
least one relevant EDJ term, finalizing the search string (for a log of search string development see
supplementary materials). Research areas were determined from these 100 articles, using the WOS
results analysis tools (Thomson Reuters, 2017). The final search string was run on the three key
academic databases, WOS, SCOPUS, and Proquest. We modified the search string to fit each
database’s particular Boolean parameters, and research areas.
The final search string produced a very large number of results (n = 46,860), and given the limited
resources available for this project, further refinements were required. In order to reduce the number
of results, we undertook a coarse screening of titles by only including in our search results those
papers which also included a distributive justice term in the title (denoted “TI” in WOS). While this
process reduced results to a manageable number (n = 2008), it likely resulted in an unknown
number of EDJ articles being missed. However, we attempted to address this by conducting a final
reliability check to ensure results were appropriate and accurate, by verifying that the final results
included several key EDJ studies (n =10) identified prior to any searches through preliminary
readings of the literature (for list see supplementary materials). As such, our review presents a
systematic insight into the EDJ literature.
Database creation, inclusion criteria, and article screening
We exported the final list of articles as citations to Endnote reference manager software (Thomson
Reuters, 2015), which we then exported to a Microsoft Excel database. We used this database to
conduct all article screening. Before commencing article screening we first developed an exhaustive
list of 19 inclusion and exclusion criteria which were used throughout the screening process to
identify which articles should be kept, and which discarded. The overall ‘theme’ of the inclusion
criteria was to: Identify peer-reviewed articles which explicitly relate to the distributive justice
dimensions of the non-sustainable exploitation of natural capital (for full list of inclusion and
exclusion criteria see supplementary materials). Examples of these criteria include: must relate to an
environmental distributive justice issue, must relate to the non-sustainable exploitation of natural
capital, must not be related to nature conservation/sustainable development, must not be related to
procedural justice etc. Article screening was conducted according to our inclusion criteria. We
Page 37
37
carried out the literature screening over four phases: the (aforementioned) coarse (keyword) title
screening, fine title screening (screened against inclusion and exclusion criteria), abstract screening,
and full text screening. At each phase of screening we discarded any articles not relevant according
to the inclusion and exclusion criteria. Throughout the screening process we also discarded any
articles which had missing information (such as title or author), or did not have full text available
online. Each stage reduced the number of total articles (fine scale title screening, n = 1107; abstract
screening, n = 492; full text screening, n = 354).
Typologies and article categorization
We read and categorized the contents of every article passed to the full text screening phase.
Articles were categorized using a list of key variables (n = 91). Variables included bibliographic
information (e.g. author names, and first author institute location), the scope of articles (e.g. article
and data types, geography, and themes), and where mentioned, the equitability of outcomes (see
supplementary materials for complete dataset, including list of variables).
To categorize the thematic scope of each article in the data extraction phase, we developed a series
of thematic typologies. Each typology consisted of a series of variables under four overarching
categories: social demographics, human well-being outcomes, natural resource exploitation issue,
and natural hazards (for complete typologies, see supplementary materials). We developed each
typology by modifying existing typologies (Bournay, 2017; McKinnon et al., 2016; Shortt et al.,
2012; Heyward, 2007; Tornblom, 1977; Cook and Hegtvedt, 1983). Each typology was updated as
appropriate during the title and abstract screening phases. For example, we used the typology
developed by (McKinnon et al., 2016) as a starting point for our social demographics typology.
However, this did not include a variable for LGBTQI+ people, who we found referred to during the
screening phase. Therefore, we added this group as a social demographic variable.
Data analysis and figure generation
All data analysis and all figure generation was conducted using R statistical software (R Core
Development Team, 2015). Linear models were used to assess: the increase in EDJ articles
published over time; and relationships between first author institute location and variable selection
(Burkholder and Edler, 2014). A Pearson’s Chi-squared test was used to assess relationships
between first author institute location and study location (Lancaster, 1969). All other analyses are
proportions.
Page 38
38
Results
The scope of EDJ article types, methods and data
Overall, we found articles relating to EDJ published between 1986-2017, and that the number of
publications per year has significantly increased over time (figure 1a). Our assessment of the EDJ
literature shows several article types (figure 1b), the use of qualitative, quantitative, and mixed data
types (figure 1c), and a diverse range of methods (figure 1d). The majority of published EDJ articles
are empirical research (n = 205, 57.91%), but there are also many essays (n = 70, 19.77%),
literature reviews (n = 54, 15.25%), commentaries (n = 14, 3.95%), and discussion articles (n = 11,
3.11%). When assessing data types, we found that a clear majority of articles used quantitative
methodologies such as desktop analyses of secondary data (n = 181, 51.13%). While we also
identified studies which utilised qualitative (n = 70, 19.77%), and mixed data methods (n = 11,
3.1%) they are less common. We discovered a high diversity of methods used in EDJ articles, but a
few dominated (for table of full methods counts see supplementary material). The most common
methods are desktop studies using secondary data (n = 73, 20.62%), GIS based spatial analysis (n =
67, 18.93%), and document analyses such as literature reviews (n = 48, 13.56%). Philosophical
argument pieces were also common (n = 54, 15.25%), which almost exclusively related to climate
change.
The geographic scope of EDJ articles
We assessed several geographical scales for EDJ literature including: the geographic scale of the
article, the location of the first author’s institute, and for research articles the primary study
location. This study is limited to English language papers only, which may have introduced some
biases.
We found a diverse use of these geographic scales used across articles (figure 2), including local (n
= 87, 24.58%), global (n = 83, 23.45%), national (n = 57, 16.1%), no scale (n = 54, 15.25%), sub-
national (n = 45, 12.7%), regional (n = 17, 4.8%), and multi-city (n = 11, 3.11%) scales. While the
majority of papers applied to the global scale, it is worth nothing that we separately classified
articles that were purely philosophical or theoretical in nature as no scale, as they could be
applicable to any human population.
Our assessment of the geographic location of the first author institutes producing EDJ literature also
revealed a relatively small number of countries (figure 3a). Furthermore, even within the countries
producing the literature, there are only a few producing the majority of the work. The United States
of America (USA), hosts a clear majority of first authors (n = 149, 42.1%), and most others come
from highly developed nations such as the United Kingdom (n = 45, 12.7%), Australia (n = 24,
Page 39
39
6.78%), and Canada (n=20, 5.65%). Even when grouped, all other nations are proportionally
underrepresented (n = 92, 25.99%; for table of all nations see supplementary material). Africa,
Central and South America, and much of Asia are least represented. Furthermore, our assessment of
articles which focused on a specific geographic area of interest shows that here too the majority are
located in a small number of highly developed countries (figure 3b). This is in clear contrast to
geographic distributions shown elsewhere (such as in the EJ Atlas (Temper et al., 2015)).
Figure 1 Total counts of articles, percent of types of articles, percent of data types, and
percent of methods used in environmental distributive justice articles, published from 1986 to
2017. (a) The total papers published per year. There has been a significant increase over time (β=
1.31, p < 0.00, r2 = 0.68), which has accelerated in recent years. 2017 is not shown, as it is an
Page 40
40
incomplete (year) record. (b) The total counts for each type of published article, for all years. (c)
The total counts of articles by data type. (d) The total counts for the methods used in the literature,
for all years - method counts are shown in two ways, the first (yellow) shows counts of non-climate
change articles only, and the second (purple) shows counts of all articles (including those related to
climate change - studies in yellow)).
Note that we have aggregated sub-national areas of interest to the national level (for complete data
set, including sub-national locations see supplementary material). Similarly, to the first author
institute location, we discovered that areas of interest are largely concentrated in only a few
countries. Here too the USA (n = 89, 25.14%), the United Kingdom (n = 18, 5.08%), Australia (n =
11, 3.11%), and Canada (n = 12, 3.39%) are the most represented nations. Areas of interest in all
other countries combined accounted for a relatively small number of articles, even when grouped (n
= 69, 13.45%). Again Africa, Central and South America, and much of Asia are least represented.
Page 41
41
Figure 1 The geographic scale of environmental distributive justice articles from 1986 to 2017.
The geographic scale (x axis) is sorted from most to least common. The percent of total articles (y
axis) is for all articles, for all years. Note that no scale refers to (mostly theoretical) papers with no
specific geography which could apply to any human population.
Page 42
42
Figure 2 World maps showing the geographic spread of environmental distributive justice
articles, from 1986 to 2017. (a) The percent of total publications aggregated at the national level,
by the location of first author institute (n = 224). Most institutes are in a small number of highly
developed nations, with a clear majority located in the USA. (b) Geographic areas of interest as
represented by the percent of total articles aggregated at the national level, by location ((n = 224),
no scale studies (n = 130) not shown). Most non-global articles regard a small number of highly
developed nations, with a majority in the USA. Scales show percent of density of articles with
highest in purple (darkest) to lowest in yellow (lightest).
The thematic scope of EDJ articles
We used theme and variable typologies to organise extracted data for four key topical areas used in
the EDJ literature: social demographics; human well-being outcomes; natural resource exploitation
issues, and; natural hazards.
Page 43
43
Demographics
We found a diverse spread of demographics either discussed in articles, or used as variables in
empirical studies (figure 4a). Economic status (n = 147, 41.52%), geographic location (n = 112,
31.64%), and race or ethnicity (n = 96, 27.19%) were the most commonly used demographic
groups. Whereas, religious (n = 1, 0.28%) and LGBTQ+ (n = 1, 0.28%) groups were comparatively
underrepresented (for complete demographic data see supplementary material). Importantly, we
also recorded several articles under the ‘other’ category as they were inadequately captured by our
social demographics typologies. The majority of these articles used unique demographic groups e.g.
regional grouping of states (Morello-Frosch and Jesdale, 2006), percentage of people employed in
manufacturing (Su et al., 2010), and percentage of single parents (Padilla et al., 2014). The only
relatively common demographic variable recorded under ‘Other’, but not captured in our typology,
was population density.
Human well-being outcomes
For human well-being outcomes, we also found a diversity of topics (figure 4b). However, there
was an overwhelming majority of articles in which health or mortality were discussed or measured
(n = 214, 60.45%). Living standards were also relatively common (n = 78, 22.03%), but equally so
were articles with no human well-being outcomes specified (n = 78, 22.03%). Less commonly
explored outcomes included education (n = 4, 1.13%), and mental health (n = 5, 1.41%). For table
with complete human well-being theme counts see supplementary material. Key examples include
the examinations of toxic substance exposure (Verde et al., 2016; Almaskut et al., 2012), disease
incidence (Szabo et al., 2016), and mental health incidence (Maantay and Maroko, 2015).
Natural resource exploitation issues
We also discovered a range of natural resource issues, but here too a few topics dominated (figure
4c). The majority of articles discussed or measured some form of climate change issue (n = 152,
42.94%) (e.g. Caney (2014)), or air pollution (n = 130, 36.72%) (e.g. Almaskut et al. (2012)), but
few discussed or measured both climate change and air pollution (n = 6, 1.69%) (e.g. Perez and
Egan (2016)). Ozone depletion (n = 2, 0.56%), hunting (n = 2, 0.56%), and hydrological
modification (n = 4, 1.13) are the least commonly explored topics. For table with complete natural
resource exploitation theme counts see supplementary material.
Page 44
44
Natural hazards
As with the other topical areas, our data extraction also revealed a spread of natural hazard topics,
but a majority for only a few topics (figure 4d). Here, the use of air quality as a discussion point or
measurement dominates (n = 143, 40.39%) (e.g. Almaskut et al. (2012)). Air quality aside, it is
common for articles to not specify any environmental hazards (n = 83, 23.45%). Other common
hazards include water quality (n = 59, 16.67%) (e.g. Babidge (2016)), natural disaster intensity (n =
56, 15.82%) and disease (n = 34, 9.6%) (e.g. Austin and McKinney (2016)). Least common hazards
explored are over grazing (n = 1, 0.28%) and extreme fires (n = 2, 0.56%). It is worth nothing that
most natural hazards can either be attributed to or be intensified by climate change (Lewis, 2017).
For table with complete natural hazard theme counts see supplementary material.
Page 45
45
Figure 4 Percent of total articles by topical areas, divided into sub-categories, for all years.
Includes articles where the sub-category was either discussed or measured. Note that articles may
discuss or measure multiple sub-categories. (a) Percent of total articles by social demographic sub-
categories (b) Percent of total articles by environmental exploitation issue sub-categories (c) Percent
of total articles by human well-being sub-categories (d) Percent of total articles by environmental
hazard sub-categories.
Equitability of outcomes
We also assessed and recorded the equitability of human well-being outcomes for each paper
(figure 5). We found that the majority of articles stated that well-being outcomes were inequitable
Page 46
46
(n = 184, 51.98%) (e.g. Sun et al. (2017)), meaning that in these cases the authors found one or
more social groups were unfairly burdened by some form of natural resource exploitation. A large
proportion of articles either did not specify an outcome or did not discuss equitability at all (n =
150, 42.37%). This was the case for all non-empirical studies, where inequities may have been
discussed, but were not measured. Very few articles stated that the outcomes were
unclear/undetermined (n = 12, 3.4%), or equitable (n = 8, 2.26%) (e.g. Rivas et al. (2017)).
Figure 5 Percent of total articles by equitability of outcomes. The outcome(s) (y axis) are sorted
from least to most common. The percent of total articles (x axis) is for all articles, for all years.
Relationships between data
In addition to quantifying the scope of the literature, we also identified several key relationships.
For instance, an assessment of the relationship between first author institute location and study
location revealed a very strong dependence (Pearson’s Chi-squared test with Monte Carlo simulated
Page 47
47
p-value (2000 replicates), χ2 = 3033, p < 0.001). As such, authors are primarily writing about EDJ
issues in countries in which they reside or work, or they focus on global issues. We also found
several statistically significant relationships between first author institute location and the inclusion
of particular social demographics, however the power of most of the tests was too weak due to
small per country sample sizes (for 0.8 power n was approximately > 300 for each demographic
variable, as such only the USA had sufficient samples). As such, the location of an institute
potentially has some effect on which demographic groups are selected during study design, however
we do not have enough data to quantify the strength of this relationship (for complete results of
linear modelling see supplementary materials).
We also identified a relationship between the environmental issue of interest and whether an article
reported on an empirical study (measurement or model of data), or simply discussed the issue (e.g.
philosophical essays, discussion, or comment pieces). In particular, authors of climate change EDJ
literature generate a large amount of discussion articles (n = 106, 29.94%), as opposed to studies
which measure or model data (n = 30, 8.47%). Furthermore, few climate change articles mention
human well-being outcomes (n = 59, 16.67%), and even fewer measure or model a climate change
effect and a human well-being outcome (n = 14, 3.95%; figure 6a). As such, the majority of the
climate change EDJ literature does not involve any empirical measurement or modelling. Similarly,
almost all air pollution articles make some mention of human health outcomes (n = 121, 34.18%).
However, we identified an uneven distribution between air pollution articles which only discuss
health outcomes (n = 83, 23.45%), and those which empirically measured or modelled health data
(n = 38, 10.73%; figure 6b).
Page 48
48
Figure 3 Total counts of key environmental exploitation issues and the disparity between the
measurement and discussion of issues. (a) Total counts of climate change articles divided by
climate change discussion or measurement, and the inclusion (or not) of human well-being
assessments. (b) Total counts of air pollution articles divided by the discussion or measurement of
human health outcomes.
Discussion
Overall, we found that while there is diversity in the EDJ literature, the majority is narrow in scope.
Our results show that the literature has, and continues to grow significantly over time at a rate well
ahead of the general increase in academic literature, which highlights the increasing importance of
this field globally. Methodologically, we found a diversity of research types, though a few method
categories dominate the literature. In addition, we found that for climate change literature
specifically, there are a much larger number of papers putting forth ethical frameworks or
arguments (e.g. how emissions ought to be calculated), than empirical studies. Furthermore, we
found that the majority of EDJ research applies quantitative methods, such as desktop analyses of
secondary data, and the use of spatial methods such as GIS. There are far fewer studies draw on
qualitative methods, such as interviews and focus groups. Even fewer are mixed methods studies.
While mixed methods research is more time and resource expensive to conduct, it has been
described as the most effective means of producing validated, complex, and rich data (Borkan,
2004; Creswell, 2014; Clark and Ivankova, 2015). As such, we argue that increased application of
mixed methods is a key opportunity area for developing robust EDJ studies.
Page 49
49
Our results show a large amount of the literature that examines issues at either the global or local
geographic scales. While examining issues at these scales is important for addressing EDJ issues,
we argue that there is also much to be done at other scales, such as at the national and or sub-
national levels. Examining EDJ issues at these scales is becoming increasingly important as
governments and scientists alike attempt more complex and nuanced approaches to solving issues,
such as ‘bottom-up’ action for mitigating greenhouse gas emissions, or reducing air pollution (Sabel
and Victor, 2017). However, the most critical issue we found in the geographic scope in the EDJ
literature, both in terms of study location and author location, is the narrow focus on a few, highly
developed nations. Interestingly, this highly contrasts with other resources showing much broader
geographic distributions of EJ issues (Temper et al., 2015). Combined with the close association
between first author institute location and area of interest, we posit that this represents a significant
geographic bias in (English language) academic literature. There is much evidence to suggest that
nations which are relatively least developed are far more likely to be socially inequitable, and to
have poor environmental exploitation standards (Adger, 2002), which in turn can lead to human
health inequalities (Drabo, 2011).
While there are certainly good reasons for a lack of academic EDJ research in developing countries
such as poor data availability, given the significant inequities observed between nations, the need
for EDJ research in developing countries is vital. It will therefore be a challenge for the field to
ensure that researchers can undertake projects in countries other than that of their research
institution, guided by attentiveness to the greatest need for EDJ research. In such situations, it will
be critical that Western researchers focus on building the capacity of researchers in those countries
thereby developing more diversity in EDJ authorship, and contributing to the broader movement of
decolonialising academia. It is worth highlighting that issues around study location and author
diversity are by no means unique to this field. Several papers have been published highlighting
similar issues across other disciplines (e.g. development studies (Cummings and Hoebink, 2017),
and Africa-based studies (Briggs and Weathers, 2016)).
While we found a diverse spread of social demographic examination across the literature, the
majority of articles focus on the economic differences between groups. While economic statistics
(such as GDP) can be used as a coarse analogue for human well-being, they offer only limited
insight into important dimensions such as capabilities (Sen, 2009). As such, it is recommended that
economic status is used with caution and only where data allowing more complex metrics, such as
the human development index (UNDP, 2017), are absent. Furthermore, we found a lack of articles
discussing minority religious groups, and people who identify as LGBTQ+, both of which are
recognized as potentially vulnerable populations (Fineman, 2014; Grim and Finke, 2007).
Page 50
50
Interestingly, we also found that primary author institute location and certain demographics
discussed in an article are potentially related, which is potentially due to the importance of cultural
context. For example, European studies tended to focus on economic status race very commonly.
However, the reasoning for demographic selection is often not clear or explicitly stated in the study
design. We suggest that (where appropriate to study design) authors utilise a more universal set of
social demographics, such as the PROGRESS framework used in the synthesis of health inequity
studies (O’Neill et al., 2014). From this universal framework authors can then tailor their study to
the appropriate cultural context.
Given the nature of EDJ, it is unsurprising that links to human well-being outcomes feature heavily
in the literature. While we did find a range of outcomes, a few areas dominated. Interestingly, while
economic status is the most commonly used social demographic in articles, livelihoods are not the
most commonly examined human well-being outcome (see results, figure 4). Rather, the majority of
articles examine health and mortality outcomes. While human health is of obvious importance,
many other well-being issues can result from environmental exploitation. Furthermore, non-health
outcomes can affect large groups of people, and themselves lead to health issues. For example,
climate change enhanced droughts and floods can destroy livelihoods, and increase disease
incidence (Gichere et al., 2013). As such, a more diverse exploration of human well-being outcomes
is a good candidate area for future EDJ research.
We found a clear tendency for EDJ articles to relate to either air pollution, or climate change which
reflects the large-scale nature of these issues and their potential for devastating impacts on human
well-being and livelihoods. However, we discovered an imbalance between the discussion, and
some form of measurement (or modelling) across these topics. This is particularly true for the
climate change literature, where only a very small proportion of papers measured some climate
change effect and a human well-being outcome. While the discussion of how we ought to decide
what might be a just distribution of climate change costs and benefits is of obvious importance,
further benefit to the field can be gained through conducting empirical studies which develop the
evidence-base for the EDJ implications of climate change. Similarly, but not to the same degree, for
the literature related to air pollution, we found many articles which discuss health outcomes from
air pollution exposure, but far fewer which attempted to measure them or record incidence.
Furthermore, health outcomes were often only discussed, rather than measured across much of the
literature. For example, statements which indicate air borne pollutants have been linked to
respiratory illness, with no accompanying respiratory illness data are common. This illustrates an
opportunity for the EDJ literature, as to maximize the applicability of EDJ research (e.g. to
influence future policy) it is not only important to continue to identify injustices in distribution of
Page 51
51
environmental issues. Rather, it is equally (if not more), important to also link such injustices to
human well-being outcomes. While causative links are difficult to establish, this could potentially
be addressed by an increase in more complex, mixed methods studies. For example, spatial
quantitative methods could be used to identify correlates, controls, and counterfactual areas across a
large space for use in the experimental design of follow on, fine scale studies.
We found that almost all papers (90.2%) which assessed the equitability of human well-being
outcomes concluded that there are distributive inequities. However, the reason for this finding is
unclear. It may be that almost all natural resource exploitation is profoundly inequitable. However,
it may alternatively represent a form of confirmation bias in the EDJ literature, where authors may
be selecting cases where they have prior knowledge of an inequity. Confirmation bias is an issue of
increasing relevance to most scientific disciplines (MacCoun and Perlmutter, 2015). As the reason
for such a high statistic is unclear, we highly recommend that future EDJ studies take precautions to
prevent potential biases where possible, and utilise methods such as counterfactual thinking and
randomized trials, quasi-experimental designs, blind analysis, and the use of control variables
(MacCoun and Perlmutter, 2015; Miguel et al., 2014; Ferraro, 2009). In either case, it was not in the
scope of this study to assess the quality of methodology within the literature, so our 90.2% inequity
statistic ought to be considered with caution. To clarify this finding, we recommend that future
reviews of the EDJ literature attempt to incorporate study quality assessment.
Study Limitations
While our overall aim with this study was to be as exhaustive in our investigation of the EDJ
literature as possible, available resources constrained us. As such, there are several limitations to
our study which need to be highlighted for disclosure. Most importantly was our lack of ability to
conduct a full, manual screening of the literature at the article title level. We did not have the time
and resources to manually screen over 40,000 articles. Relatedly, as the academic databases alone
yielded over 40,000 articles, we did not have the resources available to screen all potential sources
of literature, including very literature diverse databases such as Google Scholar, the grey literature,
and book chapters. Consequently, there is the potential that we missed some EDJ articles at each
subsequent screening level, and our results largely capture the academic peer-reviewed literature,
which may have biased our sample and results. However, given the relatively large number of full
text screened articles we were able to screen, our study can be thought of as a non-exhaustive, yet
representative sample of the EDJ literature. Furthermore, our study was limited to English language,
peer reviewed literature which no doubt introduced some biases. These potential biases may affect
the findings around geographic bias in the EDJ literature, e.g. if non-English language or non-peer
Page 52
52
reviewed studies are available in greater numbers in developing countries. It is recommended,
where possible, future reviews on this subject area address these limitations.
Conclusion
This study makes a meaningful contribution to the broader literature reviewing EDJ. Consistent
with previous reviews, we found overall, that there is rich and somewhat diverse EDJ literature
(Banzhaf et al., 2019; Brulle and Pellow, 2006). However, there are key trends apparent across the
majority of articles. This is particularly true for the geographic scope of the literature, which is
clearly misaligned to the real world scope of EDJ problems. While a few most developed nations
are very well represented, many of the world’s least developed and/or least socially equitable
societies are not. This is not to state that work which examines the very real and often harmful
environmental inequities found in developed nations. Rather, we argue this gap highlights a key
area for expanded focus by EDJ researchers, which would both advance the knowledge on EDJ and
contribute to achieving more equity in the investment of research efforts internationally.
Consequently, we encourage EDJ researchers from the world’s most developed nations to look
further afield when developing future projects, via channels such as capacity building, funding and
to develop supportive relationships with researchers in those countries of interest that are most in
need of EDJ research. In addition, there is an opportunity to direct efforts to linking EDJ
observations with human well-being outcomes in order to develop the evidence base for decision-
making, such as environmental and planning policies, informed by EDJ literature. More generally,
we strongly encourage EDJ researchers to use our results when developing future research projects
to contribute to a diverse, robust and insightful literature.
Page 53
53
Chapter 3 Global mismatch between greenhouse gas emissions and the
burden of climate change
Online publication: http://dx.doi.org/10.1038/srep20281
Full citation: Althor G, Watson JEM and Fuller RA. (2016) Global mismatch between greenhouse
gas emissions and the burden of climate change. Scientific Reports 6.
Overview
This study was published in Nature Scientific Reports in 2016. It is an innovative and novel
approach to examining the fairness dimensions of climate change. I produced the first global level
study to use data for all greenhouse gas emissions (GHGs), against the vulnerability of nations to
climate change. I argue that the economic benefits some few countries are gaining from emitting
GHGs are far outweighed by the burdens they are placing on other lower emitting nations. I use
quantitative Gini and Robin Hood coefficients to present this idea.
Research question: How are greenhouse gas emissions and the burdens of climate change
distributed globally?
Sub-questions:
1) What is the extent of global climate inequity; and
2) How is this expected to change in the near future.
This study utilises my combined principles which regard cases where some harm (or other type of
burden) is produced, for the specific benefit of a single person, or exclusive group, as follows:
P3: Where burdens upon any individual or exclusive group are the byproduct of the use of natural
capital for the betterment of another individual or exclusive group, burdens are to be accompanied
by (at least) commensurate benefits.
Highlights:
• First global study to use data for all greenhouse gas emissions to assess distributive justice
of climate change.
• Prime example of how natural capital distributive justice principles can be used to design
and evaluate empirical research.
Page 54
54
• Uses quantitative measures of inequality to support claims, and provides geographical maps
making findings easily understandable.
Page 55
55
The text appears exactly as published, with the exception of change from Nature referencing
style to Sage Harvard referencing style and placement of Methods section before Results
section.
Abstract
Countries export much of the harm created by their greenhouse gas (GHG) emissions because the
Earth’s atmosphere intermixes globally. Yet, the extent to which this leads to inequity between
GHG emitters and those impacted by the resulting climate change depends on the distribution of
climate vulnerability. Here, we determine empirically the relationship between countries’ GHG
emissions and their vulnerability to negative effects of climate change. In line with the results of
other studies, we find an enormous global inequality where 20 of the 36 highest emitting countries
are among the least vulnerable to negative impacts of future climate change. Conversely, 11 of the
17 countries with low or moderate GHG emissions, are acutely vulnerable to negative impacts of
climate change. In 2010, only 28 (16%) countries had an equitable balance between emissions and
vulnerability. Moreover, future emissions scenarios show that this inequality will significantly
worsen by 2030. Many countries are manifestly free riders causing others to bear a climate change
burden, which acts as a disincentive for them to mitigate their emissions. It is time that this
persistent and worsening climate inequity is resolved, and for the largest emitting countries to act on
their commitment of common but differentiated responsibilities.
Introduction
The current generation is the first to feel the effects of anthropogenic climate change (IPCC, 2014a;
Ipcc, 2014b). Despite their well-known harmful impacts to the world’s climate system (IPCC,
2014a; United Nations Framework Convention on Climate Change, 1992), greenhouse gases
(GHG) are deliberately emitted by countries to drive economic growth and enhance human
wellbeing (Jorgenson, 2014). Spatially localised environmental issues, such as city air pollution
(Sheehan et al., 2014), may result from high GHG emissions, but the most damaging and long
lasting consequence, that of global climate change (Montzka et al., 2011), is not constrained within
the border of the emitting country (IPCC, 2014a). Rather, by polluting the Earth’s atmosphere with
GHG emissions through fossil fuel combustion, deforestation and agricultural activities, emitting
countries are degrading the world’s climate system, a common resource shared by all biodiversity,
including people (Betts, 2008).
Page 56
56
Because the impacts of GHG emissions can be felt beyond a country’s border, and the impacts of
climate change on countries are highly variable, there is potential for some emitters to contribute
more or less to the causes of climate change than is proportionate to their vulnerability to its effects
(Cazorla and Toman, 2001; Kjellstrom et al., 2009; Trenberth, 2011). This inequity has not gone
unnoticed in international climate negotiations or global reporting (IPCC, 2014a; United Nations
Framework Convention on Climate Change, 1992). As far back as 1992, the United Nations
Framework Convention on Climate Change (UNFCCC) committed to the principle of “common but
differentiated responsibilities”, in which countries have a common responsibility in reducing GHG
emissions, but historic emissions and differences in current development levels mean that countries
have different levels of emissions reduction obligations (Cazorla and Toman, 2001). Both of the
previous IPCC Assessment Reports have acknowledged the inequity in the causes and effects of
climate change (IPCC, 2014a; IPCC, 2007) although operationalising the principle has proved
difficult (Ostrom, 2014). This is primarily because developing and developed countries continue to
disagree over the extent of each other’s responsibilities (Ostrom, 2014; Cole, 2015). One major
impediment to resolving such debates is a poor quantitative understanding of the magnitude of the
global inequity in emissions and impacts. ‘Free rider’ countries contribute disproportionately to
global GHG emissions with only limited vulnerability to the effects of the resulting climate change,
while ‘forced rider’ countries are most vulnerable to climate change but have contributed little to its
genesis (Rao, 2014; Füssel, 2010). This is an issue of environmental equity on a truly global scale
(World Resources Institute, 2014).
Here, we measure the current pattern of global climate change equity, and assess whether the
situation will improve or worsen by 2030, using data on GHG emissions (World Resources
Institute, 2014) and newly available national climate change vulnerability assessments (DARA,
2012). We address the lack of a contemporary, qualitative assessment of global climate equity that
incorporates key variables. Previous studies have been limited to CO2 emissions datasets, omitting
the most potent and long lasting GHGs (IPCC, 2014a; Montzka et al., 2011; Füssel, 2010), and
used vulnerability variables that do not capture the complexity of climate change threats, and cannot
be forecasted. Here, we use the most recently available datasets based on comprehensive national
vulnerability assessments and comprehensive GHG emissions data to produce an easily replicable
snapshot of the relationship between countries’ GHG emissions and their vulnerability to the
negative effects of climate change (World Resources Institute, 2014; DARA, 2012), and forecast
this to 2030. We employ economic metrics, the Gini and Robin Hood coefficients (Coulter, 1989b),
to quantify the present level of equity in GHG emissions. Only through a proper empirical
Page 57
57
understanding of the pattern of climate equity now, and how it will change in the near future, can
signatories of the UNFCCC make meaningful progress toward resolving the inequity in the burden
of climate change impacts.
Methods
We quantified climate change equity, defined as the distribution of climate change benefits and
burdens, using data from two publicly available datasets and national GDP data. National level data
sets suffer from some weaknesses such as a lack of accounting for sub-national variability and
scaling. Nonetheless, they are still highly useful as global metrics as they provide aggregated
assessments at the national level, which is the most meaningful for international policy negotiations.
We extracted data on national vulnerability to the negative impacts of climate change from DARA’s
Climate Vulnerability Monitor (CVM) (DARA, 2012). The CVM uses 22 climate vulnerability
indicators across four impact areas (Environmental Disasters, Habitat Change, Health Impact, and
Industry Stress) to evaluate the vulnerability of 184 countries to climate change impacts for the
years 2010 and 2030. Each of the 22 indicators is individually aggregated from various data sources
and models and then combined to determine a country’s overall climate vulnerability, measured by
impact to share of GDP and mortality (as these impacts are comparable across the wide range of
countries). The CVM calculates vulnerability projections for 2030 using human population growth,
mortality and GDP predictions. The CVM uses five vulnerability categories (low, medium, high,
severe and acute) which are determined using a mean absolute standard deviation method18. The
CVM categories do not of course capture the full complexity of national climate vulnerability, as
capturing this would require an impractical degree of data. However, we consider the 22 indicators
used by the CVM as capturing a high enough level of complexity to provide a meaningful
approximation of national vulnerability.
Data on GHG emissions (by countries) were exported from the World Resource Institute’s (WRI)
Climate Analysis Indicators Tool (CAIT) (World Resources Institute, 2014), a database of national
and international GHG emissions derived from multiple sources. The CAIT data set compiles data
for the six main GHG gasses (carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O),
hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and sulfur hexafluoride (SF6)) from 185
countries over the period from 1990-2012. We used the 2010 data for this study to match the CVM
vulnerability data. The WRI compiled GHG data from UNFCCC reports and complemented with
data from several NGO sources (World Resources Institute, 2014), including emissions data from
Page 58
58
six major sectors (Land Use Change & Forestry, Energy, Industrial Processes, Agriculture, Waste,
and International Bunkers) and several subsectors. The CAIT data set reports at the national level,
however we extrapolated per capita emissions results by dividing data by 2010 and 2030 population
data from the World Bank (World Bank Group, 2015) (see Supplementary Table S4 online).
We excluded the ten countries (Cook Islands, Federated States of Micronesia, Marshall Islands,
Montenegro, Nauru, Niue, Saint Kitts & Nevis, Serbia, Somalia and Taiwan) with data missing in
any dataset, and 179 remained for analysis. In addition, there were also insufficient data available
for many of the world’s island and archipelagic countries. Given the negligible GHG emissions and
high climate change vulnerability of such countries, the majority are highly likely to qualify as
climate forced riders (Nurse et al., 2014; Widlansky et al., 2013) and as such, we expect that climate
forced rider countries are likely underrepresented in our results. National GDP (measured in Current
US$) was extracted from the World Bank Group (World Bank Group, 2015), who measure GDP as
the gross value of all resident producers in an economy plus taxes.
We created a Lorenz curve to represent the variation of GHG emissions among countries using the
CAIT dataset, and calculated the Gini index to measure inequity in GHG emissions among
countries, and the Robin hood index to measure how much of the total global emissions would have
to be redistributed to achieve equity among countries (see Supplementary Fig. S2 online).
We compared the CAIT GHG data and the CVM vulnerability data both in 2010 and 2030 to assess
whether the most heavily polluting countries were also those least vulnerable to the negative effects
of climate change. We divided the CAIT GHG emissions into quintiles, matching the CVM data, to
enhance comparability between the datasets and enable visualisation of climate equity in the recent
past (2010) and near future (2030). We placed the emissions quintiles on a scale between the
highest (acute emissions) and the lowest (low emissions) emitting countries. We also tested the
correlations between GHG emissions and GDP against vulnerability to climate change by treating
vulnerability categories as ordinal data and undertaking spearman’s rho tests using R statistical
software (R Core Development Team, 2015). R has a computational limitation for p-values lower
than 2.2e-16, as such, where values this small were reported we wrote “p = 0”. Additionally, we
counted countries in each CVM category and compared them between each time period.
In common with other studies of inequity in climate change (Otto et al., 2015), we used terminology
from the economics literature to define ‘free riders’ and ‘forced riders’ (Ahnlid, 1992), recognising
Page 59
59
that a strict definition of these terms often applies only to situations where one agent’s use of a
resource does not directly incur a cost to another agent. We define climate free riders as those
countries in the ‘acute’ GHG emissions quintile and the ‘low’ vulnerability category, as they
disproportionately receive benefits from climate change (via the national wellbeing generated by
GHG emissions) but pay few costs in the sense they are the least vulnerable to negative climate
change effects (Keohane, 2015; Kennedy and Basu, 2014). Conversely, we define climate forced
riders as those countries that fall within the ‘acute’ vulnerability category and the ‘low’ GHG
emissions quintile, as they are the most susceptible to the negative consequences of climate change
but receive the least benefits. Those countries that we define as equitable, fall in the same emissions
quintile and vulnerability category (for example, low emissions quintile, low vulnerability
category), as their emissions benefits are concomitant with their climate change burden.
Results
Greenhouse gas emissions are spread highly unevenly across the world’s countries (Fig. 1), with the
top ten GHG emitting countries generating > 60% of total emissions, and three countries, China
(21.1%), the United States of America (14.1%) and India (5.2%) being by far the largest
contributors. A Gini coefficient of 80.9 indicated extreme inequality in the distribution of emissions
among countries, given that the index can only vary between 0 (perfectly even responsibility) and
100 (one country responsible for all emissions) (Coulter, 1989b). A Robin Hood index of 64
indicated that 64% of GHG emissions would need to be redistributed to achieve an even distribution
among countries (Coulter, 1989b). Vulnerability to the impacts of climate change was also
unevenly spread among countries, with 17 countries acutely vulnerable to climate change impacts in
2010 (Fig. 2). The majority of these were island countries located in the Atlantic, Pacific and Indian
oceans (n=7, 35.3%) and African countries (n=8, 47%). By 2030 the number of acutely vulnerable
countries is predicted to rise dramatically (n=62; Fig. 2), and the majority of these will again be
island (n=20, 32.8%) and African (n=27, 44.2%) countries.
Page 60
60
Figure 4. Global inequity in the responsibility for climate change and the burden of its
impacts. (a) Climate change equity for 2010. (b) Climate change equity for 2030. Countries with
emissions in the highest quintile and vulnerability in the lowest quintile are shown in dark red (the
climate free riders), and those countries with emissions in the lowest quintile and vulnerability in
the highest quintile are shown in dark green (the climate forced riders). Intermediate levels of equity
are shown in graduating colours, with countries in yellow producing GHG emissions concomitant
with their vulnerability to the resulting climate change. Data deficient countries are shown as grey.
Maps generated using ESRI ArcGIS (ESRI ArcGIS, 2011).
Page 61
61
Figure 5. Vulnerability to climate change, mean GHG emissions, and mean GDP. (a) Number
of countries in each climate change vulnerability category, derived from DARA vulnerability data
(DARA, 2012), for 2010 (blue bars) and 2030 (green bars). (b) Mean GHG emissions for 2010,
derived from CAIT GHG emissions data (World Resources Institute, 2014), shown in CO2
equivalent units and climate vulnerability categories for 2010 (blue bars, with standard error) and
2030 (green bars, with standard error). (c) GDP shown in current US$ (in billions), derived from the
World Bank GDP 2010 data (World Bank Group, 2015), and 2010 GHG emissions. (d) Mean GDP
for 2010 shown in current US$ (in billions) and climate change vulnerability for 2010 (blue bars,
with standard error) and 2030 (green bars, with standard error).
Countries least vulnerable to the impacts of climate change were generally the highest GHG
emitters, and conversely those most vulnerable to climate change were the least responsible for its
genesis. This inequity held true for both 2010 and 2030, with a negative relationship between
emissions and climate vulnerability in both years (2010: ρ = -0.4, n = 175, p = 0; 2030: ρ = -0.37, n
= 175, p = 0). The only exception is in 2030, where countries acutely vulnerable to climate change
will have slightly higher average emissions than those in the severe category (2030: severe = 48.83
mtCO2e, acute = 103.13 mtCO2e).
Page 62
62
In 2010, of the 179 countries assessed, 28 (15.6%) were in the same quintile for GHG emissions
and vulnerability to the negative impacts of climate change. This indicates that their vulnerability to
climate change approximately matched their relative contribution to its genesis (Fig. 1). Ninety
countries (50.3%) had GHG emissions in a higher quintile than their 2010 climate vulnerability, and
20 (11.2%) countries were free riders, with GHG emissions in the highest quintile and climate
vulnerability in the lowest quintile (Fig. 1; see Supplementary Table S4 online). Sixty-one (34%)
countries had GHG emissions in a lower quintile than their climate vulnerability, and six (3.4%)
countries were forced riders, with GHG emissions in the lowest quintile and climate vulnerability in
the highest quintile (Comoros, Gambia, Guinea-Bissau, São Tomé and Príncipe, Solomon Islands
and Vanuatu; see Supplementary Table S4 online).
By 2030, climate change inequity will rise further, with an increase in the proportion of countries
that are forced riders (n=20; 11.2%), but fewer free riders (n=16; 8.9%) and equitable countries
(n=23; 12.8%; see Supplementary Table S4 online). Free riders are typically located in the world’s
sub-tropical and temperate regions, while forced riders are frequently located in tropical regions
(Fig. 1).
Greenhouse gas emissions were positively correlated with GDP (2010: ρ = 0.84, n=175, p = 0; Fig
2c), while climate vulnerability declined with increasing GDP (2010: ρ = -0.69, n = 175, p = 0;
2030: ρ = -0.65, n = 175, p = 0; Fig. 2d). Our analysis considers the absolute contribution of each
country to climate change, but we also examined climate change equity in per capita terms to
provide a more complete picture of emissions responsibilities. The patterns were broadly similar,
with, for example, Australia, Russia and the United States of America remaining free riders (see
Supplementary Fig. S3 online). However, several populous major emitters (e.g. United Kingdom,
China, and Brazil) were no longer categorised as free riders.
Discussion
Climate change inequity is globally pervasive, and correlated with economic output. Some
countries, such as China and the United States of America, are in a win-win position of achieving
economic growth through fossil fuel use with few consequences from the resulting climate change,
while many other, mostly Island and African, countries suffer low economic growth and severe,
negative climate change impacts (see Supplementary Table S4 online). The beneficiaries of this
climate inequity have few incentives to meaningfully reduce or halt their GHG emissions. Despite
many of the broad issues around climate equity being well known (IPCC, 2014a), well-funded
Page 63
63
global mechanisms that are being implemented still do not exist. This has serious consequences for
our ability to slow the rate of climate change, and reduce the wellbeing implications for forced rider
countries.
There are several global policy frameworks currently being debated that could address elements of
the problem. The Paris Agreement (UNFCCC, 2015a), secured at the 21st UNFCCC Conference of
the Parties (COP21), for example, sets an ambitious target of limiting global warming to 1.5C
above preindustrial levels. However, the 160 indicative nationally determined contributions
(INDCs) pledges submitted by signatories to the UNFCCC prior to COP21 (UNFCCC, 2015b),
indicate that current targets for GHG emissions are unlikely to limit warming to below 2C
(Jackson et al., 2015) With no binding agreement established at COP21 for INDCs, there is no clear
indication of how successful the Paris Agreement will be (UNFCCC, 2015a). Addressing GHG
emissions is clearly an important first step in ensuring the burden of climate change is not amplified
in the future. However, the historic commitment to GHG emissions reduction by key free riders has
been slow. Only 50 countries ratified the previous Doha Amendment to the Kyoto protocol, which
did not include key free riders such as the United States and Russia (Fekete et al., 2013).
Furthermore, some countries have actually backtracked on their commitments to emissions
reductions (e.g. Canada and Australia) (Pizer and Yates, 2015; Harrison, 2012).
Likewise, the Paris Agreement calls for urgent and adequate financing of US$100 billion per year
by 2020 for climate mitigation and adaptation through the Financial Mechanism of the Convention
(FMC) (UNFCCC, 2015a). However, there is no legally binding mechanism under which parties are
responsible for providing this funding. History suggests such funding goals are not always met. For
example, the Green Climate Fund (GCF) was established in 2010 under the UNFCCC to mobilise
funding support for the least developed countries that are most vulnerable to climate change, yet it
remains poorly funded, with only US$10.2 billion received in pledges by November 2015 (Green
Climate Fund, 2015). Addressing these issues around climate funding will play a critical role in
addressing climate inequity (Pickering et al., 2015).
Conclusion
It is clear climate change inequity must be addressed. If the commitment to the principle of
common but differentiated responsibilities that was widely accepted early on in the UNFCCC is to
be acted upon, member states now need to do much more to hold climate free riders to account. To
ensure equitable outcomes from climate negotiations, there needs to be a meaningful mobilization
Page 64
64
of policies, such as the Paris Agreement, that achieve national level emissions reductions, and to
ensure the vulnerable forced-rider countries are able to adapt rapidly to climate change. The
provisioning of these policy mechanisms will require a distribution of resources and responsibilities
and we believe our results provide one way to understand where these responsibilities lie. The Paris
Agreement may be a significant step forward in global climate negotiations. However, as the
Agreement’s key policies are yet to be realized, member states have both an exceptional
opportunity and a moral impetus to use these results to address climate change equity in a
meaningful manner.
Author contributions
G.A., J.E.M.W. and R.A.F. designed the analysis. G.A. performed the analysis and analysed the
results. G.A., J.E.M.W. and R.A.F. wrote the paper.
Competing Financial Interests statement
The authors declare no competing financial interests.
Page 65
65
Chapter 4 The distribution of industrial pollution across socio-
economic groups in Australia; a ten-year, a fine scale distributive
justice analysis
Full citation: Althor G, Witt B and Colvin R. (forthcoming) The distribution of industrial pollution
across socio-economic groups in Australia; a ten-year, a fine scale distributive justice analysis.
Target journal: Environmental pollution.
Overview
This study is forthcoming, but as presented, is appropriate for submission to a peer-reviewed
journal. With this study I build on existing work which assessed the distribution of harmful
industrial pollution across Australian socio-economic groups. It is the finest scale study of its kind
undertaken in Australia, and the first to add a temporal dimension. Furthermore, I address the
limitations of previous work in weighting emissions by toxicity, which is critical for understanding
the potential impacts on people’s well-being.
Research question: Are there social inequities in the distribution of exposure to harmful industrial
pollution across Australia?
Sub-questions:
1) Have these distributions changed over the last 10 years; and
2) If any, which social or economic groups are more likely to be exposed to industrial pollution.
This study specifically utilises my principle of strict equality, as follows:
P2: Where harm is accepted as a tolerable by-product of natural capital use for the betterment of
society, exposure levels of harm ought to be equal among all persons.
Highlights:
• Highest resolution study of its kind undertaken for Australia.
• First study of its kind to introduce a temporal, as well as geographic element.
• Provides the most robust method to date for weighting emissions by toxicity, specifically for
use with data provided by the Australian Government.
Page 66
66
The following text appears exactly as manuscript prepared for peer-review submission.
Abstract
Environmental distributive justice assesses the cost and benefit trade-offs among affected social and
economic groups in situations of natural resource use. Industrial pollution is a wide spread
phenomenon, and the largely unavoidable byproduct of many economic activities. Industrial
pollution can have negative effects on human health when people are exposed to emissions. As such
understanding the distribution of, and human exposure to, industrial pollutants has been
increasingly studied across the world, and within Australia. In particular there is a focus on
understanding if specific social groups (such as racial or economic groups) are more or less likely to
be situated near industrial pollution sources, and as such more or less likely to suffer negative health
effects. However, an up to date, comprehensive and nationally representative study assessing the
distribution of industrial pollutants and nearby social groups has not been undertaken in Australia.
Here we present a study using data gathered at three intervals, over 10-years for all of Australia
where we assessed emissions for all 93 industrial pollution substances that are currently monitored
in Australia and weighted each substance according to its potential to cause harm to people. We
show that there are no significant correlations between industrial pollutant sources and social
groups within Australia. This result is in contrast to previous Australian studies which have shown
distributive injustices between industrial pollution and social groups.
Introduction
In Australia industrial pollution occurs as a side effect of a highly diverse range of economic
activities such as power stations, mining and chemical manufacturing (Keywood et al., 2016).
These activities broadly improve human well-being in Australia through their contribution to Gross
Domestic Product (GDP). This is best illustrated by the approximately $1.69 Trillion (AUD) value
of Australian industry (ABS, 2017a). However, via pollution emissions into the environment,
industry can also have serious negative well-being effects on people, such as causing disease and
premature death (Moore and Hotchkiss, 2016; Kampa and Castanas, 2008). For example, a recent
global commission on pollution and health states that pollution caused an estimated nine million
mortality related diseases in 2015, which accounted for some 16 percent of global deaths
(Landrigan et al., 2017). As such, the costs and benefits of industry can be described as a complex
trade-off between improving well-being via economic growth and degrading it via toxic emissions.
Furthermore, extant evidence indicates these trade-offs are rarely evenly distributed among social
and economic groups within populations.
Page 67
67
In many of the world’s nations, industrial pollution disproportionately causes death and disease
among least advantaged groups, such as low-income countries globally, or among vulnerable socio-
economic groups at various sub-national scales (Padilla et al., 2014; Charafeddine and Boden, 2008;
Germani et al., 2014). These fairness implications of such distributions make this an issue of
environmental distributive justice (EDJ). Distributive justice can broadly be described as principles
which argue that the distribution of social benefits and burdens ought to be guided by moral
principles of fairness (Allingham, 2014). EDJ applies distributive justice principles to
environmental issues. When choosing a pathway to increased GDP through the utilization of, and
impact on, the natural environment, a society (or its decision-makers) should seek to ensure an
equitable burden of any hazards or impacts on human wellbeing. This argument holds true at any
scale from the fine (i.e. between individuals in a village) up to the global (i.e. between nations). As
such, EDJ principles can be applied to industrial pollution in order to: 1) understand the resultant
trade-offs between positive and negative well-being outcomes; and 2) identifying any unjust, socio-
economic distributive patterns of pollution.
Since the 1980s, increasing public awareness of human and environmental impacts has resulted in a
distinct increase in scholarship and activism around EDJ and industrial pollution, particularly in the
United States and Europe (Padilla et al., 2014; Holnicki et al., 2017; Charafeddine and Boden,
2008). However, in countries such as Australia there is much less research dedicated to the
investigation of EDJ and industrial pollution (Chakraborty and Green, 2014). Pollution (industrial
and otherwise) is a serious issue in Australia with around 3,000 deaths per year attributed to urban
pollution alone (Keywood et al., 2017). Furthermore, it has been estimated that the health costs of
pollution-related premature deaths are approximately $2.5 billion AUD annually (Broome et al.,
2015). While the overall disease and mortality burdens related to pollution in Australia are
relatively well understood, far less is known about the distribution of this pollution, and if it is
correlated with socio-economic factors such as ethnicity or income.
We found only three studies based in Australia that have assessed the EDJ implications of pollution
nationally (Chakraborty and Green, 2014; Knibbs and Barnett, 2015; Dobbie and Green, 2015).
Chakraborty & Green (2014) aggregated a sub-set of air pollution data from the Australian
Government’s National Pollution Inventory (NPI), and four aggregates of demographic data from
several data sets from the Australian Bureau of Statistics (ABS). The authors state that their study
shows socio-economically disadvantaged populations are more likely to be distributed near
concentrated sources of toxic air pollutants. However, in this study the emissions data were
Page 68
68
“toxicity-weighted volume of air pollution”, based on an out of date toxicity rating which was not
developed for use in this manner (NEPC 1999, appendix III) (Chakraborty and Green, 2014: pg. 2).
Furthermore, details of the actual weighting method used are unclear making replication of the
study difficult. As such, the validity of weighting the data in such a way may be problematic. In
addition, the study used the relatively large Australian standard Statistical Area Level 2 geographic
units. Recent research has shown that spatial analysis which utilises such large units can produce
unreliable results (Hanigan et al., 2017), such as aggregation errors and other measurement errors
(Anderton et al., 1994).
Knibbs & Barnett (2015), also assessed the distribution of exposure to air pollutant NO2 by social
stratification, including variables for socio-economic advantage and disadvantage, and percent of
people identifying as Aboriginal or Torres Strait Islander. This robust study utilises high quality
complex pollution and socio-economic data at the smallest available spatial scale. However, the
authors only assessed exposure to a single pollutant (nitrogen dioxide (NO2)) and did not assess the
distribution of sources of the pollutant (i.e. whether or not it was generated by industry). Their
results show evidence of some distributive inequalities in NO2 exposure across Australia, with
people identifying as Aboriginal or Torres Strait Islander more likely to be exposed to NO2.
Dobbie & Green (2015), focused on the regulation of air pollution in Australia and the geographic
distribution of breaches of pollution licence conditions. In particular, they assessed two case studies
to illustrate the nature of these distributions. Dobbie & Green (2015) suggest that results of their
study indicate disproportionate human health impacts present in the case study communities, and
argue that this is largely the result of failures in existing regulatory systems.
These previous studies suggest that there are pollution related cases of distributive injustice in
Australia, in particular they suggest least economically advantaged groups and Indigenous
Australians are more likely to be exposed to pollution. However, in an earlier meta-analysis of EDJ
studies (mostly from the USA), Bowen (2002) found, contrary to the dominant opinion, that there
was little statistically significant evidence to suggest patterns of distributive injustices in the
location of pollutant emitting sites. Given Bowen’s analysis and the limitations with the existing
Australian literature, we present here a quantitative spatial analysis of industrial emissions in
relation to EDJ principles using Australian nationwide, fine-scale, and the most recently available
data. As such, this study aims to understand how industrial emissions are distributed among social
and economic groups in Australia.
Page 69
69
The results of this study will highlight any current industrial pollution injustices, and the methods
used will provide future researchers with a replicable framework to re-assess EDJ and industrial
pollution in the future.
1.1 A testable principle for distributive justice
In this study we apply a relatively simple strict equality distributive justice principle drawn from the
theoretical argumentation in broader distributive justice literature (Allingham, 2014), as such we
posit:
1) exposure to harmful pollution (i.e. pollution beyond a safe threshold) will always reduce human
well-being by causing disease and/or death;
2) all people residing in Australia of any socio-economic group are morally and legally equal, and
lastly;
3) given all people residing in Australia are equal, the rates of exposure to industrial pollution ought
to be the same for all socio-economic groups.
Using these premises our research offers an analysis determine whether our strict equality
distributive justice principle is violated for industrial pollution in Australia. As such, our research
question is:
Is there evidence of social inequities in exposure to harmful industrial pollution in Australia?
In order to answer this research question, and in light of the findings of previous research in
Australia, we will test the following Hypothesis:
H1 = The prevalence of exposure to industrial pollution is related to socio-economic status.
H0 = The prevalence of exposure to industrial pollution is not related to socio-economic status.
It is important to note that with this study we are not aiming to prove a causative effect between
pollution exposure and associated health burdens, rather we are seeking to identify whether a
relationship between these variables is present.
Methods
We use spatial statistics to explore the relationship between industrial pollution and socio-economic
status, Australia wide. Our analyses required the aggregation and adjustment of several large, data
rich, and fine scale databases (figure 1). For each data set we use annual data taken at three time-
points: 2006, 2011, and 2016, to analyse any overall changes over the 2006-2016 decade. The script
for all data aggregation and analysis methods undertaken using R statistical software and associated
packages can be found in Supplementary file 1.
Page 70
70
Figure 1 Data analysis process. For each year (2006, 2011, and 2016) three large, Australia wide
data sets were used to create total toxicity potential ratings and analysed using regression models.
2.1 Spatial data
To best understand the distribution of industrial pollution within Australia, we have used fine scale
spatial data produced by the Australian Bureau of Statistics (ABS). For 2006, we use Collection
Districts (CD). There are 38,200 CDs Australia wide, with each representing an average of 225
dwellings (ABS, 2011). In 2011 the ABS structure changed to the Australian Statistical Geography
Standard, which resulted in CDs changing to finer scale Statistical Areas (SA), as such we use SA1
units (the smallest SA unit) for years 2011 and 2016. The SA1 units have approximately 400 people
each, and as such the number of SA1 units will change as the Australian population changes. At the
time of writing, there are 57,523 SA1 units (ABS, 2017b). Research has shown that using fine scale
data such as CD and SA1 produces the most reliable results for data analysis (Hanigan et al., 2017).
Page 71
71
2.2 Industrial pollutant emissions data
To examine the distribution of industrial pollution, we use data from the Australian National
Pollutant Inventory (NPI) (NPI, 2015). The NPI is a sub-division of the Australian Department of
the Environment and Energy and provides a self-reporting and tracking system for industrial
pollutants in Australia. It primarily acts as a publicly accessible, online database of emissions data.
Australian law requires that all facilities which exceed particular (substance dependent) emissions
thresholds for 93 defined substances must self-report emissions to the NPI (NPI, 2014). Emissions
records include both fugitive and point source. Here we utilise all point source emissions data from
the NPI (from the appropriate annual reports), which includes emissions data for every polluting
facility in the country, in order to estimate a potential path of pollutant exposure to surrounding
communities. The 93 substances are those that potentially harm people or the environment. The NPI
annually collects, records, and reports data for emissions to air, water and land. It is critical to note a
key limitation to the NPI data is that the reported data are available only above the threshold at
which reporting is required (NPI, 2015), meaning very low level emissions from individual
pollution sources are not recorded.
2.3 Total Toxicity Potential
The NPI records the mass of emissions from each source but does not give any indication of their
human toxicity potential. As such, adjusting the NPI emissions data for toxicity is critical for any
meaningful use in impact analyses, as a small volume of a highly toxic pollutant is potentially much
more harmful than a large amount of a less toxic substance (Wu et al., 2009). To address this we
have utilised the Human Toxicity Potentials (HTP) defined by Hertwich et al. (2001) to weight NPI
emissions data. The HTP is an index that was specifically designed for use with emissions
inventories such as the NPI. Its primary use is to understand the potential harm caused by pollutant
release. HTP scores reflect the potential harm of a unit of chemical released, as each HTP score is
based on the intrinsic toxicity of a substance, and accounts for a number of exposure routes. The
HTP classifies chemicals into two different vectors (air and water) and two types of health effects
(cancer causing and non-cancer-causing). While it is very extensive, the HTP is not exhaustive.
There are many substances (several of which are known to be harmful to human health) which are
not included in the HTP, as such we have excluded from our analysis any pollutants present in the
NPI which do not have a HTP score. We have weighted the NPI point source emissions data by the
HTP scores to produce a Total Toxicity Potential (TTP), per each emissions point source ( ).
Page 72
72
For simplification, we have combined the air, water, and cancer and non-cancer-causing scores into
a single toxicity potential value , which is described by:
Total toxicity potential for each point source = Sum of (air emissions*cancer effects) + (water
emissions*cancer effects) + (air emissions*non-cancer effects) + (water emissions*non-cancer
effects)
NPI emissions are represented by E, Cancer effects by C and Non-cancer by N, and all emissions
and health effects per point source i are separated into air borne a, and water borne w emissions. It
is important to note that as each chemical can have both, either, or neither C and N health effects,
emissions can potentially contribute nil, once, or twice to the TTP. Any potential interactions
between C and N are not accounted for.
2.3.1 Mapping the distribution of Total Toxicity Potentials ( ) across Australia
Using ArcGIS Desktop (ESRI, 2011), for each year included in the study, we assigned TTP values
to each CD (for 2006) or SA1 unit (for 2011 and 2016) based on proximity to each emissions point
source. For each emissions point source (provided with NPI data, and represented as point
coordinates with six decimal places of accuracy) we added a one kilometre buffer to compensate for
the ‘edge effect problem’ as per Chakraborty and Green (2014), which is methodologically in line
with an analytical review of environmental justice studies (Bowen, 2002). This provides a realistic
area of effect of pollution around the point source. Any SA1 unit which fell (even partially) within
this one kilometre buffer was assigned the of the point. There were many spatial units which
intersected with multiple buffered points, in these cases, all were summed.
2.4 Socio-demographic Advantage and Disadvantage
To examine social and economic groups within each spatial unit, we have used four indicators
which make up the Socio-Economic Indexes for Areas (SEIFA) in line with previous studies
(Knibbs and Barnett, 2015; Chakraborty and Green, 2014). SEIFA is an index developed by the
Australian Bureau of Statistics (ABS) to score areas (e.g. SA1 units) according to their relative
socio-economic disadvantage and/or advantage (ABS, 2018a). The index consists of four sub-
indices created from weighted social and economic data, gathered from the five-yearly Australian
Census (ABS, 2018a). The four indices in SEIFA are: Index of Relative Socio-Economic
Page 73
73
Disadvantage (IRSD); Index of Relative Socio-Economic Advantage and Disadvantage (IRSAD);
Index of Economic Resources (IER); and Index of Education and Occupation (IEO). SEIFA is
particularly useful for Australian studies as it was specifically designed to capture those variables
which are most representative of Australian culture and diversity. These indices consist of values
which are standardised to a mean of 1000 and a standard deviation of 100 (ABS, 2018a).
In addition to SEIFA data, we have included include percent Indigenous population per spatial unit,
using data from the Australian Census (ABS, 2018b). Previous work on environmental justice in
Australia highlighted a correlation between Indigenous Australian populations and pollution (e.g.
Knibbs and Barnett, 2015). It is worth noting that the SEIFA 2006 IRSD sub-index included percent
of people identifying as Indigenous. However, this was removed for later iterations (SEIFA 2011,
2016). In sum, we analysed both socio-economic advantage/disadvantage via the SEIFA indices and
the percent Indigenous per spatial unit against the pollution data.
For each regression model we used TTP as a dependent variable and the socio-demographic
advantage and disadvantage indices plus percent Indigenous population as separate, independent
variables. We constructed each model to understand if any particular indicators of social advantage
or disadvantage can predict the levels of TPP, or have more or less explanatory power for variance
within the TTP.
Due to the truncation of the NPI pollution data (i.e. no values below the threshold for reporting to
the NPI) we have analysed the data using two separate regression methods. Initially we analysed the
data using a series of truncated regression models, using the truncreg R package (Croissant and
Zeileis, 2017). We created models for each of the five socio-demographic advantage and
disadvantage variables per year (four SEIFA sub-variables plus percent identification as
Indigenous), and estimated R2 values by squaring the correlations between TTP values and
predicted values (Croissant and Zeileis, 2017). Truncated regression models are useful to
accommodate for variables being truncated (in this case left truncation of TTPi < min(TTP)), and
address the bias introduced when using ordinary least squares regression, however they do have
limitations. For example, for truncated models the variance of outcome variables is reduced
compared to the non-truncated distribution. Also, the mean of the truncated variables will differ
from non-truncated variables.
Page 74
74
Due to the limitations of truncated regression, we also constructed standard linear regression
models, using ad hoc imputation for TTPi < min(TTP) observations. This imputation method
simulates a complete, non-truncated dataset. Due to the nature of how the NPI pollution data are
only collected once they go over particular thresholds, the missing data are not random. We used
the R Multiple Imputation by Chained Equations (MICE) package to impute values (van Buuren
and Groothuis-Oudshoorn, 2011), using the predictive mean matching (PPM) method. For each
missing value PPM randomly fills in a value from the observed values using an observation which
has similar regression predicted values based on a simulated regression model (Heitjan and
Roderick, 1991). As such, data imputed with PPM closely maintain the distribution of the observed
values.
For each instance where TTP values were missing (i.e. TTP = 0 due to the reporting thresholds), we
generated multiple imputations (n = 50) (see supplementary file 2 for counts of ‘missing’ data per
year). We used the five socio-demographic advantage and disadvantage variables (i.e. four SEIFA
sub-variables plus percent identification as Indigenous) to generate imputed values. In effect, this
analysed the relationship between existing TTP data and socio-demographic data and used this to
provide an estimate of a likely value for the missing TTP data. The validity of the 50 imputed data
sets was assessed using means, confidence intervals (.95), and by visually comparing the
distribution of imputed and non-imputed data (see supplementary file 3 for validity figures).
Following confirmation of validity of the imputed data, analysis for a relationship between TTP and
socio-demographic variables was conducted. The 50 imputed data sets were pooled and analysed using
simple linear regression models, using log(TTP) and each socio-demographic advantage and
disadvantage variable. Note that the regression model for 2006 differs from the 2011 and 2016
models due to later removal of percept Indigenous from the IRSD sub-index of the SEIFA (see
section 2.4). As a result, 2011 and 2016 were analysed via multiple regression as they also included
the percent Indigenous population variable.
Results
We analysed total toxicity potentials (TTP) against socio-demographic measures of advantage and
disadvantage across Australia’s smallest available statistical unit (SA1). A spatial representation of
the distribution of TTP in Australia’s SA1 units is presented in Figure 2.
Page 76
76
c)
Figure 2 Relative TTP 2006 CDs (a), 2011 SA1s (b), and 2016 SA1s (c). Each year is displayed
with each spatial unit’s TTP rating (normalised between 0 to 1, and using 10 natural breaks, for
readability). Melbourne and Sydney are inset to demonstrate the very small spatial units in high
population density areas.
3.1 Summary statistics
Table 1 summarises the descriptive statistics for all data used in the models. Note that distribution
of the TTP values are heavily left skewed in each year.
Table 1 Descriptive summary statistics for all variables. TTP = total toxicity potential; IRSAD =
Index of Relative Socio-Economic Advantage and Disadvantage; IRSD = Index of Relative Socio-
Economic Disadvantage; IER = Index of Economic Resources; IEO = Index of Education and
Occupation (IEO); Pct. Indig. = Percent Indigenous in the population.
Variable n Mean Std. dev. Min Max Std. err.
2006 TTP 37,457 134,341,185.28 6,417,649,383.35 0.00 498,070,637,541.00 33,159,615.88
IRSAD 37,457 1,000.01 100 474.00 1,306.00 0.52
IRSD 37,457 1,000.01 100 205.00 1,199.00 0.52
IER 37,457 1,000 100 272.00 1,276.00 0.52
IEO 37,457 1,000 100 597.00 1,365.00 0.52
Pct. Indig. 37,457 2.53 7.98 0.00 104.30 0.04
Page 77
77
2011 TTP 46,743 60,463,819.19 2,443,508,383.30 0.00 194,048,594,161.00 11,302,006.23
IRSAD 46,743 999.66 99.92 298.00 1,246.00 0.46
IRSD 46,743 1,000.12 99.70 121.00 1,184.00 0.46
IER 46,743 999.68 98.52 283.00 1,290.00 0.46
IEO 46,743 1,000.58 101.04 528.00 1,289.00 0.47
Pct. Indig. 46,743 2.56 6.88 0.00 102.47 0.03
2016 TTP 55,014 23,309,284.45 841,280,692.70 0.00 73,741,562,651.50 3,586,776.46
IRSAD 55,014 999.99 100.01 188.00 1,186.00 0.43
IRSD 55,014 999.99 100.00 400.00 1,239.00 0.43
IER 55,014 1,000.04 99.95 245.00 1,281.00 0.43
IEO 55,014 999.99 99.96 523.00 1,283.00 0.43
Pct. Indig. 55,014 2.91 7.09 0.00 101.98 0.03
3.2 Proportions of variance explained and statistical significance
The statistical analysis we conducted to interrogate whether there is a relationship between TTP and
socio-demographic variables produced a number of models that we use to interpret the results. One
model was produced for each year with each socio-demographic variable and for both the truncated
and imputed analysis (Table 2). As such, thirty models were produced in this analysis.
Several models report statistically significant results (α = 0.05), however the estimated R2 sizes are
very small (R2 < 0.01) for all models (Sawilowsky, 2009). Socio-demographic advantage and
disadvantage categories accounted for a maximum of 0.11 percent of TTP variance under the
truncated models, and a maximum of 0.06 percent under the pooled imputed models. This indicates
that less than one percent of variance in the exposure of CD and SA1 units to industrial pollution
can be explained by differences in the socio-demographic variables we included in this analysis.
Based on this and following Sullivan and Feinn (2012) and Nakagawa and Cuthill (2007), although
some models yielded results that were by definition statistically significant, given the very small R2
values for those significant models we fail to reject the null hypothesis, which states the prevalence
of exposure to industrial pollution is not related to socio-economic status (H0).
Table 2 Proportions of TTP variance explained by socio-demographic advantage and
disadvantage for years 2006, 2011, and 2016 (α = 0.05). Note the 2006 SEIFA included percent
indigenous people as a variable within the IRSD index.
Page 78
78
Truncated regression Pooled imputed data
regression
Year Variables Estimated R2
Truncated p-value
Estimated R2
Pooled p-value
2006 IRSAD 1.80E-06 0.94 7.20E-05 0.8
IRSD* 6.70E-06 0.88 4.70E-05 0.72
IER 1.20E-04 0.53 5.08E-05 0.95
IEO 4.50E-06 0.9 8.63E-05 0.66
Pct. Indig. 1.10E-03 0.04* 3.21E-04 0.06
2011 IRSAD 6.70E-05 0.63 2.40E-05 0.71
IRSD 1.05E-03 0.7 5.60E-04 0.88
IER 3.90E-05 0.71 5.80E-05 0.32
IEO 3.30E-07 0.97 2.29E-05 0.88
Pct. Indig. 6.90E-04 0.1 4.26E-04 0.03*
2016 IRSAD 2.00E-04 0.38 3.43E-05 0.34 IRSD 6.13E-05 0.62 1.31E-04 0.45 IER 7.60E-04 0.09 3.53E-05 0.4 IEO 9.80E-07 0.95 1.36E-05 0.89 Pct. Indig. 1.20E-03 0.02* 9.52E-05 0.07
*denotes statistically significant result (α = 0.05).
Discussion
With this study we aimed to understand if there are implications for environmental distributive
justice from industrial pollution in Australia. Based on previous studies we hypothesized that the
prevalence of exposure to industrial pollution would have a relationship with socio-economic
variables. However, our results show there is little evidence to support this relationship.
Only models which used percent of people identifying as Indigenous yielded any statistically
significant results. While any type of distributive justice issue that disproportionately burdens
Indigenous people is a critical issue, the analysis yielded very small estimated R2 values even at an
overly generous type I error threshold of 0.05. The largest of these estimated R2 values was 0.0011
(2011 IRSD), meaning that the proportion of people identifying as Indigenous explained only 0.1%
to the variability in the Total Toxicity Potential (TTP) of industrial pollution. Given the very small
effect size of this result, we do not consider this to be a finding meaningful to the question of
environmental distributive justice concerning industrial pollution in Australia. However, we note
that this finding may indicate that Indigenous Australians may be most prone to risk should the
spatial distribution of industrial pollution change in a way that concentrates pollution in higher-TPP
areas as reduces pollution in lower-TPP areas.
Page 79
79
This suggests that indicators of socio-economic demographics (included in this study as the SEIFA
sub-indices and percent of people identifying as Indigenous) do not have a meaningful relationship
with the distribution of harmful industrial emissions. This finding held true for both analytical
approaches (truncation or imputation). Using the best available data, at the finest scale available, we
did not find sufficient evidence to suggest that the prevalence of exposure to industrial pollution is
related to socio-economic advantage and disadvantage. As such, and although we did find some
statically significant relationships, due to the very small effect sizes, we fail reject the null
hypothesis.
This finding is somewhat contrary to previous studies on the distributive patterns of pollution in
Australia (in particular, we did not find a large effect size on Indigenous Australians cf. Knibbs &
Barnett (2015)). However, this does not suggest that there are no pollution based inequities in
Australia, (e.g. our results are not incompatible with Knibbs and Barnett (2015), as they specifically
assessed ambient and poorly regulated air emissions of a specific substance (NO2). Furthermore,
our results do not suggest that there are no inequities in exposure to pollution from industrial
emissions sources. Case study based work in Australia (e.g. Higginbotham et al.(2010)) highlights
many situations where individual communities suffer a highly disproportional and unfair burden
from local industry. What our results do show is that in aggregate across Australia there is no
violation of the strict equality distributive justice principle for industrial pollution when examined
via quantitative spatial analysis.
These findings indicate that Australia is equitably governing the distribution of industrial pollution
when it comes to broad, national-level trends (e.g. Keywood et al., 2016). Either by regulatory
design or by chance of the history of industrial development and activity (or perhaps a combination
of both), there is no evidence of a pattern of disproportionate impacts on the most socio-
demographically disadvantaged groups; a key goal for environmental distributive justice efforts.
This is contrary to findings of a similarly-designed quantitative spatial study of pollution and socio-
demographic advantage and disadvantage, where analyses have found a clear trend of greater
impacts on the socially disadvantaged across different countries (e.g. Li et al., 2018; Cushing et al.,
2015; Stewart et al., 2014; Delpla et al., 2015; Fecht et al., 2015). Nevertheless, the regulatory
framework for industrial pollution in Australia is not without its limitations. The National Pollutant
Inventory (NPI) sits within a portfolio of policy instruments, nested beneath the newly developed
National Clean Air Agreement – a framework for cooperation between the Australian
Page 80
80
Commonwealth and State Governments that is in phase 2 of development (2018-2020) (Australian
Government, 2015). As a key instrument for the regulation of industrial pollution, the NPI has been
critiqued for being an industry-friendly mechanism lacking stringency in order to encourage
industry buy-in and for failing to fulfil its legislated goals (Howes, 2001; Cooper et al., 2017). This
is important as special and business interests have affected the design of air pollution policy in the
past (Crepaz, 1995). Dobbie and Green (2015) also found that when the network of national and
state systems for regulating industrial pollution have succeeded in identifying breaches, few
prosecutions follow, suggesting deficiencies with enforcement. These perspectives highlight the
importance of interpreting our results correctly positioned within their policy setting and in the
context of qualitative case study examination of clear instances of inequitable distribution of the
impacts of industrial pollution (e.g. Dobbie and Green (2015), Cooper et al. (2018), and
Higginbotham et al. (2010). From this positioning, we report the undoubtedly positive finding that
in aggregate and at the national level there is no evidence of a violation of the strict equality
distributive justice principle for industrial pollution when examined via qualitative spatial analysis.
But this does not mean there are no instances of unjust distribution of environmental burdens in
specific cases, or in instances that are not defined by socio-demographic categories that describe
social disadvantage. Similarly, this does not suggest the Australian policy framework cannot be
improved, and we encourage ongoing and critical development of the National Clean Air
Agreement.
4.1 Study limitations
The aim of this study was to provide a thorough and replicable method with which to understand
the relationship between industrial pollution and socio-demographic advantage and disadvantage
across Australia. However, it does have several limitations which must be considered before using
our findings in a decision-making context. Most importantly the recognition that this study pertains
to risks to human health only (e.g. TTP is total toxicity potential), and does not make any
correlations with actual burdens of disease (such as work by Holniki et al. (2017)). In addition,
while we aimed to use the highest resolution, and most detailed data available, the data are not
without limitations. The list of 93 substances includes most harmful pollutants emitted within
Australia, the list is by no means exhaustive. Some substances reported to the NPI are not rated in
the HPI including highly toxic substances and as such cannot be weighted. Furthermore, the HPI
treats unknown values for cancer and noncancer impacts of a substance as 0, due to a lack of
knowledge on the specific human health impacts. However, some of these substances may actually
Page 81
81
be carcinogenic or disease causing, as such our data likely underestimates the threat to human
health caused by industrial emissions in Australia.
The available data, too, may be insufficient. Dobbie and Green (2015) argue that records of
pollution developed via collection of time point data may obscure major spikes in pollution between
the points of reporting, and there are risks that self-reporting by industry bodies may undermine the
quality of available data (Howes, 2001). In lieu of agencies such as the ABS and the NPI producing
more detailed data, or independent data collection of emissions sources of every harmful substance
from a nationally representative sample, combined with a comprehensive update of the HPI, our
study (at time of completion) is likely the highest resolution and most detailed possible for
investigating this issue. Future research should examine methodological issues, such as the use of
different buffer sizes, spatial units and pollution dispersion pathways.
Conclusions
Our study aimed to determine prevalence of, and identify, any associations between exposure to
industrial pollution and socio-demographic advantage and disadvantage. We find no compelling
evidence of association. This result indicates that Australian industrial emissions conform to strict
equality distributive justice principles. Understanding whether this conformity is deliberate and by
design, or a positive accident is outside of the scope of this study, though we have discussed some
of the interactions with Australian pollution policy and highlighted contrasting and supporting
insights from the extant literature. Further research in this area should address this knowledge gap
and examine why our finding differs from similar studies in similar countries (e.g. other Western
liberal democracies) and what other nations can (and can not) learn from Australian industrial
pollution and planning regulations. Furthermore, if higher resolution or more detailed data become
available, this study ought to be replicated, addressing the data limitations.
Acknowledgements
We would like to extend thanks to the following people whose contributions made this project
possible. Staff from the Australian Bureau of Statistics for their patient and thorough advice. Staff
at the National Pollutant Inventory for their advice and knowledge around toxicity weighting data.
Dr. Phil Kokic, and Dr. Grace Chiu from the Australian National University for their kindness and
advice for statistical modelling of the data.
Page 82
82
Author Contributions
G.A., designed and performed the study, analysed results. G.A. and R.C. wrote the manuscript. B.W
contributed to editing the manuscript.
Page 83
83
Chapter 5 Large-scale environmental degradation results in
inequitable impacts to already impoverished communities: A case
study from the floating villages of Cambodia
Online publication: http://dx.doi.org/10.1007/s13280-018-1022-2
Full citation: Althor G, Mahood S, Witt B, Colvin R, Watson J. (2018) Large-scale environmental
degradation results in inequitable impacts to communities: a case study from the floating villages of
Cambodia. Ambio 47(7): 747-759.
Overview
This study was published in Ambio in 2018. This study is somewhat unique in this thesis as it is
based on a purely qualitative method. This work is a result of time that I spend in the in remote
Cambodian floating villages. During my time in Cambodia I was able to meet with local fishing
community members and representatives to understand their lived experiences in the context of a
rapidly changing environment, which they are highly dependant on for subsistence needs. This
study fits within a broader literature which has examined the bio-physical and social impacts on the
Tonle Sap Great Lake as a result of climate change, land degradation, and damming along the
Mekong.
Research question: How are the fishing communities of the Tonle Sap floating villages experiencing
major environmental changes?
Sub-questions:
1) How are floating village communities experiencing a change in fisheries; and
2) How are floating village communities responding to these changes.
This study specifically utilises my modified (from Rawl’s second principle of justice (Rawls,
1971)) distributive justice principle, as follows:
P1: Social and economic inequalities, resulting from the utilisation of natural capital, are to be of
the greatest benefit of the least advantaged members of society.
Highlights:
• Adds an updated and lived experience narrative to a well-established body of literature on
this topic.
Page 84
84
• Demonstrates the injustice of how over 100,000 rural Cambodians are being impacted by the
degradation of natural capital from which they receive no benefit.
• Further highlights the applicability for distributive justice principles in investigating issues
of natural capital utilisation.
Page 85
85
The following text appears exactly as published.
Abstract
Cambodian subsistence communities within the Tonle Sap Great Lake area rely on resource extraction from
the lake to meet livelihood needs. These fishing communities – many of which consist of dwellings floating
on the lake – face potentially profound livelihood challenges because of climate change and changing
hydrology due to dam construction for hydroelectricity within the Mekong Basin. We conducted interviews
across five village communities, with local subsistence fisher people in the Tonle Sap in 2015, and
used thematic analysis methods to reveal a fishery system that is undergoing rapid ecological
decline, with local fishing communities increasingly experiencing reductions in available fish
stocks. As a result, over 100,000 people living in these communities are experiencing a direct loss
of well-being and livelihood. We discuss these losses and consider their implications for the future
viability of Cambodian floating village communities.
Keywords
Climate change, distributive justice, human well-being, Mekong basin, subsistence livelihoods,
Tonle Sap Great Lake.
Introduction
Globally, significant changes to ecosystems and the services they provide to people are occurring.
The diffuse effects of climate change are altering entire natural systems such as hydrological
regimes, which is compounded by the impacts of land use change and development. A range of
diverse and complex factors drives these changes, including economic growth in the developed
world (Leal Filho et al., 2017), to gazettement of protected areas for nature conservation (Watson et
al., 2014), and natural resource development for poverty alleviation in the developing world
(Manorom et al., 2017).
Common to these drivers of major global change are distributive justice implications. Viewed from
an environmental distributive justice perspective, such changes commonly produce inequitable
outcomes for the world’s poorest people, particularly in least developed countries. Rarely are the
distributions of well-being benefits derived from ecosystem degradation proportionate to the
resultant burdens that the poorest people are forced to bear (Pauline Dube and Sivakumar, 2015).
Furthermore, there is strong evidence which suggests that by contributing to human well-being,
ecosystem services (that is those natural, biophysical processes people use to support their well-
Page 86
86
being) can reduce poverty (MEA, 2005), as examined by Fisher et al. (2014) in their framework for
analysing ecosystem services and poverty alleviation (ESPA) . As such, not only are benefits and
burdens resulting from ecosystem exploitation inequitably distributed to poor communities, their
degradation also commonly results in a double inequity by reinforcing local poverty (Berbés-
Blázquez et al., 2016).
While the environmental justice literature has a distinct focus on the justice implications of global
climate change, there are far fewer studies examining the impacts of large-scale environmental
changes on small, subsistence communities. Here we aim to partially address this gap by examining
the state of environmental distributive justice within the Tonle Sap Great Lake (hereafter ‘Tonle
Sap’). To do so, we use Rawls’ (Rawls, 1971: , pp. 5–6) second principle of distributive justice
(Rawls, 1971: , pp. 5–6), applied through an environmental justice lens, that is: Social and
economic inequalities, resulting from the degradation of natural resources, are to be to the greatest
benefit of the least advantaged members of society.
Located in Cambodia, the Tonle Sap is a prime example of a highly productive, valuable and
vulnerable wetland. Globally, wetlands are a key ecosystem type that provide many essential
ecosystem services, such as supporting and regulating biodiversity habitat, provisioning food and
other material goods, and cultural services (UNESCO, 1994). Wetlands such as the Tonle Sap are
generally highly productive, and are among the most valuable ecosystems on the planet (Al-Obaid
et al., 2017). Due to being highly productive, wetlands around the world commonly support the
well-being of large groups of people via ecosystem services, and as such, the well-being of these
people are reliant on a healthy ecosystem (Fisher et al., 2014). However, wetlands are also
characterised by their high level of vulnerability to systemic changes.
The Tonle Sap is located in rural Cambodia, where people often live subsistence lifestyles, meaning
they are highly dependent on the direct provisioning of ecosystem services to meet their livelihood
needs (Keskinen, 2006). Over one million Cambodians directly depend on the Tonle Sap system,
where they fish and farm food resources for day to day consumption and livelihood generation
(Sokhem and Sunada, 2006; FAO, 2017). The lake is particularly critical for the livelihoods of over
100,000 people who live in ‘floating villages’; communities which consist of dwellings that
permanently float on the water of the Tonle Sap (Figure 1; (Keskinen, 2006)). The livelihoods of
people living within these communities are based predominantly on fishing activities (Keskinen,
2006). As such, fish extracted from the lake are the main source of protein and livelihood
Page 87
87
generation for residents (CDRI, 2013; Ratner et al., 2017). Given this dependence on the lake’s
natural resources, floating village communities are particularly vulnerable to any environmental
changes that affect the fishery.
Unfortunately for these communities, environmental change in the Tonle Sap has been rapidly
increasing over recent decades. Some of these changes are driven by the activities of the large
human population living within the system (Keskinen, 2006), and the high sensitivity of fresh
wetlands to disturbances (Middleton and Souter, 2016). However, the most significant
environmental pressures are large-scale and external to the system. Specifically, climate change and
the modification of the Mekong River are the most substantial drivers of declines in the health of
the lake’s natural systems (Arias et al., 2014; Arias et al., 2012). While the activities which generate
these pressures (such as manufacturing and damming) are undertaken to produce economic growth
and enhance social well-being for some, it is likely that the environmental declines they also
generate result in inequitable outcomes for floating villages. While these communities feel the
effects of environmental changes most directly and profoundly, they receive few benefits from the
development driving such change. As such, our present research examines the environmental
distributive justice implications of major environmental changes on people living in the floating
villages of the Tonle Sap. We aim to contribute to an understanding of how environmental inequity
may affect subsistence communities.
Page 88
88
Figure 1 Prek Toal, is one of many floating villages located within the Tonle Sap. Most float
on the water using barrels, or bundles of bamboo attached beneath the dwellings.
The biophysical dimensions of the Tonle Sap Great Lake
In order to understand the livelihoods of people living within the Tonle Sap Great Lake, it is
important to understand its unique and complex biophysical dimensions. The Tonle Sap is the
largest lake in South East Asia and is part of a complex hydrological system that includes the Tonle
river and the Mekong River which is the lake’s primary tributary (Kummu and Sarkkula, 2008).
The most distinguishing feature of the lake is its exceptional annual flood cycle. Each year, the
Tonle Sap floods and expands from ~2,500-3,000 km2 in area and an average ~1.5m in depth, up to
~10,000-16,000 km2 in area and an average ~10m in depth (Figure 2). This flood occurs each year,
between July and October, and is driven by the South East Asian monsoonal regime. During this
period the monsoon generates over 60% of the Mekong Basin’s total annual rainfall (Arias et al.,
2014). The rains drive a vast flow of water into the Mekong river, which in turn reverses the flow of
the Tonle river. During the flooding large amounts of nutrients are pumped into the lake, which is
the foundation of its highly diverse set of freshwater ecosystems and fish species (FAO, 2001;
Kummu et al., 2008b; Ratner et al., 2017; FAO, 2017). As such, the lake is characterised by its
fisheries, which are among the most productive in the world (Keskinen, 2006). The most recent
estimates of total annual catch are between 180,000 – 250,000 tonnes (Chea et al., 2016). Catch
from the Tonle Sap contributes around 50 percent of Cambodia’s total catch, and around 16 percent
of the national GDP (van Zalinge et al., 2000). Therefore the annual flood pulse is critical for the
health of the livelihoods of people living within it (Keskinen, 2006), and for the lake’s natural areas
(Arias et al., 2014).
Page 89
89
Figure 2 The location and seasonal extents of the Tonle Sap Great Lake, tributaries, and key
conservation areas. Map created using ESRI ArcGIS Pro with data provided by Wildlife
Conservation Society, Cambodia (ESRI ArcGIS Pro, 2017).
The Tonle Sap’s fisheries are dependent on a sustainable ecological system (van Zalinge et al.,
2000). As such, the livelihoods of villagers living within the Tonle Sap are completely reliant on the
fitness of the entire system’s functionality, particularly its regular hydrological flood pulse system
(Arias et al., 2014). Furthermore, much of the lake’s biodiversity have life cycles that are
synchronised to the flood pulse (Arias et al., 2014). Within these different ecosystems that range
from floodplains to seasonally flooded grasslands, shrub-lands, and gallery forests (Middleton and
Souter, 2016), hundreds of fish, plant, bird, reptile, invertebrate, and mammal species occur, many
of which have high nutritional or conservation value (Roos et al., 2007; Arias et al., 2014; Campbell
et al., 2006). For a significant majority of Tonle Sap species the seasonally flooded ecosystems
function as key habitat, which makes them critical for the ecological health of the lake (Arias et al.,
2012).
Page 90
90
Environmental Protection within the Tonle Sap
In addition to the complex biophysical dimensions of the Tonle Sap, and due to its important social
and ecological values the lake is also host to a diverse and intricate network of protected areas
(figure 2). Understanding these protected areas is important as the lake’s floating village
communities often occur within and around them. Among the largest protected areas is the Tonle
Sap Biosphere Reserve. Designated in 2001 by the United Nations Educational, Scientific and
Cultural Organization (UNESCO) the Tonle Sap Biosphere Reserve officially recognizes the lake’s
ecological significance (UNESCO, 2007). The biosphere reserve aims to promote the conservation
and sustainable use of the Tonle Sap’s natural resources, with a particular focus on the lake’s
flooded ecosystems (UNESCO, 2015). It hosts three of the lake’s most ecologically diverse areas,
which UNESCO designated as ‘core zones’: Prek Toal, Boeng Tonle Chhmar, and Stoeng Sen
(UNESCO, 2007). Core zones act as ‘strict protected areas’ where any exploitation of natural
resources is prohibited under Cambodian law. Surrounding the core zones are buffer and transition
zones, which allow increasing degrees of human activity. The transition zones are of particular
importance to many floating village communities as this is where people are permitted to engage in
economic activities, such as fishing. Overlapping the core zones at Prek Toal, and Boeng Chhmar
are two Ramsar designated sites (Ramsar, 2017). Ramsar designation highlights these sites as
wetlands of international ecological significance, but confers no legal protection. In addition to the
globally recognized conservation sites are a network of over 400 locally managed freshwater
community fisheries and fish sanctuaries (Sok et al., 2012). Finally, there are three concentric
ecological zones around the lake: the terrestrial zone where development activities are permitted;
the floodplain zone where only specified development activities are permitted, and; the aquatic zone
where habitat is strictly protected (Mak, 2015). Adding to the complexity of the protected areas is
the overlapping jurisdictional and governance structures of the Cambodian authorities responsible
for their management. For example, the biosphere reserve and Ramsar sites are managed by the
Ministry of Environment, whereas the Fisheries Administration of the Ministry of Agriculture,
Forestry and Fisheries is responsible for the management of the community fisheries and fish
sanctuaries. Furthermore, the Tonle Sap Authority is responsible for ensuring coordinated natural
resource management across the entire Tonle Sap within the three ecological zones. All together the
various protected areas and management authorities operating within the Tonle Sap make
understanding its management and protection challenging for local people (Gillespie, 2016).
Page 91
91
Threats to the ecosystem and fishery
Regardless of the many layers of environmental protection within the Tonle Sap, it is still a system
under significant threat. Since the 1980s the Tonle Sap has undergone significant environmental
change due to clearing of vegetation, declines in water quality, and increased extraction of resources
from people (Cooperman et al., 2012; Arias et al., 2014; Ratner et al., 2017). Furthermore, the lake
is affected by major ongoing changes to its hydrology, flow regime, and flood pulse, which are due
to rapid development of the upper reaches of the Mekong River (Lamberts, 2008; Arias et al., 2012;
Ziv et al., 2012).
Countries in the upper region of the Mekong River are increasingly constructing hydropower dams
and reservoirs (Arias et al., 2012). For instance, there are more than 40 mainstream and tributary
dams, irrigation, and water supply projects planned for implementation across Lao PDR, Thailand,
China, Vietnam and Cambodia in the next 20 years (Mekong River Commission, 2016). The
Mekong infrastructure projects put stress on the entire socio-ecological Mekong system, and in
particular the Tonle Sap (Lauri et al., 2012). Of specific concern for the health of the Tonle Sap is
the alteration of flow regimes (Arias et al., 2014). Damming of the Mekong dampens the annual
flood pulse, which is vital for the health of the lake’s ecosystems (Arias et al., 2012). Predictive
modelling forecasts that dampening the flood pulse will create a twofold effect: in the dry season
the average water levels will undergo a permanent increase, while in in the wet season average
levels will be permanently decreased (Piman et al., 2013). This will result in the average,
permanently flooded area increasing between 17% and 40% (Keskinen et al., 2011). This is a
serious issue as previous research suggests that even small changes in seasonal water extent are
likely to compromise the lake’s productivity (Arias et al., 2012; Baron et al., 2002; Arias et al.,
2014). This is largely a result of the increase in permanently flooded areas dramatically reducing
seasonally flooded habitats, which are the most significant for species of high economic and
conservation values (Kummu et al., 2008a; Middleton and Souter, 2016; Cunningham et al., 2011).
Reductions in flooded habitats are likely to be rapid, as wetlands, in general, degrade much faster
than other ecosystems (MEA, 2000).
While modification of the Mekong is expected to be the primary driver of the Tonle Sap’s declining
health, climate change is also predicted to affect the system (Middleton and Souter, 2016; Arias et
al., 2014). Within the Tonle Sap, climate change is increasing both atmospheric and water
temperatures, the frequency of extreme weather events, and the variability of rainfall (Nuorteva et
al., 2010). In fresh water systems, climate change modifies hydrological patterns such as
Page 92
92
precipitation, evaporation, and flooding, (Erwin, 2009; Lemieux et al., 2014). These modifications
result in a scarcity of fresh water, and hydrological systems become strained.
Previous research has modelled many scenarios for both the effects that alteration of the Mekong
system and climate change will have on the Tonle Sap ecosystems (Arias et al., 2014; Arias et al.,
2012; Keskinen et al., 2011; Kummu and Sarkkula, 2008; Lamberts, 2008). Although there is a
degree of variability between studies, all agree that large tracts of the lake’s seasonally flooded
habitats will be lost. This will result in widespread reduction in the lake’s productivity (Lamberts,
2008). For floating village communities, reductions in the lake’s productivity is a serious cause for
concern. A decrease in the productivity of the lake will result in a decline in the fisheries, which is
the only source of livelihood generation for most households (CDRI, 2013; van Zalinge et al.,
2000). While there is a body of research highlighting how ongoing anthropogenic change will
adversely affect the natural systems of the Tonle Sap (Arias et al., 2014; Arias et al., 2013; Arias et
al., 2012; Kummu et al., 2008b; Kummu and Sarkkula, 2008; Orr et al., 2012), comparatively little
research has been conducted to understand how these changes will affect the lives of the
communities living within the lake’s floating villages.
As such, our research question is: How are the fishing people of the Tonle Sap floating villages
experiencing major environmental changes? To achieve this we interviewed people in five Tonle
Sap floating village communities using thematic analysis methods to record: (1) people’s
experiences living in this complex natural system; (2) how they perceive ongoing environmental
changes in the system, and; (3) if and how they respond to these changes. By collecting and
analysing these experiences and perceptions our aim was to understand the observations of, and
responses by, communities to ongoing changes in the lake’s ecosystems.
Methods
Prek Toal Biosphere Reserve and surrounding communities
We undertook an interview-based study in a selection of communities to understand and record the
experiences and perceptions of people living in floating villages within the Tonle Sap. We elected
to visit the five communities surrounding the Prek Toal Core Area of the Tonle Sap Biosphere
Reserve (Figure 3), as these communities were recommended to us by Cambodia-based
conservation practitioners as representative of the demographic makeup and livelihood model of the
Tonle Sap (fishing people), and due their proximity to other areas of interest. These villages
(Thvang, Kampong Prahok, Anlong Ta Uor, Prek Toal, and Kbal Taol - henceforth Prek Toal
Page 93
93
communities) are subsistence communities permanently floating within the Tonle Sap floodplain,
and are adjacent to several protected area and fishery management zones, and as such, are of
interest in understanding how environmental change affects people in the Tonle Sap. The Prek Toal
communities are adjacent to various Community Fisheries on one side, and to the former Fishing
Lot No. 2, now a Fish Sanctuary (no-take zone) that overlaps and surrounds the Prek Toal Core
Area of the Tonle Sap Biosphere Reserve (itself also the Prek Toal Ramsar site) within which there
is a Community Protected Area. Over 10,000 people make up these communities, and by
international standards many are living below the poverty line (Varis et al., 2006). This makes the
Prek Toal communities highly vulnerable to changes to the resources that provide for their
livelihoods. The five communities represent a relevant sample for understanding the complex trade-
offs and interactions between livelihoods, nature, resource management, and conservation in the
Tonle Sap.
Page 94
94
Figure 3. Map of floating villages, and other features surrounding the Prek Toal Biosphere
Reserve. Note that the background blue area in the main map indicates that the entire area is
flooded in the wet season. Map created using ESRI ArcGIS Pro with data provided by Wildlife
Conservation Society, Cambodia (ESRI ArcGIS Pro, 2017).
Data Collection
We conducted semi-structured interviews with subsistence fisher people (n=19) using qualitative
interview methods (Guest et al., 2006). Before any recorded interviews took place, we met with the
chief of each of the villages, and the commune chief. Village chiefs are responsible for governance
and community leadership within their village. Communes are a multi-village Cambodian
Page 95
95
administrative division. The commune chief we spoke to governs the Kaoh Chiveang commune,
which includes all five villages. There were several purposes for these meetings. First, we explained
the nature of our research, and importantly, asked for their permission and endorsement to conduct
research within their communities. Once permission was granted we asked the chiefs to assist us in
understanding the communities, and the issues people face, relevant to our research. As such, the
village chief interviews were vital for participant recruitment, and to design our interview questions
and interview guide (supplementary material 1).
With the assistance of the chiefs we were able to use purposive sampling, targeting those people
with fishing as their primary means of livelihood generation, who had long-term residence within
the community. This was decided as fishing is by far the most common and important livelihood
activity in the communities, and long-term residence would avoid some of the bias that occurs with
shifting baselines (Sáenz-Arroyo et al., 2005). Further participants were engaged using snowball
sampling (Lewis-Beck et al., 2004), by asking participants for recommendations following
interviews. Participant sample size (n=19) was established based on data saturation techniques
(Bernard and Gravlee, 2014; Bryman, 2012; Francis et al., 2010; Guest et al., 2006). We then
conducted interviews with all participants in their villages using participatory tools including
seasonal fishing charts (see supplementary materials 2), and participant-generated maps of the local
area and features (see supplementary materials 3). The creation and discussion of these charts were
critical as discussion tools during the interviews, however the sample size is too small for any
quantitative analyses. As some of the issues discussed in our interviews were potentially
contentious within the community or could place participants at risk, we collected no identifiable
information from participants. All discussions with chiefs and interviews with participants were
subject to ethics approval, granted by the University of Queensland (number 20150701). We
conducted all interviews in the Cambodian language, Khmer, with the assistance of a translator. All
interviews were recorded for later translation and transcription. It is important to note that as
interviews were translated, some meaning from either interviewer or participants may have been
unclear. In order to reduce this issue, we ensured interview prompts and probing questions were
simple and easy to communicate.
Data Analysis
The lead author undertook all analyses, including designing the strategy, developing the code book,
coding the data, and performing the thematic analyses. The data corpus consisted of (1) the
transcribed interviews, and participant-generated (2) fishing charts and (3) participant-generated
Page 96
96
maps. Professional Cambodian translators were contracted to translate and transcribe audio data
from Khmer into English. The field translator conducted translations of fish charts and maps. Data
were not quantified due to the small sample size suited to qualitative analysis, and the semi-open
ended structure of the interviews. The purpose of the analysis was to explore participant responses,
in the context of our research aims. The interview data largely informed the analytical strategy, and
we used inductive analysis to explore and code participant responses.
Thematic identification for creation of our code book was used (see supplementary materials 4).
Codes were developed using the literature, and key words, trends, themes, and ideas within the data
corpus (as per Guest et al., 2012). Throughout the screening processes only data considered relevant
to the research aims were analysed. All data coding was undertaken using R v. 3.3.3 (R Core
Development Team, 2015), with qualitative data analysis package RQDA v. 0.2-8 (Huang, 2014).
Coding required three data screening phases (as per Frith and Gleeson, 2004):
1. An initial screening of data for the purposes of identifying key and sub-themes, and
organisation of themes into a draft code book.
2. A second screening of data to test the precision of, and to refine, the code book.
3. A final screening of data, where we re-coded all data using the final, refined code book.
Thematic analysis methods were used to identify, analyse, and describe thematic patterns within the
coded data sets (Guest et al., 2006). As the aim was to capture community perceptions and
experiences, we determined thematic prevalence based on the number of interviewees who
discussed any particular topic. Data were analysed and described using a narrative approach (Braun
and Clarke, 2006; Kitzinger and Willmott, 2002), whereby we extracted an understanding of
participants’ realities from the data.
Results
Our analysis revealed several key themes from the data: (1) fishing is increasingly critical to local
livelihoods, (2) people believe that the fishery is highly vulnerable to environmental change, and;
(3) people believe the fishery is undergoing rapid declines.
Fishing is critical to local livelihoods
All participants explained that the majority of floating village families catch fish as their primary
source of food and income. Some villagers sometimes buy fish, but most villagers generally catch
Page 97
97
fish to meet their needs. For those who buy fish, buying is mainly restricted to the dry season, when
the catch is reduced.
There was little relevance of the fish species caught. Participants described how families will both
eat and sell any fish species that they catch. However, they tend to only eat the smaller fish, which
have less market value, whereas larger fish are sold. Some participants described how the majority
of the income derived from selling fish is spent on basic food, and in particular rice which is eaten
at each meal, year-round. Participants explained that the fishery changes from month to month, and
that different fish species occur at different times of the year, with the dry season being lengthened
and increasingly hot. As different species have different monetary values, this has the effect of
making family incomes seasonal.
Participants also discussed the local practice of aquaculture (fish and, very rarely, crocodile
farming). They explained that aquaculture is usually restricted to those in the community who are
relatively wealthy. This is due to the capital required to set up the fish farms, which require timber
and other resources which are scarce in the area. However, many participants also explained that eel
farming is a very common practice, as it does not need any capital. To raise caught eel people
simply use the hollows of bamboo stalks, submerged in water. The majority of fish produced in the
farms is sold to traders, with few farmed fish being eaten by farm owners. Many participants
explained that fish farmers will still catch some fish for consumption, but also for use as feed in the
farms. The majority of fish farmed in the region are Snakehead species (Channidae), meaning this
farming is technically illegal. However, participants spoke of how these species are of the highest
value for resale and as such, very desirable. There was no mention or evidence that these
restrictions are in any way enforced.
People believe the fishery is highly vulnerable to environmental change
When people where asked about when and where they go to fish, participants talked about how
villagers fish in all open aquatic and flooded habitats (grasslands, shrub lands and gallery forests) in
the area. However, access to different areas is seasonal, and can be restricted by Cambodian
authorities. Participants described differences in fishing between the wet and dry seasons. During
the dry season, villagers fish within and along the shore of the Tonle Sap, within and along the
shore of local streams and channels, and along the border of the protected areas. In the wet season,
people will fish anywhere that the water has risen. However, communities reduce fishing efforts in
the deepest areas of the lake during the wet season due to the danger of large waves and storms.
Page 98
98
While most people told us that they can never fish in the protected areas in either season, some
stated that “a little poaching” occurs within the protected areas as enforcement is either lax or prone
to minor corruption.
We also spoke with participants about when and where fish are the most abundant, and where the
greatest catch volume occurs. Participants explained that the greatest catch occurs during the
transition between the end of the dry season and the beginning of the wet season, as the water levels
rise. Participants told us that fish are most abundant in and around the nearby protected areas, and
that the best catch occurs along the borders of the protected areas. Participants also told us that the
catch is better along the shores of the lake and local streams than in other areas. When asked why
they thought the protected areas have the most fish, people told us that it is due to government
prohibitions, and because armed guards patrol these areas. Furthermore, they spoke of how these
areas are most abundant as this is where fish migrate to during the dry season. Many participants
told us that the protected areas are important for the local fish, however some participants said that
they are dissatisfied with prohibitions and that they would prefer complete fishing access. We asked
participants about the effects of extreme weather on the communities, including drought, very high
floods, and storms. Participants explained how drought (such as in 2002 and 2015) has terrible
effects on communities. The majority of people described that 2015 was the hottest and driest year
in living memory, and how the water level increased slowly during the start of the wet season. They
also explained how the lake water can become polluted and ‘smell bad’ during drought. Many
people described how during a drought the fish catch is much smaller than the average year. They
attributed this to many of the dry season fishing areas, such as local streams, completely drying up
or becoming stagnant, which kills the fish. Furthermore, participants spoke of how high water
temperature in local streams causes a high rate of fish mortality. Fish that die in this manner are
rendered inedible. We asked participants about the effect that unusually high annual floods (such as
in 2011 and 2013) have on families in the villages. All participants stated that during these floods
the fish catch is greater, and many also told us that the size of fish is larger. A consistent comment
from many participants was ‘big flood, big fish’. Additionally, we asked participants about the
effect of the severe storms which occur in the area. Most participants told us that storms are not
very disruptive, but can result in a short-term suspension of fishing activities, and damage fishing
gear.
Page 99
99
People believe the fishery is undergoing rapid declines
The participants spoke of perceptions about changes to the fishery in recent history (the last 10-20
years). There was a very strong consensus that the fishery has rapidly declined, both in terms of
catch size and individual fish size. Participants explained how every fish species they catch has
declined in abundance by anywhere between 30% and 90%, and that some species had disappeared
altogether. Many participants showed us that the size of some species had gone from the girth of
their forearm down to the girth of a few fingers. When prompted about why they thought that the
fishery had declined, participants gave a range of reasons although virtually all participants told us
that the main causes for declines in the fishery were due to an (unspecified) increase in the local
human population, resulting in overfishing, and that many villagers were now using modern fishing
gear, such as electric nets. Other less common reasons given for the declines were the increasing hot
weather, water pollution, and illegal fishing in restricted areas.
We prompted the villagers to discuss how they respond to the declines in the fishery, and how they
adapt to extreme weather such as drought and flood. Most people told us that the communities do
not respond or adapt at all. Due to the much-reduced catch, many families now need to supplement
their catch by buying fish to eat from local traders. However, often families cannot afford to buy
fish. A few participants reported that during drought they can go through periods where “people
almost have nothing to eat” as they have no alternate source of food or income. We were told that
during extreme conditions such as drought, even though people know there is little catch, most will
still try to undertake fishing activities. Few fisher people will change to alternative livelihood
activities. This was attributed to villagers not having any choice for alternative livelihood
generation in the floating villages. Some participants explained that families who can afford the
expense of buying food will leave the village for factory work in Thailand.
Discussion
The ongoing and projected changes in the Tonle Sap’s ecosystems are expected to increasingly
compromise the lake’s ecosystems (Arias et al., 2014). As such, it is very likely that the Tonle Sap
fisheries will become increasingly less productive. As a result, floating village communities will
suffer from significantly reduced livelihoods. This outcome clearly fails the requirements of
environmental distributive justice as the well-being outcomes of the developments driving these
changes are not equitably distributed to those most in need. Rather, our findings demonstrate that
floating village communities are the most likely people to suffer from ecosystem service declines
Page 100
100
in the Tonle Sap. We demonstrate that people believe there are rapid changes and declines in the
ecological system. Furthermore, we show how floating villagers believe that the entire fishery has
declined rapidly over the past two decades, a phenomenon which is likely true given the consistent
reports in interviews, and supporting empirical studies (Arias et al., 2014; Ratner et al., 2017).
Interestingly, our results show how the effects of climate change and hydrological alteration, such
as reductions in seasonally flooded habitats (Arias et al., 2012), are to some extent invisible to
floating villagers. Rather, they largely attribute fishery declines to overpopulation and the increased
use of illegal fishing gear.
Importantly, our research also demonstrates how people living in floating villages have very little
capacity, opportunity or means to adapt to these events and changes in situ. While some families
can seek international factory work, or trade fish, the majority of people describe themselves as too
poor to afford the capital required for such activities. We also found that in many cases, people may
not attempt to adapt at all, rather they simply continue fishing in much declined waters. This finding
is similar to previous survey based research in Cambodia, where over 40% of participants stated
they responded to severe drought by either organising religious ceremonies, or planting crops as
usual (MoE, 2005). This lack of adaptive capacity in floating villages is very likely due to economic
poverty (Keskinen, 2006). Poverty has been shown in general to make societies highly vulnerable to
changes in livelihoods (Shepherd and Brunt, 2013). For floating village communities, a distinct lack
of alternative livelihood options further exacerbates this vulnerability. The vulnerability of floating
villages is a serious concern as communities with little capacity to adapt to threats can fall into
poverty traps, which if left unresolved can cause catastrophic social breakdown (Carpenter and
Brock, 2008). As such, if the Tonle Sap’s ecosystems continue to degrade alongside the conditions
of poverty suffered by floating villagers, it is unclear if and how these communities will be able to
survive into the future.
While the overall adaptive capacity of floating villages is low, the protected areas near these
villages may play a role in buffering their vulnerability to change. While the role of the protected
areas in conserving biodiversity is obvious, the co-benefits to communities can be less so (Adams et
al., 2004). By maintaining healthy habitat the protected areas also provide important ecosystem
services to nearby communities (Middleton and Souter, 2016). In the Tonle Sap, protected areas
offer sanctuary, and spawning grounds for the fish on which the communities depend. Furthermore,
participants described an effect very similar to the fishery spill over seen around marine protected
areas (Russ and Alcala, 2011). In well managed marine reserves, species thrive in the absence of
Page 101
101
human pressure and as a result, fish can ‘spill over’ into the surrounding waters (Goñi et al., 2010).
However, the extent of such an effect in the Tonle Sap is poorly understood and should be the
subject of future research in this region. Regardless of the potential spill over, previous research
(and one participant who we interviewed) describe the protected areas as a burden on communities
(Gillespie, 2016). For local people, the complex nature of the over-lapping layers of protection and
regulation for the environment can be confusing. Gillespie (2016) reported that while many local
people value the services protected areas provide, their management was described as ‘clunky’ and
‘confusing’. This confusion is a concern, as local communities are those most affected (positively or
negatively) by the protected areas. This is also of concern from a biodiversity conservation
perspective, as a lack of buy-in from local communities have been shown to potentially undermine
the effectiveness of protected areas (Beger et al., 2004). As such, there is a need for an improved
understanding and communication of human and biodiversity conservation relationships in the
Tonle Sap. This presents an important research need which may identify opportunities for attaining
benefits for both people and nature in Cambodia.
Future studies assessing protected area spill over, or the human-nature relationships in the Tonle
Sap would benefit from improved data collection methods. In our study, the use of translation from
Khmer to English to conduct interviews is a time consuming and exhausting task which introduced
the potential for issues of miscommunication and misunderstanding. We therefore recommend that
native Khmer speakers undertake future interview based research with floating village communities.
While the communities and the biophysical nature of the Tonle Sap are highly unique, the lack of
distributive justice is not. Globally, these wetland ecosystems provide the majority of resources
required by local people to support their livelihoods and well-being. However, many (if not all)
major wetland systems are in a state of decline due to large scale anthropogenic changes, which
result in the unfair distribution of burdens to local communities. Recent examples include
subsistence or poor communities impacted by the Belo Monte dam in the Amazon Basin (Abers et
al., 2017), and the Grand Ethiopian Renaissance Dam on the Blue Nile River (Nalepa et al., 2017).
Even within the Mekong the situation in the Tonle Sap is not unique, with agrarian communities
throughout the Mekong suffering similar injustices (Tilt and Gerkey, 2016). As such, understanding
and addressing issues of distributive justice in wetland ecosystems is a problem of global concern.
The drivers of large scale environmental change in the world’s wetlands are happening at (regional
and global) scales which are beyond the ability of local communities to affect. As such, the
environmental degradation of wetlands is also an issue of social equity and distributive justice. The
people, businesses, and nations who are receiving the benefits from degrading these systems (i.e.
Page 102
102
gross domestic product growth from producing greenhouse gas emissions, or the sale of hydro-
power) are forcing others, such as the Tonle Sap floating village communities, to bear the costs.
However, there are efforts being made to address these inequities. The Mekong River Commission
(the inter-governmental organisation which manages the development of the Mekong River) state
they will make up for the losses and spread the benefits expected from hydropower development to
small communities via benefit sharing mechanisms (MRCS, 2014). Likewise, climate adaptation
funds, such as the Adaptation Fund, the Green Climate Fund, and the Special Climate Change Fund
aim to commit extensive resources to helping developing countries adapt to climate change.
However, there is no clear indication if (or how) any such funds may or may not be distributed to
specific communities (such as floating villages within the Tonle Sap). The Mekong River
Commission documentation related to benefit sharing offer little to no detail regarding beneficiaries
or actual distribution mechanisms, and this was reflected in our interviews with participants making
no mention of any such arrangements. Similarly, the climate adaptation funds offer little detail on
actual fund distribution mechanisms and are struggling to gain financial commitment from wealthy
nations (Kumar, 2015). As such, there is a critical need for clear and detailed mechanisms for
benefit sharing to Tonle Sap communities. These mechanisms urgently need to be formed and
implemented before livelihoods in the lake become further compromised. The parties driving
environmental degradation in the Tonle Sap have a clear moral obligation to share the benefits of
their activities with those bearing the costs. These issues also have implications at a broader scale
as these issues are hardly unique to the Tonle Sap. Previous research shows that inequitable impacts
of large-scale environmental degradation can be found the world over (Althor et al., 2016b), and
much like the case of the Tonle Sap, the main beneficiaries of environmental exploitation elsewhere
pay relatively few costs (O'Faircheallaigh, 2015; Ehrlich et al., 2012).
Conclusion
Globally, large scale environmental changes threaten the viability of subsistence communities.
Development along the Mekong basin will permanently change the ecological structure of the Tonle
Sap. Global climate change will increase temperatures within the lake, and increase the incidence of
extreme weather events. The development of dams on the Mekong River will affect the Tonle Sap’s
hydrology, leading to poor consequences for sustainability of the fishery. Furthermore, the effects
of Mekong basin development and climate change on the Tonle Sap are likely to intensify each
other. As a result, the lake’s ecosystems will become increasingly stressed, and some habitats such
as the lake’s seasonally flooded habitats will potentially disappear altogether. Permanent loss of
Page 103
103
habitat in the Tonle Sap could lead to a collapse of the lake’s (already) stressed fisheries. If the
lake’s fisheries fail, life in floating villages will become even more difficult, if not impossible. As
such, climate change and the development of the Mekong basin will dramatically, and potentially
catastrophically, affect these people. Furthermore, given the low adaptive capacity of these
communities, it is unclear how they might respond or adapt to a loss of their main livelihood source.
Efforts to secure a sustainable future are needed to correct the inequities of the manifold pressures –
climate change, Mekong basin development, fishery decline, population growth – affecting the
people of the Cambodian Tonle Sap Great Lake.
Page 104
104
Chapter 6 Thesis summary and conclusion
In this chapter I summarise how each research chapter of this thesis has overall contributed to how
distributive justice theories can be used to evaluate fairness for issues of natural capital distribution
while also producing rigorous scientific analyses of contemporary issues. In addition, I outline the
advantages and limitations, both broad and specific, of this approach, and provide suggestions for
how this work may be extended in the future.
Overall, throughout this thesis I have first argued then empirically demonstrated how distributive
justice principles can be adapted and modified to inform, develop, and understand empirical
research. More specifically, I have shown how research of this nature can be undertaken to examine
and understand natural capital distribution case studies through a justice lens.
A key foundation for understanding the quality and quantity of evidence regarding any particular
academic topic is the systematic review. I firmly believe that (high quality and rigorous) systematic
reviews are vital to gain an understanding of the whats, whens, wheres, and whos of particular topic
areas. Through these wonderful tools, we can systematically assess our fields of interest and
understand key unanswered questions, and how existing evidence paints the known truth of the
world. In chapter 2 I have demonstrated how a theoretical framework can be used alongside
systematic review methods for study design and to analyse peer reviewed data specific to
distributive justice and the environment. Moreover, I have shown how the results of this kind of
review can be used to identify knowledge gaps, which can in turn be used to guide the direction the
researcher takes. In my case, this review guided me toward the topics and cases investigated with
empirical methods in chapters 3, 4, and 5.
Issues of just or fair natural capital distribution can be empirically examined at the widest possible
scales, that is between nation states at the global level. While studies at this level do not allow for
the nuance of on-ground and inter-personal methods, they are nonetheless vital for the well-being of
many of the world’s people. Global issues which can and are affecting all of Earth’s human and
non-human inhabitants, generally require a global response. This response manifests in global
treaties, conventions, and other such political mechanisms. In chapter 3 I have shown how climate
change, perhaps the most concerning issue our species has faced in written history, is demonstrably
unjust. Essentially only a few of the world’s nations are driving the climate change phenomenon in
order to grow their economies, and in doing so are forcing unprecedented environmental change on
Page 105
105
the majority. This is clearly a repugnant situation, which must be quantified through studies such as
that which I undertook in order to inarguably bring such injustices to light. From this perspective, I
would argue that such studies can be seen as providing essential empirical tools for the burdened to
call out and argue against the unfair cost and benefit distributions of natural capital exploitation
during global negotiations such as the UN Climate Change Conferences.
Logically, distributive justice issues also occur at the national scale, and can be examined as such.
In the absence of any truly egalitarian society, most if not all nations are characterised by some
degree of unfair natural capital distribution. This is exemplified in chapter 2, the systematic review,
where I have shown that the vast majority of studies in this area have concluded that social
characteristics such as race and ethnicity often have a deterministic relationship with poor
environmental quality. Such work is critical, as it can give voice to issues faced by groups that often
also tend to have the least power and self-determination within a nation. Where there is data
available, there is a way to use distributive justice theories to critically examine these issues. In
chapter 4 I demonstrated how to examine the potential issue of unjust distributions of industrial
pollution across an entire nation, even within serious data quality constraints. As such, this method
can be replicated elsewhere to examine similar issues. The development and application of such
methods is critical to produce an evidence base upon which policy design and analysis can be
conducted, particularly where distributive justice is of concern. Furthermore, this chapter
demonstrates how important the ecological fallacy can be in studies based on spatial methods,
where disaggregating spatial units used in previous studies can lead to very different conclusions.
Additionally, cases of unjust or unfair natural capital distribution occur and can be examined at the
scale of impacts to individuals and their immediate communities. Such examinations are critical as
they convey the richness and nuance of real-world lived experiences that are missing from broad or
large scale quantitative analyses. While large scale analyses can quantify injustices, they are absent
the humanity of how real people experience, regard, and reflect on injustices. In the scientific realm,
it is only through these qualitative methods that we can gain some degree of understanding how, for
example, a person experiences living next to heavily polluting industry. Or, as I have examined in
chapter 5, how subsistence fisher folk in rural Cambodia are experiencing the literal evaporation of
their primary source of livelihood so that others, elsewhere in the world, might power their homes,
or contribute to climate change through unchecked consumerism and economic growth.
Page 106
106
I believe that this thesis adds an important, and at times absent, moral dimension to rigorous,
empirical studies, which in turn adds weight to their importance. It is my intention that this thesis
might be used by other researchers to inform the theory underlying similar studies, and to also
provide examples for how distributive justice theories can be utilised in empirical application. I
argue that this gives researchers a clear framework for discussing the results of such cases and
provides a baseline for drawing conclusions. As demonstrated, this process of theory adaptation and
study design can be applied to a wide range of problems, data types, and research methods.
Also importantly, throughout this thesis I have shown that it is not only appropriate for academic
research to be transparently normative in its execution, but also that this is unavoidable and as such
ought to be done in a transparent manner, which makes the researcher’s positionality obvious to the
reader. I believe that this is a critical for the interpretability of empirical research, and ought to be
undertaken when and where possible.
Modifying distributive justice principles
A key idea driving this thesis was my desire to transparently utilise philosophical theories in
combination with empirical research to evaluate real-world natural capital distribution issues. Very
early on, I came across the work on distributive justice by John Rawls (Rawls, 1971). His principles
fit very comfortably within the natural capital distribution context in which I was working. In
particular, I saw a lot of value in utilising part b of Rawls’ second principle of distributive justice:
“Social and economic inequalities are to satisfy two conditions: (a)… ; and (b), they are to be to
the greatest benefit of the least advantaged members of society”
As demonstrated in this thesis, and as argued elsewhere, the pragmatism and practicality of this
principle is well-suited to evaluating the way that natural capital goods and services are distributed
across different groups of people. Additionally, I have demonstrated the potential for the principle
of strict egalitarianism to be useful in such settings:
Every person should have the same level of material goods (including burdens) and services.
This principle, though, is highly unpragmatic when applied to general concepts of natural capital
distributions. As such, I have modified the principle to capture the idea that every person is of equal
Page 107
107
moral worth and that harm to persons is universally immoral. Therefore, if we are to use natural
capital to benefit society in a manner which generates harm as a tolerable by-product of that benefit,
then harm ought strictly to be distributed equally across all beneficiaries.
To apply the work of Rawls and incorporate the principle of strict egalitarian in the context of harm,
I elected to modify these principles into distinct forms that are explicit in their relationship to the
distribution of natural capital goods and services, as follows:
P1: Social and economic inequalities, resulting from the utilisation of natural capital, are to be of
the greatest benefit of the least advantaged members of society.
P2: Where harm is accepted as a tolerable by-product of natural capital use for the betterment of
society, exposure levels of harm ought to be equal among all persons.
P1 and P2 can be utilised separately, or combined, as follows:
P3: Where burdens upon any individual or exclusive group are the byproduct of the use of natural
capital for the betterment of another individual or exclusive group, burdens are to be accompanied
by (at least) commensurate benefits.
These three principles make up the underlying theoretical approaches used throughout the thesis.
They were applied separately and in combination across research chapters.
Contribution of each research chapter
The following is a chapter by chapter breakdown of each study’s key research question and sub-
questions, a summary of how they utilised distributive justice principles, and finally the key
outcomes of each.
Chapter 2: A quantitative systematic review of distributive environmental justice literature; a rich
history, and the need for an enterprising future
Page 108
108
Research question: What is the depth and breadth of the distributive justice and natural capital
literature?
Sub-questions:
1) Where are the research institutions of the authors of this research located;
2) What has been the geographic focus of the literature;
3) What has been the demographic, environmental, and human well-being scope of the literature;
4) What types of publications and study methods have been used; and
5) What are the gaps and areas for future focus in the literature.
In this Chapter I utilised all three modified principles (P1, P2, and P3) within a systematic review of
the distributive justice and natural capital literature. I used the principles to develop the study aims,
and methods such as the search strings, and inclusion/exclusion criteria. Specifically, P1, P2, and P3
were all used to develop the scope of the literature review, and were critical for the structure of the
introduction and arguing for the purpose and importance of the study. P1 and P2 were used to
develop the initial search string by using keywords such as “social inequality”, and “natural capital”
in the initial testing phase. Furthermore, I utilised P1 and P2 to develop the typologies used in the
review to ensure that all data I was capturing were related to the thesis principles.
Key outcome 1
With this study I provided a rigorously developed database of the peer-reviewed literature which
relates to distributive justice and natural capital. This database can be used to identify gaps in the
literature and to quickly and easily identify literature using cross tabulated sub-categories.
Key outcome 2
This literature is rapidly growing, and somewhat diverse, but is dominated by particular
geographies, methods, demographics, livelihood indicators, and topics. I recommend that a more
comprehensive approach is taken in the future, and suggest that the tabulated results of this review
can be used to identify these kinds of gaps.
Key outcome 3
Studies in this area are dominated by findings of distributive inequities (90 percent plus). However,
it is unclear if this is the true nature of natural capital distributions, or a result of biases such as
researcher and design bias, or the general bias in all sciences against the publication of negative
Page 109
109
results. Based on this finding I recommend that future studies in this area apply the highest degree
of rigour possible, particularly in terms of controls.
Chapter 3: Global mismatch between greenhouse gas emissions and the burden of climate change
Research question: How are greenhouse gas emissions and the burdens of climate change
distributed globally?
Sub-questions:
1) What is the extent of global climate inequity; and
2) How is this expected to change in the near future.
In this Chapter I utilised my third modified distributive justice principle (P3) to empirically describe
and evaluate how the benefits and burdens associated with climate change are distributed.
Specifically, I utilised this principle in the design and results interpretation of this study. P3 posits
that cases where groups or individuals might be burdened by the use of natural capital by other
groups, these burdens ought to be accompanied by commensurate benefits (at a minimum). This is
clearly applicable to global climate change, where some few nations are improving their nation
economies through greenhouse gas emitting industries, while others are suffering from resulting
environmental burdens with few to no benefits.
Key outcome 1
Climate change inequity is globally pervasive, and correlated with economic output. The world’s
highest GHG emitting nations are driving climate change in order to grow their national economies,
but are (relatively) the least vulnerable (partially due to their large economies) to the negative
impacts of climate change. This is inversely true for low emitting nations, who are benefiting very
little from the processes driving climate change, but are inordinately suffering from their effects.
Key outcome 2
Climate change inequity is expected to worsen in the future. I project that by 2030 the distributions
of benefits and burdens will continue along a trajectory of inequity.
Page 110
110
Key outcome 3
Global policy frameworks have been developed to address these kinds of inequities, but are grossly
inadequate, underfunded, and lack enforceability. As such, there are few incentives for the highest
emitting nations to curtail activities.
Chapter 4: The distribution of industrial pollution across socio-economic groups in Australia; a ten-
year, a fine scale distributive justice analysis
Research question: Are there social inequities in the distribution of exposure to harmful industrial
pollution across Australia?
Sub-questions:
1) Have these distributions changed over the last 10 years; and
2) If any, which social or economic groups are more likely to be exposed to industrial pollution.
In this Chapter I utilised my second modified distributive justice principle (P2) to empirically
describe and evaluate how harm from industrial pollution is distributed across social demographics
Australia wide. Specifically, I used P2 in the design and results interpretation of this study. P2
argues that cases where harm is accepted as a by-product of using natural capital to improve
society, then harms ought to be distributed equally. This principle is less generalisable than P1, but
does have a specific purpose for use in cases of harm distribution, such as the distribution of
industrial pollution. Industrial pollution (within regulated thresholds) is accepted as a by-product of
economic activities that is known to be harmful. As such, it provides a prime model for how P2 can
be used to examine the distribution of harm across social and demographic groups.
Key outcome 1
My results show there is little evidence aligned with previous studies which claimed that exposure
to industrial pollution has a clear relationship with socio-economic variables in aggregate. I found
some statistically significant relationships, however the effect sizes were too small to attribute any
causal relationship. I argue that this may be a result of overly large spatial units being used in
previous studies, and improperly applied toxicity weightings.
Page 111
111
Key outcome 2
The findings may indicate that Indigenous Australians are the most at risk of exposure to industrial
pollution, but further research needs to clarify this relationship.
Chapter 5: Large-scale environmental degradation results in inequitable impacts to already
impoverished communities: A case study from the floating villages of Cambodia
Research question: How are the fishing communities of the Tonle Sap floating villages experiencing
major environmental changes?
Sub-questions:
1) How are floating village communities experiencing a change in fisheries; and
2) How are floating village communities responding to these changes.
In this chapter I utilised my first modified distributive justice principle (P1) using qualitative
research methods to empirically describe and evaluate the lived experiences of Cambodian
subsistence fisher people, in regards to how upstream environmental degradation is affecting their
well-being. Specifically, I used P1 in the design and results interpretation of this study. P1 argues
that social and economic inequalities, resulting from the utilisation of natural capital, are to be of
the greatest benefit of the least advantaged members of society. In this case there are gross
inequalities experienced by Tonle Sap fishing communities, resulting from Mekong damming and
climate change, however these communities are receiving little to no benefit from these activities.
As such, this chapter is exemplary of how to examine distributions of costs and benefits resulting
from natural capital exploitation.
Key outcome 1
I demonstrate that people living within the Tonle Sap Great Lake believe there are rapid changes
and declines in the ecological system, aligning with existing ecological and environmental science
research.
Key outcome 2
However, my results also show how the effects of climate change and hydrological alteration, such
as reductions in seasonally flooded habitats, are to some extent invisible to floating villagers.
Page 112
112
Instead, they largely attribute fishery declines to overpopulation and the increased use of illegal
fishing gear.
Key outcome 3
I also demonstrate how people living in floating villages have very little capacity, opportunity or
means to adapt to a rapidly changing environment. For floating village communities, a distinct lack
of alternative livelihood options exacerbates their vulnerability to these changes. In most cases,
people are unlikely to attempt to adapt, but rather they will simply continue fishing in much
declined waters.
Advantages and limitations of the distributive justice approach, and future
research
This section contains the conclusions I have reached throughout my PhD about the advantages and
limitations of the distributive justice approach I have used (in its modified form), and future
research needs. As each research chapter discusses its own specific limitations and future research
needs, I have not replicated these here. Rather, this section pertains to the overall advantages,
limitations, and future research.
Advantages and limitations
The distributive justice approach provides a clear, practical, and easy to understand philosophical
framework with which to evaluate empirical research findings related to the fairness in how natural
capital is distributed. This is important for researchers interested in distributive justice issues, as it
not only provides a useable framework, it also allows them to be transparent in acknowledging their
positionality, and how this may affect their decisions and interpretations of empirical findings.
Furthermore, distributive justice studies have the advantage of simplicity, although this is also a
limitation (as outlined below). As I have demonstrated throughout this thesis, distributive justice
evaluations and their underlying moral arguments are intuitive, easy to understand, and
communicate. For example, practically anyone intuitively understands that it is unfair for some
nations to suffer from a global climate crisis they contributed very little to creating. The value of
this is manifold, not only does it provide a formalised and rational argument for intuitive ideas, by
reinforcing intuitive arguments it can be used to produce outputs including graphical figures that are
Page 113
113
easily interpretable by a wide range of audiences, importantly including policy makers and the
general public.
Another key advantage is its scalability across geographies. With Chapters 3, 4, and 5. I have
demonstrated how these principles can be applied at a range of vastly different scales. Between
these chapters, I show how distributive justice principles can be applied anywhere from and
between the global scale (in Chapter 3), to the national scale (in Chapter 4), and all the way down to
evaluating the lived experiences of individuals in a local community setting (Chapter 5). Moreover,
this approach is highly useful as a guide in selecting and developing research methods. The
dimensions of distributive justice, such as costs and benefits, can be used by researchers and
practitioners in their selection of appropriate research methods. Throughout this thesis, I have
demonstrated the use of distributive justice principles in methodological design across a wide range
of method types. In Chapter 2 I used these principles to develop screening criteria for a systematic
literature review, in Chapters 3 and 4 I used them to develop statistical methods and modelling to
interpret quantitative data, and in Chapter 5 I demonstrated how they can be used to guide the
design and interpretation of qualitative research.
The most critical limitation of singularly distributive justice studies is their simplicity (although also
an advantage as discussed above), and lack of multidimensional justice perspectives. As discussed
throughout, due to this simplicity I caution against uses of distributive justice that treat it as an all-
encompassing evaluation of justice. There are many complex forms of justice which interact and
overlap to form a complete picture of what might universally be considered just or unjust.
Distributive justice is of most value and use if it is combined with other approaches to justice,
including but not limited to procedural, relational, restorative, or legal. While these approaches
were beyond the scope of the work contained in this thesis, it is critical that I acknowledge their
value alongside distributive approaches. More specific to this thesis, this work is also limited in the
scope of distributive justice forms examined. By basing this thesis almost solely on the principles
created by John Rawls, I have not rigorously examined all forms of distributive justice itself. While
Rawls is certainly the most prolific writer in this space, he is by no means the only writer. There are
several other forms of distributive justice which would likely have been just as legitimate for
developing my evaluations – e.g. the luck egalitarian approaches of Dworkin, Arneson, and Cohen,
or Nozick’s entitlement theory (Allingham, 2014).
Page 114
114
Future research
As is likely obvious, this thesis is limited in its ability to holistically capture the ideas that it has
touched on. As discussed, the forms of justice examined herein are very limited, and are by no
means comprehensive. For future work undertaken in this in this area, rather than examine many
cases at different scales and using different methods, there would be great benefit in critically
examine a single case from a multi-dimensional and holistic justice approach.
As it stands, I have made many claims for what is fair, unfair, just, or unjust throughout, but they
are singularly limited to a distributive perspective, and as such cannot be called universally just or
not. Furthermore, and based on the findings of the systematic review outlined in Chapter 2,
distributive justice approaches clearly need to be more diverse, and more inclusive. For example, it
would be highly desirable to see more research dedicated to Indigenous people around the world,
LGBTQI+ people, and other under-represented minorities. Lastly, while I have argued that
distributive justice lends itself to an ease of understanding and real-world applicability (via its
simplicity), I have not empirically demonstrated this idea. It would be excellent for future work in
this area to address this issue by directly engaging with policy makers or science communicators
about problems they are facing, and work with them to see if this approach is truly as useful as I
have hypothesized.
Concluding remarks
I opened this thesis noting my own positionality as a researcher and reasserting the claim that every
human being has equal moral worth and that we must attach value to the well-being of each and
every person. My lived experiences, disciplinary background and training in environmental
management, and belief in equality guided my PhD research toward the questions of distributive
justice and natural capital examined in this thesis.
I modified principles of distributive justice and examined these across a range of studies and scales
via a diversity of methods. Through this I have come to learn, and my thesis has shown, that
distributive justice is flexible and practical for use in analysing empirical research related to the way
in which natural capital is distributed. I did not set out to capture a holistic perspective on justice as
related to natural capital, but instead sought a pragmatic and practical way to look at how benefits
and burdens are shared, and critically, the impact this has on the well-being of people.
Based on these studies, I cannot assert that the world is fair, nor is it unfair. Each study explored a
specific case of natural capital distributions and their social impacts. The nuances of each of these
Page 115
115
studies are shaped by their own social and geographical contexts. Rather, and unsurprisingly, the
world is a complex place that cannot be universally essentialised. However, through this thesis I
have demonstrated that there is a practical value in applying singular dimensions of justice when
examining real world cases. Furthermore, I have met my personal goal of ensuring my work
transparently and honestly reflects my positionality as a researcher – a person who believes in a
universal moral framework for the equality of rights and fair treatment of all people.
Page 116
116
Bibliography
Abers RN, Oliveira MSd and Pereira AK. (2017) Inclusive Development and the Asymmetric State:
Big Projects and Local Communities in the Brazilian Amazon. The Journal of Development
Studies 53: 857-872.
ABS. (2011) 1259.0.30.002 - Statistical Geography - Australian Standard Geographical
Classification (ASGC), Digital Boundaries, 2006. Available at:
http://www.abs.gov.au/AUSSTATS/[email protected] /Lookup/1259.0.30.002Main+Features12006?
OpenDocument.
ABS. (2017a) 1345.0 - Key Economic Indicators, 2018 Available at:
http://www.abs.gov.au/AUSSTATS/[email protected] /mf/1345.0.
ABS. (2017b) Australian Statistical Geography Standard (ASGS): Volume 1 - Main Structure and
Greater Capital City Statistical Areas. Available at:
http://www.abs.gov.au/AUSSTATS/[email protected] /DetailsPage/1270.0.55.001July%202016?Ope
nDocument.
ABS. (2018a) Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA),
Australia, 2011. Available at:
http://www.abs.gov.au/ausstats/[email protected] /DetailsPage/2033.0.55.0012011?OpenDocument.
ABS. (2018b) TableBuilder. Available at:
http://www.abs.gov.au/websitedbs/D3310114.nsf/Home/2016%20TableBuilder.
Adams WM, Aveling R, Brockington D, et al. (2004) Biodiversity conservation and the eradication
of poverty. Science 306: 1146-1149.
Adger WN. (2002) Inequality, Environment, and Planning. Environment and Planning A 34: 1716-
1719.
Ahnlid A. (1992) Free or Forced Riders?: Small States in the International Political Economy: The
Example of Sweden. Cooperation and Conflict 27: 241-276.
Al-Obaid S, Samraoui B, Thomas J, et al. (2017) An overview of wetlands of Saudi Arabia: Values,
threats, and perspectives. Ambio 46: 98-108.
Allingham M. (2014) Distributive Justice, Florence: Taylor and Francis.
Almaskut A, Farrell PJ and Krewski D. (2012) Statistical methods for estimating the environmental
burden of disease in Canada, with applications to mortality from fine particulate matter.
Environmetrics 23: 329-344.
Althor G, Watson J and Fuller R. (2016a) Global mismatch between greenhouse gas emissions and
the burden of climate change. Scientific Reports 6.
Page 117
117
Althor G, Watson JEM and Fuller RA. (2016b) Global mismatch between greenhouse gas
emissions and the burden of climate change. Scientific Reports 6: 20281.
Anderton DL, Anderson AB, Rossi PH, et al. (1994) Hazardous Waste Facilities:"Environmental
Equity" Issues in Metropolitan Areas. Evaluation Review 18: 123-140.
Arias ME, Cochrane TA and Elliott V. (2014) Modelling future changes of habitat and fauna in the
Tonle Sap wetland of the Mekong. Environmental Conservation 41: 165-175.
Arias ME, Cochrane TA, Norton D, et al. (2013) The flood pulse as the underlying driver of
vegetation in the largest wetland and fishery of the Mekong Basin. Ambio 42: 864-876.
Arias ME, Cochrane TA, Piman T, et al. (2012) Quantifying changes in flooding and habitats in the
Tonle Sap Lake (Cambodia) caused by water infrastructure development and climate change
in the Mekong Basin. J Environ Manage 112: 53-66.
Austin KF and McKinney LA. (2016) Disaster Devastation in Poor Nations: The Direct and Indirect
Effects of Gender Equality, Ecological Losses, and Development. Social Forces 95: 355-
380.
Australian Government. (2015) National Clean Air Agreement. In: Energy DoEa (ed). online:
Department of Environment and Energy.
Ayres CJ. (2012) The international trade in conflict minerals: coltan. Critical perspectives on
international business 8: 178-193.
Babidge S. (2016) Contested value and an ethics of resources: Water, mining and indigenous people
in the Atacama Desert, Chile. Australian Journal of Anthropology 27: 84-103.
Bagchi S and Svejnar J. (2015) Does wealth inequality matter for growth? The effect of billionaire
wealth, income distribution, and poverty. Journal of Comparative Economics 43: 505-530.
Banzhaf S, Ma L and Timmins C. (2019) Environmental Justice: The Economics of Race, Place,
and Pollution. The Journal of Economic Perspectives 33: 185-208.
Baron JS, Poff NL, Angermeier PL, et al. (2002) Meeting ecological and societal needs for
freshwater. Ecological Applications 12: 1247-1260.
Beger M, Harborne AR, Dacles TP, et al. (2004) A framework of lessons learned from community-
based marine reserves and its effectiveness in guiding a new coastal management initiative
in the Philippines. Environ Manage 34: 786-801.
Berbés-Blázquez M, González JA and Pascual U. (2016) Towards an ecosystem services approach
that addresses social power relations. Curr Opin Environ Sustain 19.
Bernard HR and Gravlee CC. (2014) Handbook of methods in cultural anthropology: Rowman &
Littlefield.
Betts R. (2008) Comparing apples with oranges. 7-8.
Page 118
118
Borkan JM. (2004) Mixed Methods Studies: A Foundation for Primary Care Research. Annals of
Family Medicine 2: 4-6.
Bournay E. (2017) Typology of Hazards. Available at: http://www.grida.no/resources/7805.
Bowen W. (2002) An Analytical Review of Environmental Justice Research: What Do We Really
Know? Environmental Management 29: 3-15.
Bradley B. (2006) Two Concepts of Intrinsic Value. Ethical Theory and Moral Practice 9: 111-130.
Braun V and Clarke V. (2006) Using thematic analysis in psychology. Qualitative Research in
Psychology 3: 77-101.
Briggs RC and Weathers S. (2016) Gender and location in African politics scholarship: The other
white man's burden? African Affairs 115: 466-489.
Broome RA, Fann N, Cristina TJN, et al. (2015) The health benefits of reducing air pollution in
Sydney, Australia. Environmental Research 143: 19-25.
Brulle RJ and Pellow DN. (2006) Environmental justice: human health and environmental
inequalities. Annu. Rev. Public Health 27: 103-124.
Brunekreef B and Hoffmann B. (2016) Air pollution and heart disease. The Lancet 388: 640-642.
Bryman A. (2012) Social research methods, Oxford: Oxford university press.
Burkholder I and Edler L. (2014) Linear Model: Overview. Wiley StatsRef: Statistics Reference
Online: John Wiley & Sons, Ltd.
Campbell IC, Poole C, Giesen W, et al. (2006) Species diversity and ecology of Tonle Sap Great
Lake, Cambodia. Aquatic Sciences 68: 355-373.
Caney S. (2014) Climate change, intergenerational equity and the social discount rate. Politics
Philosophy & Economics 13: 320-342.
Carney N. (2016) All Lives Matter, but so Does Race: Black Lives Matter and the Evolving Role of
Social Media. Humanity & Society 40: 180-199.
Carpenter SR and Brock WA. (2008) Adaptive capacity and traps. Ecology and Society 13: online.
Cazorla M and Toman M. (2001) International Equity and Climate Change Policy. Climate Change
Economics and Policy: An RFF Anthology. Washington DC: RFF Press, 235.
CDRI. (2013) Climate Change Adaptation and Livelihoods in Inclusive Growth: A Review of
Climate Change Impacts and Adaptive Capacity in Cambodia. In: CDRI (ed). Phnom Penh:
CDRI.
CEE. (2013) Guidelines for Systematic Reviews in Environmental Management, Bangor, UK:
Collaboration for Environmental Evidence.
Chakraborty J and Green D. (2014) Australia’s first national level quantitative environmental
justice assessment of industrial air pollution. Environmental Research Letters 9: 044010.
Page 119
119
Charafeddine R and Boden LI. (2008) Does income inequality modify the association between air
pollution and health? Environmental Research 106: 81-88.
Chea R, Guo C, Grenouillet G, et al. (2016) Toward an ecological understanding of a flood-pulse
system lake in a tropical ecosystem: Food web structure and ecosystem health. Ecological
Modelling 323: 1-11.
Clark VLP and Ivankova NV. (2015) Mixed methods research: A guide to the field: Sage
Publications.
Cole DH. (2015) Advantages of a polycentric approach to climate change policy. Nature Clim.
Change 5: 114-118.
Cook KS and Hegtvedt KA. (1983) Distributive justice, equity, and equality. Ann Rev Sociol 9.
Cooper N, Green D and Meissner JK. (2017) The Australian National Pollutant Inventory Fails to
Fulfil Its Legislated Goals. International Journal of Environmental Research and Public
Health 14.
Cooper N, Green D, Sullivan M, et al. (2018) Environmental justice analyses may hide inequalities
in Indigenous people’s exposure to lead in Mount Isa, Queensland. Environmental Research
Letters 13: 084004.
Cooperman MS, So N, Arias M, et al. (2012) A watershed moment for the Mekong: newly
announced community use and conservation areas for the Tonle Sap Lake may boost
sustainability of the world’s largest inland fishery. Cambodian Journal of Natural History
2012: 101-106.
Copeland DC. (1996) Economic Interdependence and War: A Theory of Trade Expectations.
International Security 20: 5-41.
Coulter PB. (1989a) Measuring inequality: a methodological handbook, Boulder: Westview Press.
Coulter PB. (1989b) Measuring inequality: a methodological handbook, Boulder: Westview Press.
Crase L and Gillespie R. (2008) The impact of water quality and water level on the recreation
values of Lake Hume. Australasian Journal of Environmental Management 15: 21-29.
Crepaz MML. (1995) Explaining national variations of air pollution levels: Political institutions and
their impact on environmental policy‐making. Environmental Politics 4: 391-414.
Creswell JW. (2014) A concise introduction to mixed methods research: SAGE publications.
Crick B. (1987) Socialism, Milton Keynes: Open University Press.
Croissant Y and Zeileis A. (2017) truncreg: Truncated Gaussian Regression Models. Available at:
http://r-forge.r-project.org/projects/truncreg/, Version = 0.2-4.
Page 120
120
Cummings S and Hoebink P. (2017) Representation of Academics from Developing Countries as
Authors and Editorial Board Members in Scientific Journals: Does this Matter to the Field
of Development Studies? The European Journal of Development Research 29: 369-383.
Cunningham SC, Thomson JR, Mac Nally R, et al. (2011) Groundwater change forecasts
widespread forest dieback across an extensive floodplain system. Freshwater Biology 56:
1494-1508.
Cushing L, Faust J, August LM, et al. (2015) Racial/Ethnic Disparities in Cumulative
Environmental Health Impacts in California: Evidence From a Statewide Environmental
Justice Screening Tool (CalEnviroScreen 1.1). Am J Public Health 105: 2341-2348.
DARA. (2012) Methodological Documentation for the Climate Vulnerability Monitor, Madrid:
DARA.
Davis AY. (1981) Women, race & class: New York : Vintage eBooks.
Delpla I, Benmarhnia T, Lebel A, et al. (2015) Investigating social inequalities in exposure to
drinking water contaminants in rural areas. Environmental Pollution 207: 88-96.
Dobbie B and Green D. (2015) Australians are not equally protected from industrial air pollution.
Environmental Research Letters 10.
Drabo A. (2011) Impact of Income Inequality on Health: Does Environment Quality Matter?
Environment and Planning A 43: 146-165.
Ehrlich PR, Kareiva PM and Daily GC. (2012) Securing natural capital and expanding equity to
rescale civilization. Nature 486: 68-73.
Erwin KL. (2009) Wetlands and global climate change: the role of wetland restoration in a changing
world. Wetlands Ecology and management 17: 71.
ESRI. (2011) ArcGIS Desktop: Release 10.5. Redlands, CA: Environmental Systems Research
Institute.
ESRI ArcGIS. (2011) Release 10. Redlands, CA: Environmental Systems Research Institute.
ESRI ArcGIS Pro. (2017) 1.4.1. Redlands, CA: Environmental Systems Research Institute.
Essoka JD. (2010) The Gentrifying Effects of Brownfields Redevelopment. Western Journal of
Black Studies 34: 299-315.
European Environment Agency. (2007) Europe's environment - The fourth assessment.
Copenhagen: European Environment Agency.
FAO. (2001) Tonle Sap Fisheries: A Case Study on Floodplain Gillnet Fisheries. In: Lamberts D
(ed). online: FAO Regional Office for Asia and the Pacific.
FAO. (2017) Fishery and Aquaculture Country Profiles: The Kingdom of Cambodia. Available at:
http://www.fao.org/fishery/facp/KHM/en.
Page 121
121
Fecht D, Fischer P, Fortunato L, et al. (2015) Associations between air pollution and socioeconomic
characteristics, ethnicity and age profile of neighbourhoods in England and the Netherlands.
Environmental Pollution 198: 201-210.
Fekete H, Vieweg M, Rocha M, et al. (2013) Analysis of Current Greenhouse Gas Emission
Trends, Berlin: Climate Analytics.
Ferraro PJ. (2009) Counterfactual thinking and impact evaluation in environmental policy. New
Directions for Evaluation 2009: 75-84.
Findlay R and O’Rourke K. (2012) War, Trade, and Natural Resources: A Historical Perspective.
In: Garfinkel M and Skaperdas S (eds) The Oxford Handbook of the Economics of Peace
and Conflict. Oxford: Oxford University Press.
Fineman MA. (2014) Vulnerability, Resilience, and LGBT Youth. Temple Political & Civil Rights
Law Review 23: 307-330.
Fisher JA, Patenaude G, Giri K, et al. (2014) Understanding the relationships between ecosystem
services and poverty alleviation: A conceptual framework. Ecosystem Services 7: 34-45.
Francis JJ, Johnston M, Robertson C, et al. (2010) What is an adequate sample size?
Operationalising data saturation for theory-based interview studies. Psychology and Health
25: 1229-1245.
Freischlag JA and Faria P. (2018) It Is Time for Women (and Men) to Be Brave: A Consequence of
the #MeToo MovementPutting an End to Sexual Harassment in Health CarePutting an End
to Sexual Harassment in Health Care. JAMA 319: 1761-1762.
Frith H and Gleeson K. (2004) Clothing and embodiment: Men managing body image and
appearance. Psychology of men and masculinity 5: 40-48.
Füssel H-M. (2010) How inequitable is the global distribution of responsibility, capability, and
vulnerability to climate change: A comprehensive indicator-based assessment. Global
Environmental Change 20: 597-611.
Gerlagh R and Keyzer MA. (2001) Sustainability and the intergenerational distribution of natural
resource entitlements. Journal of Public Economics 79: 315-341.
Germani AR, Morone P and Testa G. (2014) Environmental justice and air pollution: A case study
on Italian provinces. Ecological Economics 106: 69-82.
Gichere SK, Olado G, Anyona DN, et al. (2013) Effects of drought and floods on crop and animal
losses and socio-economic status of households in the Lake Victoria Basin of Kenya.
Journal of Emerging Trends in Economics and Management Sciences 4: 31.
Gillespie J. (2016) Catch 22: wetlands protection and fishing for survival. Geographical Research
54: 336-347.
Page 122
122
Gilmartin M. (2009) Colonialism/Imperialism Key Concepts in Political Geography. London:
SAGE Publications Ltd.
Goñi R, Hilborn R, Díaz D, et al. (2010) Net contribution of spillover from a marine reserve to
fishery catches. Marine Ecology Progress Series 400: 233-243.
Green Climate Fund. (2015) Background.
GRID Arendal. (2017) Typology of Hazards. Available at: www.grida.no/resources/7805.
Griffith University. (2019) Systematic Quantitative Literature Review. Available at:
https://www.griffith.edu.au/griffith-sciences/school-environment-
science/research/systematic-quantitative-literature-review.
Grim BJ and Finke R. (2007) Religious Persecution in Cross-National Context: Clashing
Civilizations or Regulated Religious Economies? American Sociological Review 72: 633-
658.
Gross C. (2008) A measure of fairness: An investigative framework to explore perceptions of
fairness and justice in real-life social conflict. Human Ecology Review 15: 130-140.
Guest G, Bunce A and Johnson L. (2006) How Many Interviews Are Enough?: An Experiment with
Data Saturation and Variability. Field Methods 18: 59-82.
Guest G, MacQueen K and Namey E. (2012) Themes and Codes. Applied Thematic Analysis.,
Thousand Oaks, CA: SAGE Publications, Inc.
Gusenbauer M and Haddaway N. (2019) Which Academic Search Systems are Suitable for
Systematic Reviews or Meta-Analyses? Evaluating Retrieval Qualities of Google Scholar,
PubMed and 26 other Resources. Research Synthesis Methods.
Hanigan IC, Cochrane T and Davey R. (2017) Impact of scale of aggregation on associations of
cardiovascular hospitalization and socio-economic disadvantage. PLoS ONE 12: e0188161.
Harrison K. (2012) A Tale of Two Taxes: The Fate of Environmental Tax Reform in Canada.
Review of Policy Research 29: 383-407.
Hayek F. (1976) Law, Legislation and Liberty, Volume 2: The Mirage of Social Justice, Chicago:
The University of Chicago Press.
Heitjan DF and Roderick JAL. (1991) Multiple Imputation for the Fatal Accident Reporting
System. Journal of the Royal Statistical Society. Series C (Applied Statistics) 40: 13-29.
Hertwich EG, Mateles SF, Pease WS, et al. (2001) Human toxicity potentials for life-cycle
assessment and toxics release inventory risk screening. Environ Toxicol Chem 20: 928-939.
Heyward M. (2007) Equity and international climate change negotiations: a matter of perspective.
Climate Policy 7: 518-534.
Page 123
123
Higginbotham N, Freeman S, Connor L, et al. (2010) Environmental injustice and air pollution in
coal affected communities, Hunter Valley, Australia. Health & Place 16: 259-266.
Holnicki P, Tainio M, Kałuszko A, et al. (2017) Burden of Mortality and Disease Attributable to
Multiple Air Pollutants in Warsaw, Poland. International Journal of Environmental
Research and Public Health 14: 1359.
Howes M. (2001) What's Your Poison? The Australian National Pollutant Inventory versus the US
Toxics Release Inventory. Australian Journal of Political Science 36: 529-552.
Huang R. (2014) RQDA: R-based Qualitative Data Analysis. 0.2-7 ed.
IPCC. (2007) Climate Change 2007: Impacts, Adaptation and Vulnerability : Working Group ii
Contribution to the Fourth Assessment Report of the Ipcc [M.L. Parry, O.F. Canziani, J.P.
Palutikof, P.J. Van Der Linden And C.E. Hanson (Eds)], Cambridge: Cambridge University
Press.
IPCC. (2014a) Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and
Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change [Field, C.B., V.R. Barros, D.J. Dokken, K.J.
Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatterjee, K.L. Ebi, Y.O. Estrada, R.C. Genova,
B. Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, and L.L. White (eds.)],
Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press.
Ipcc. (2014b) Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part B: Regional
Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change [Barros, V.R., C.B. Field, D.J. Dokken, M.D.
Mastrandrea, K.J. Mach, T.E. Bilir, M. Chatterjee, K.L. Ebi, Y.O. Estrada, R.C. Genova, B.
Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, and L.L. White (eds.)],
Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press.
Jackson RB, Canadell JG, Le Quere C, et al. (2015) Reaching peak emissions. Available at:
http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate2892.html.
Jorgenson AK. (2014) Economic development and the carbon intensity of human well-being.
Nature Clim. Change 4: 186-189.
Kampa M and Castanas E. (2008) Human health effects of air pollution. Environmental Pollution
151: 362-367.
Kennedy M and Basu B. (2014) An analysis of the climate change architecture. Renewable and
Sustainable Energy Reviews 34: 185-193.
Kenny DC, Costanza R, Dowsley T, et al. (2019) Australia's Genuine Progress Indicator Revisited
(1962–2013). Ecological Economics 158: 1-10.
Page 124
124
Keohane RO. (2015) The Global Politics of Climate Change: Challenge for Political Science. PS:
Political Science & Politics 48: 19-26.
Keskinen M. (2006) The Lake with Floating Villages: Socio-economic Analysis of the Tonle Sap
Lake. International Journal of Water Resources Development 22: 463-480.
Keskinen M, Kummu M, Salmivaara A, et al. (2011) Baseline results from hydrological and
livelihood analyses, Exploring Tonle Sap Futures. online: Aalto University.
Keywood M, Emmerson K and Hibberd M. (2016) Atmosphere: Australia state of the environment
2016. In: Energy AGDotEa (ed). Canberra: Australian Government Department of the
Environment and Energy.
Keywood M, Hibberd M and Emmerson K. (2017) Australia state of the environment 2016:
atmosphere, independent report to the Australian Government Minister for the Environment
and Energy. online: Commonwealth of Australia.
Kitzinger C and Willmott J. (2002) 'The thief of womanhood': women's experience of polycystic
ovarian syndrome. Soc Sci Med 54: 349-361.
Kjellstrom T, Kovats RS, Lloyd SJ, et al. (2009) The Direct Impact of Climate Change on Regional
Labor Productivity. Archives of Environmental & Occupational Health 64: 217-227.
Knibbs LD and Barnett AG. (2015) Assessing environmental inequalities in ambient air pollution
across urban Australia. Spat Spatiotemporal Epidemiol 13: 1-6.
Kumar S. (2015) Green Climate Fund faces slew of criticism. Nature 527: 419-420.
Kummu M, Keskinen M and Varis O. (2008a) Modern Myths of the Mekong: A critical review of
water and development concepts, principles and policies. In: Kummu M, Keskinen M and
Varis O (eds) Modern myths of the Mekong. Online: Water & Development Publications -
Helsinki University of Technology, 206.
Kummu M, Penny D, Sarkkula J, et al. (2008b) Sediment: curse or blessing for Tonle Sap Lake?
Ambio 37: 158-163.
Kummu M and Sarkkula J. (2008) Impact of the Mekong River flow alteration on the Tonle Sap
flood pulse. Ambio 37: 185-192.
Kyne D and Bolin B. (2016) Emerging Environmental Justice Issues in Nuclear Power and
Radioactive Contamination. International Journal of Environmental Research and Public
Health 13: 700.
Lamberts D. (2008) Little impact, much damage: The consequences of Mekong River flow
alterations for the tonle sap ecosystem. In: Kummu M (ed) Modern myths of the Mekong.
online: Water & Development Publications - Helsinki University of Technology, 3-18.
Page 125
125
Lamont J and Favor C. (2017) Distributive Justice. In: Zalta EN (ed) The Stanford Encyclopedia of
Philosophy.
Lancaster HO. (1969) The chi-squared distribution, New York: Wiley.
Landrigan PJ, Fuller R, Acosta NJR, et al. (2017) The Lancet Commission on pollution and health.
The Lancet.
Lauri H, de Moel H, Ward PJ, et al. (2012) Future changes in Mekong River hydrology: impact of
climate change and reservoir operation on discharge. Hydrology and Earth System Sciences
16: 4603-4619.
Leal Filho W, Pociovalisteanu D-M and Al-Amin AQ. (2017) Sustainable Economic Development
Green Economy and Green Growth, Online: Springer International Publishing.
Lemieux CJ, Gray PA, Douglas AG, et al. (2014) From science to policy: The making of a
watershed-scale climate change adaptation strategy. Environmental Science & Policy 42:
123-137.
Lewis-Beck M, Bryman A and Futing Liao T. (2004) The SAGE Encyclopedia of Social Science
Research Methods, online: SAGE Publications Ltd.
Lewis S. (2017) Extreme Climate Change: Damage and Responsibility. AQ - Australian Quarterly
88: 3-8.
Li VOK, Han Y, Lam JCK, et al. (2018) Air pollution and environmental injustice: Are the socially
deprived exposed to more PM2.5 pollution in Hong Kong? Environmental Science & Policy
80: 53-61.
Maantay J and Maroko A. (2015) 'At-risk' places: inequities in the distribution of environmental
stressors and prescription rates of mental health medications in Glasgow, Scotland.
Environmental Research Letters 10.
MacCoun R and Perlmutter S. (2015) Blind analysis: Hide results to seek the truth. Nature 526:
187-189.
Mak S. (2015) The Governance of Wetlands in the Tonle Sap Lake, Cambodia. Journal of
Environmental Science and Engineering B 4: 331-346.
Manorom K, Baird IG and Shoemaker B. (2017) The World Bank, Hydropower-based Poverty
Alleviation and Indigenous Peoples: On-the-Ground Realities in the Xe Bang Fai River
Basin of Laos. Forum for Development Studies 44: 275-300.
McKinnon MC, Cheng SH, Dupre S, et al. (2016) What are the effects of nature conservation on
human well-being? A systematic map of empirical evidence from developing countries.
Environ Evid 5.
Page 126
126
MEA. (2000) Ecosystems and human well-being wetlands and water synthesis. Online: Millennium
Ecosystem Assessment
MEA. (2005) Millennium Ecosystem Assessment, Ecosystems and human well-being, Washington,
D.C: Island Press.
Mekong River Commission. (2016) Integrated water resources management-based basin
development strategy for the Lower Mekong Basin 2016-2020. online: Mekong River
Commission.
Middleton BA and Souter NJ. (2016) Functional integrity of freshwater forested wetlands,
hydrologic alteration, and climate change. Ecosystem Health and Sustainability 2: e01200-
n/a.
Miguel E, Camerer C, Casey K, et al. (2014) Promoting Transparency in Social Science Research.
Science 343: 30-31.
MoE C. (2005) Vulnerability and Adaptation to Climate Hazards and to Climate Change: A Survey
of Rural Cambodian Households, Phnom Penh: Ministry of Environment, Cambodia.
Mohai P, Pellow D and Roberts JT. (2009) Environmental Justice. 34: 405-430.
Mohai P and Saha R. (2015) Which came first, people or pollution? A review of theory and
evidence from longitudinal environmental justice studies. Environmental Research Letters
10: 125011.
Montzka SA, Dlugokencky EJ and Butler JH. (2011) Non-CO2 greenhouse gases and climate
change. Nature 476: 43-50.
Moon K, Blackman DA, Adams VM, et al. (2019) Expanding the role of social science in
conservation through an engagement with philosophy, methodology, and methods. Methods
in Ecology and Evolution 0.
Moore RJH and Hotchkiss JL. (2016) The importance of toxicity in determining the impact of
hazardous air pollutants on the respiratory health of children in Tennessee. Environmental
Pollution 216: 616-623.
Morello-Frosch R and Jesdale BM. (2006) Separate and Unequal: Residential Segregation and
Estimated Cancer Risks Associated with Ambient Air Toxics in U.S. Metropolitan Areas.
Environmental Health Perspectives 114: 386-393.
MRCS. (2014) National-to-Local benefit sharing options for hydropower on Mekong tributaries
evaluated and reported by 2014. onine: Mekong River Commission Secretariat.
Nakagawa S and Cuthill IC. (2007) Effect size, confidence interval and statistical significance: a
practical guide for biologists. Biological Reviews 82: 591-605.
Page 127
127
Nalepa RA, Short Gianotti AG and Bauer DM. (2017) Marginal land and the global land rush: A
spatial exploration of contested lands and state-directed development in contemporary
Ethiopia. Geoforum 82: 237-251.
Nozick R. (1974) Anarchy, State and Utopia, New York: Basic Books.
NPI. (2014) Substance list and thresholds. Available at:
http://www.npi.gov.au/substances/substance-list-and-thresholds.
NPI. (2015) National Pollutant Inventory Guide. online: National Pollutant Inventory.
Nuorteva P, Keskinen M and Varis O. (2010) Water, livelihoods and climate change adaptation in
the Tonle Sap Lake area, Cambodia: learning from the past to understand the future. Journal
of Water and Climate Change 1: 87-101.
Nurse LA, McLean RF, Agard J, et al. (2014) Small Islands. In: Barros VR, Field CB, Dokken DJ,
et al. (eds) Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part B: Regional
Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel of Climate Change. Cambridge, United Kingdom and New York,
NY, USA: Cambridge University Press, 1613-1654.
O'Faircheallaigh C. (2015) Social Equity and Large Mining Projects: Voluntary Industry Initiatives,
Public Regulation and Community Development Agreements. Journal of Business Ethics
132: 91-103.
O'Neill J, Tabish H, Welch V, et al. (2014) Applying an equity lens to interventions: using
PROGRESS ensures consideration of socially stratifying factors to illuminate inequities in
health. Journal of Clinical Epidemiology 67: 56-64.
O’Neill J, Tabish H, Welch V, et al. (2014) Applying an equity lens to interventions: using
PROGRESS ensures consideration of socially stratifying factors to illuminate inequities in
health. J Clin Epidemiol 67.
OHCHR. (2019) World Conference on Human Rights, 14-25 June 1993, Vienna, Austria. Available
at: https://www.ohchr.org/EN/ABOUTUS/Pages/ViennaWC.aspx.
Olsaretti S. (2018) Introduction: The Idea of Distributive Justice. In: Olsaretti S (ed) The Oxford
Handbook of Distributive Justice. Oxford: Oxford University Press.
Orr S, Pittock J, Chapagain A, et al. (2012) Dams on the Mekong River: Lost fish protein and the
implications for land and water resources. Global Environmental Change 22: 925-932.
Ostrom E. (2014) A polycentric approach for coping with climate change. Ann. Econ. Finance 15:
71-108.
Otto FEL, Frame DJ, Otto A, et al. (2015) Embracing uncertainty in climate change policy. Nature
Clim. Change 5: 917-920.
Page 128
128
Padilla CM, Kihal-Talantikite W, Vieira VM, et al. (2014) Air quality and social deprivation in four
French metropolitan areas—A localized spatio-temporal environmental inequality analysis.
Environmental Research 134: 315-324.
Pauline Dube O and Sivakumar M. (2015) Global environmental change and vulnerability of Least
Developed Countries to extreme events: Editorial on the special issue. Weather and Climate
Extremes 7: 2-7.
Pearce DW and Turner RK. (1990) Economics of natural resources and the environment, Hemel
Hempstead: Harvester Wheatsheaf.
Pearce JR, Richardson EA, Mitchell RJ, et al. (2010) Environmental justice and health: the
implications of the socio-spatial distribution of multiple environmental deprivation for
health inequalities in the United Kingdom. Transactions of the Institute of British
Geographers 35: 522-539.
Perez VW and Egan J. (2016) Knowledge and Concern for Sea-Level Rise in an Urban
Environmental Justice Community. Sociological Forum 31: 885-907.
Pickering C, Grignon J, Steven R, et al. (2014) Publishing not perishing: how research students
transition from novice to knowledgeable using systematic quantitative literature reviews.
Studies in Higher Education: 1-14.
Pickering J, Jotzo F and Wood PJ. (2015) Splitting the difference: can limited coordination achieve
a fair distribution of the global climate financing effort? Environmental Politics 15: 4.
Piketty T. (2014) Capital in the twenty-first century, Cambridge Massachusetts: The Belknap Press
of Harvard University Press.
Piman T, Lennaerts T and Southalack P. (2013) Assessment of hydrological changes in the lower
Mekong Basin from Basin-Wide development scenarios. Hydrological Processes 27: 2115-
2125.
Pizer WA and Yates AJ. (2015) Terminating links between emission trading programs. Journal of
Environmental Economics and Management 71: 142-159.
R Core Development Team. (2015) R: A language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing.
Ramsar. (2017) Cambodia Country Profile. Available at: http://www.ramsar.org/wetland/cambodia.
Rao ND. (2014) International and intranational equity in sharing climate change mitigation burdens.
International Environmental Agreements-Politics Law and Economics 14: 129-146.
Ratner BD, So S, Mam K, et al. (2017) Conflict and collective action in Tonle Sap fisheries:
adapting governance to support community livelihoods. Natural Resources Forum 41: 71-
82.
Page 129
129
Rawls J. (1971) A theory of justice, Cambridge, Mass: Cambridge, Mass., Belknap Press of Harvard
University Press.
Rawls J. (1993) Political liberalism: New York : Columbia University Press.
Rawls J. (2001) Justice as fairness : a restatement, Cambridge: Harvard University Press.
Rivas I, Kumar P and Hagen-Zanker A. (2017) Exposure to air pollutants during commuting in
London: Are there inequalities among different socio-economic groups? Environment
International 101: 143-157.
Robeyns I. (2017) Wellbeing, freedom and social justice the capability approach re-examined,
Cambridge, UK: Open Book Publishers.
Rodriguez-Lara I. (2013) An Experimental Study of Gender Differences in Distributive Justice.
IDEAS Working Paper Series from RePEc.
Roos N, Chamnan C, Loeung D, et al. (2007) Freshwater fish as a dietary source of vitamin A in
Cambodia. Food Chemistry 103: 1104-1111.
Russ GR and Alcala AC. (2011) Enhanced biodiversity beyond marine reserve boundaries: the cup
spillith over. Ecol Appl 21: 241-250.
Sabel CF and Victor DG. (2017) Governing global problems under uncertainty: making bottom-up
climate policy work. Climatic Change 144: 15-27.
Sáenz-Arroyo A, Roberts C, Torre, J., , Cariño-Olvera M, et al. (2005) Rapidly Shifting
Environmental Baselines among Fishers of the Gulf of California. Proceedings: Biological
Sciences 272: 1957-1962.
Sawilowsky SS. (2009) New effect size rules of thumb. Journal of Modern Applied Statistical
Methods 8.
Scanlon T. (2018) Why Does Inequality Matter?: Oxford University Press.
Sen A. (1995) Inequality reexamined, Cambridge, Mass: Harvard University Press.
Sen A. (1999) Development as Freedom, London: Zed Books.
Sen A. (2009) The idea of justice, London ; New York: Allen Lane/Penguin Books.
Sheehan P, Cheng E, English A, et al. (2014) China's response to the air pollution shock. Nature
Clim. Change 4: 306-309.
Shehzad K, Qamer FM, Murthy MSR, et al. (2014) Deforestation trends and spatial modelling of its
drivers in the dry temperate forests of northern Pakistan — A case study of Chitral. Journal
of Mountain Science 11: 1192-1207.
Shepherd A and Brunt J. (2013) Chronic Poverty: Concepts, Causes and Policy, online: Palgrave
Macmillan UK.
Page 130
130
Shortt NK, Richardson EA, Pearce J, et al. (2012) Mortality inequalities by environment type in
New Zealand. Health Place 18: 1132-1136.
Sok S, Yu X and Wong KK. (2012) Impediments to community fisheries management: Some
findings in Kompong Pou commune, Krakor District in Cambodia's Tonle Sap. Singapore
Journal of Tropical Geography 33: 398-413.
Sokhem P and Sunada K. (2006) The Governance of the Tonle Sap Lake, Cambodia: Integration of
Local, National and International Levels. International Journal of Water Resources
Development 22: 399-416.
Stewart IT, Bacon CM and Burke WD. (2014) The uneven distribution of environmental burdens
and benefits in Silicon Valley's backyard. Applied Geography 55: 266-277.
Su JG, Larson T, Gould T, et al. (2010) Transboundary air pollution and environmental justice:
Vancouver and Seattle compared. GeoJournal 75: 595-608.
Sullivan GM and Feinn R. (2012) Using Effect Size—or Why the P Value Is Not Enough. Journal
of Graduate Medical Education 4: 279-282.
Sumner A. (2016) The Geography of Poverty: How has global poverty changed since the end of the
Cold War? Global Poverty. Oxford: Oxford University Press.
Sun C, Kahn ME and Zheng SQ. (2017) Self-protection investment exacerbates air pollution
exposure inequality in urban China. Ecological Economics 131: 468-474.
Szabo S, Hajra R, Baschieri A, et al. (2016) Inequalities in Human Well-Being in the Urban Ganges
Brahmaputra Meghna Delta. Sustainability 8.
Temper L, Bene D and Martinez-Alier J. (2015) Mapping the frontiers and front lines of global
environmental justice: The EJAtlas. Journal of Political Ecology 22: 254-278.
Thomson Reuters. (2015) EndNote X7. New York, NY: Thomson Reuters.
Thomson Reuters. (2017) Web of Science. Available at: http://wokinfo.com/.
Tilt B and Gerkey D. (2016) Dams and population displacement on China’s Upper Mekong River:
Implications for social capital and social–ecological resilience. Global Environmental
Change 36: 153-162.
Tornblom KY. (1977) Distributive Justice: Typology and Propositions. Human Relations 30: 1-24.
Trenberth K. (2011) Changes in precipitation with climate change. Climate Research 47: 123-138.
UN General Assembly. (1948) Universal Declaration of Human Rights.
UNDP. (2017) Human Development Index. Available at: http://hdr.undp.org/en/content/human-
development-index-hdi.
Page 131
131
UNESCO. (1994) Ramsar, Convention on Wetlands of International Importance especially as
Waterfowl Habitat. Online: United Nations Educational, Scientific and Cultural
Organization.
UNESCO. (2007) Biosphere Reserve Information, Cambodia, Tonle Sap. Available at:
http://www.unesco.org/mabdb/br/brdir/directory/biores.asp?code=CAM+01&mode=all.
UNESCO. (2015) Mab Strategy 2015-2025. online: United Nations Organization for Education,
Science and Culture.
UNFCCC. (2015a) Adoption of the Paris Agreement. Conference of the Parties to the United
Nations Framework Convention on Climate Change. Paris: UNFCCC.
UNFCCC. (2015b) INDCs as communicated by Parties. Available at:
http://www4.unfccc.int/submissions/indc/Submission%20Pages/submissions.aspx.
United Nations. (2017) The Sustainable Development Goals Report 2017. online: United Nations
Publications.
United Nations Framework Convention on Climate Change. (1992) Text of the Convention, Rio:
United Nations Framework Convention on Climate Change.
van Buuren S and Groothuis-Oudshoorn K. (2011) mice: Multivariate Imputation by Chained
Equations in R. 45: 1-67.
van Zalinge N, Loeung D, Pengbun N, et al. (2000) Mekong flood levels and Tonle Sap fish
catches. Meeting of the Department of Fisheries. Phnom Penh: Thai National Mekong
Committee Secretariat, 28.
Varis O, Kummu M, Keskinen M, et al. (2006) Tonle Sap Lake, Cambodia: Nature’s affluence
meets human poverty. In: Report UNHD (ed) 1 ed. online: United Nations Human
Development Report.
Verde M, Martinez-Carrion JM and Martinez-Soto AP. (2016) Biological Welfare and Inequality
During the Mining Boom: Rio Tinto, 1832-1935. Revista De Historia Industrial: 149-181.
Ward JD, Sutton PC, Werner AD, et al. (2016) Is Decoupling GDP Growth from Environmental
Impact Possible? PLoS ONE 11: e0164733.
Watson JEM, Dudley N, Segan DB, et al. (2014) The performance and potential of protected areas.
Nature 515: 67-73.
Widlansky MJ, Timmermann A, Stein K, et al. (2013) Changes in South Pacific rainfall bands in a
warming climate. Nature Clim. Change 3: 417-423.
Willcox S. (2016) Climate change inundation, self-determination, and atoll island states. Human
Rights Quarterly 38: 1022-1037.
World Bank Group. (2015) World DataBank.
Page 132
132
World Resources Institute. (2014) Climate Analysis Indicators Tool: WRI’s Climate Data Explorer.
Wu C-f, Wu S-y, Wu Y-H, et al. (2009) Cancer risk assessment of selected hazardous air pollutants
in Seattle. Environment International 35: 516-522.
Ziv G, Baran E, Nam S, et al. (2012) Trading-off fish biodiversity, food security, and hydropower
in the Mekong River Basin. Proceedings of the National Academy of Sciences 109: 5609-
5614.
Page 133
133
Supplementary materials
Materials in this section are separated by chapter. Unless otherwise stated, they appear and are titled
either exactly as they appear in published works, or as they are expected to appear for forthcoming
publications. For example, Nature uses a style of Sx (where x equals material number) to represent
supplementary materials.
Chapter 2 Supplementary materials
Additional file 1 - List of key EDJ studies
This list of papers was identified prior to undertaking literature searches as key types of EDJ
studies. After each iteration of the search process was completed, we ensured that results included
these papers (along with all other search modifications outlined in the main manuscript Methods
section).
1. Althor, G, Watson, JEM & Fuller, RA 2016, 'Global mismatch between greenhouse gas
emissions and the burden of climate change', Scientific Reports, vol. 6.
2. Babidge, S. (2016), Contested value and an ethics of resources: Water, mining and
indigenous people in the Atacama Desert, Chile. Aust J Anthropol, 27: 84–103.
doi:10.1111/taja.12139
3. Cassidy, E., Judge, R., & Sommers, P. (2000). The Distribution of Environmental Justice: A
Comment. Social Science Quarterly, 81(3), 877-878.
4. Chakraborty, J & Green, D 2014, 'Australia’s first national level quantitative environmental
justice assessment of industrial air pollution', Environmental Research Letters, vol. 9, no. 4,
p. 044010.
5. Füssel, H-M 2010, 'How inequitable is the global distribution of responsibility, capability,
and vulnerability to climate change: A comprehensive indicator-based assessment', Global
Environmental Change, vol. 20, no. 4, pp. 597-611.
6. Ghosh, R, Lurmann, F, Perez, L, Penfold, B, Brandt, S, Wilson, J, Milet, M, Kunzli, N &
McConnell, R 2016, 'Near-Roadway Air Pollution and Coronary Heart Disease: Burden of
Disease and Potential Impact of a Greenhouse Gas Reduction Strategy in Southern
California', Environ Health Perspect, vol. 124, no. 2, pp. 193-200.
7. Gochfeld, M & Burger, J 2011, 'Disproportionate exposures in environmental justice and
other populations: the importance of outliers', Am J Public Health, vol. 101 Suppl 1, pp.
S53-63.
Page 134
134
8. Knibbs, LD & Barnett, AG 2015, 'Assessing environmental inequalities in ambient air
pollution across urban Australia', Spat Spatiotemporal Epidemiol, vol. 13, pp. 1-6.
9. Lukasiewicz, A & Dare, M 2016, 'When private water rights become a public asset:
Stakeholder perspectives on the fairness of environmental water management', Journal of
Hydrology, vol. 536, no. Supplement C, pp. 183-91.
10. Nkomo, S & van der Zaag, P 2004, 'Equitable water allocation in a heavily committed
international catchment area: the case of the Komati Catchment', Physics and Chemistry of
the Earth, Parts A/B/C, vol. 29, no. 15, pp. 1309-17.
Page 135
135
Additional file 2 - Search string development log
Each search was undertaken and subsequently modified as per the Methods section of the main
manuscript.
Web of science
Search #1
You searched for: (TS= (natur*) AND TS= (environment*) AND TS= ("distributive justice" OR
equity) NOT TS=(health)) AND LANGUAGE: (English)
Refined by: DOCUMENT TYPES: ( ARTICLE OR BOOK CHAPTER ) AND WEB OF
SCIENCE CATEGORIES: ( ENVIRONMENTAL STUDIES OR ENVIRONMENTAL
SCIENCES OR MULTIDISCIPLINARY SCIENCES OR GEOGRAPHY OR SOCIAL
SCIENCES INTERDISCIPLINARY OR WATER RESOURCES OR POLITICAL SCIENCE OR
AGRONOMY OR FORESTRY OR AGRICULTURE MULTIDISCIPLINARY OR FISHERIES
OR AGRICULTURAL ECONOMICS POLICY OR ETHICS OR SOCIAL SCIENCES
MATHEMATICAL METHODS )
Timespan: 1900-2017. Indexes: SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, BKCI-S,
BKCI-SSH, ESCI, CCR-EXPANDED, IC.
Search #2
Results: 258
You searched for: (TS= (natur*) AND TS= (environment*) AND TS= ("distributive justice" OR
equity) NOT TS=(health)) AND LANGUAGE: (English)
Refined by: DOCUMENT TYPES: ( ARTICLE OR BOOK CHAPTER OR REVIEW ) AND
[excluding] PUBLICATION YEARS: ( 1973 ) AND WEB OF SCIENCE CATEGORIES: (
ENVIRONMENTAL STUDIES OR ENVIRONMENTAL SCIENCES OR ECONOMICS OR
MATHEMATICS INTERDISCIPLINARY APPLICATIONS OR PLANNING DEVELOPMENT
OR GEOGRAPHY OR SOCIAL SCIENCES INTERDISCIPLINARY OR PSYCHOLOGY
SOCIAL OR PHILOSOPHY OR STATISTICS PROBABILITY OR SOCIOLOGY OR WATER
RESOURCES OR URBAN STUDIES OR DEMOGRAPHY OR ENERGY FUELS OR
POLITICAL SCIENCE OR FORESTRY OR AGRICULTURE MULTIDISCIPLINARY OR
FISHERIES OR ETHICS OR ETHNIC STUDIES OR SOCIAL ISSUES OR
MULTIDISCIPLINARY SCIENCES OR AREA STUDIES OR AGRICULTURAL ECONOMICS
POLICY OR SOCIAL SCIENCES MATHEMATICAL METHODS OR GEOGRAPHY
PHYSICAL ) AND DOCUMENT TYPES: ( ARTICLE OR BOOK CHAPTER OR REVIEW )
Page 136
136
AND RESEARCH AREAS: ( BUSINESS ECONOMICS OR PUBLIC ADMINISTRATION OR
GEOGRAPHY OR MATHEMATICS OR MATHEMATICAL METHODS IN SOCIAL
SCIENCES OR SOCIOLOGY OR GOVERNMENT LAW OR WATER RESOURCES OR
URBAN STUDIES OR INTERNATIONAL RELATIONS OR AGRICULTURE OR ENERGY
FUELS OR FISHERIES OR SOCIAL SCIENCES OTHER TOPICS OR FORESTRY OR SOCIAL
ISSUES )
Timespan: All years. Indexes: SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, BKCI-S,
BKCI-SSH, ESCI, CCR-EXPANDED, IC
Search #3
Results: 606
You searched for: (TS= ("distributive justice" OR *equity) AND TS = ("climate change" OR
"ozone depletion" OR "ecosystem service*" OR Urbanization OR biodiversity OR desertification
OR "land degradation" OR "water pollution")) AND LANGUAGE: (English)
Refined by: WEB OF SCIENCE CATEGORIES: ( ENVIRONMENTAL STUDIES OR
ENVIRONMENTAL SCIENCES OR ECONOMICS OR GEOGRAPHY OR ENERGY FUELS
OR WATER RESOURCES OR FISHERIES OR URBAN STUDIES OR POLITICAL SCIENCE
OR SOCIAL ISSUES OR MULTIDISCIPLINARY SCIENCES OR FORESTRY OR ETHICS OR
INTERNATIONAL RELATIONS OR AGRICULTURE MULTIDISCIPLINARY OR SOCIAL
SCIENCES MATHEMATICAL METHODS OR AGRICULTURAL ECONOMICS POLICY OR
SOCIAL SCIENCES INTERDISCIPLINARY ) AND RESEARCH AREAS: ( BUSINESS
ECONOMICS OR SCIENCE TECHNOLOGY OTHER TOPICS OR PUBLIC
ADMINISTRATION OR GEOGRAPHY OR FISHERIES OR SOCIAL ISSUES OR WATER
RESOURCES OR GOVERNMENT LAW OR URBAN STUDIES OR SOCIAL SCIENCES
OTHER TOPICS OR AGRICULTURE OR FORESTRY OR INTERNATIONAL RELATIONS )
AND RESEARCH AREAS: ( BUSINESS ECONOMICS OR FISHERIES OR PUBLIC
ADMINISTRATION OR GEOGRAPHY OR WATER RESOURCES OR SOCIAL ISSUES OR
URBAN STUDIES OR SOCIAL SCIENCES OTHER TOPICS OR AGRICULTURE OR
FORESTRY OR INTERNATIONAL RELATIONS OR ENERGY FUELS )
Timespan: 1980-2017. Indexes: SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, BKCI-S,
BKCI-SSH, ESCI, CCR-EXPANDED, IC.
Search #4
Results: 436
Page 137
137
You searched for: (TI= ("distributive justice" OR equit* OR inequit* OR fair* OR justice) AND TS
= ("Climate change" OR "deforestation" OR "ozone depletion" OR "land degradation" OR
"Reclamation" OR "Mining" OR "air pollution" OR "water pollution" OR "land pollution" OR
"Fishing" OR "Water degradation") NOT TS=(conservation OR ecology OR biology OR species))
AND LANGUAGE: (English)
Refined by: WEB OF SCIENCE CATEGORIES: ( ENVIRONMENTAL STUDIES OR
AGRICULTURE MULTIDISCIPLINARY OR ENVIRONMENTAL SCIENCES OR POLITICAL
SCIENCE OR ECONOMICS OR METEOROLOGY ATMOSPHERIC SCIENCES OR ENERGY
FUELS OR SOCIAL SCIENCES MATHEMATICAL METHODS OR INTERNATIONAL
RELATIONS OR GEOSCIENCES MULTIDISCIPLINARY OR URBAN STUDIES OR
FISHERIES OR AGRICULTURAL ECONOMICS POLICY OR WATER RESOURCES OR
SOCIAL ISSUES OR MINING MINERAL PROCESSING ) AND DOCUMENT TYPES: (
ARTICLE OR REVIEW ) AND DOCUMENT TYPES: ( ARTICLE OR REVIEW ) AND
RESEARCH AREAS: ( ENVIRONMENTAL SCIENCES ECOLOGY OR SOCIAL ISSUES OR
GOVERNMENT LAW OR SOCIOLOGY OR DEMOGRAPHY OR BUSINESS ECONOMICS
OR METEOROLOGY ATMOSPHERIC SCIENCES OR AGRICULTURE OR GEOGRAPHY
OR SOCIAL SCIENCES OTHER TOPICS OR MATHEMATICAL METHODS IN SOCIAL
SCIENCES OR ENERGY FUELS OR INTERNATIONAL RELATIONS OR FISHERIES OR
URBAN STUDIES OR WATER RESOURCES OR MINING MINERAL PROCESSING )
Timespan: 1980-2017. Indexes: SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, BKCI-S,
BKCI-SSH, ESCI, CCR-EXPANDED, IC.
Search #5
Results: 1,175
(TI= ( "distributive justice" OR equit* OR inequit* OR *fair* OR justice OR burden OR just OR
*equal* OR ethic*) AND TS = ( "Climate change" OR "deforestation" OR "ozone depletion" OR
"land degradation" OR "Reclamation" OR "Mining" OR “hunting” OR "air pollution" OR "water
pollution" OR "land pollution" OR "Fishing" OR "Water degradation") NOT TS=(conservation OR
ecology OR biology OR species)) AND LANGUAGE: (English)
Refined by: DOCUMENT TYPES: ( ARTICLE ) AND WEB OF SCIENCE CATEGORIES: (
ENVIRONMENTAL STUDIES OR AGRICULTURE MULTIDISCIPLINARY OR
ENVIRONMENTAL SCIENCES OR AREA STUDIES OR ECONOMICS OR ETHICS OR
GEOGRAPHY OR POLITICAL SCIENCE OR METEOROLOGY ATMOSPHERIC SCIENCES
OR LAW OR FORESTRY OR SOCIOLOGY OR ETHNIC STUDIES OR INTERNATIONAL
Page 138
138
RELATIONS OR ENERGY FUELS OR SOCIAL SCIENCES MATHEMATICAL METHODS
OR MULTIDISCIPLINARY SCIENCES OR WATER RESOURCES OR WOMEN S STUDIES
OR SOCIAL SCIENCES INTERDISCIPLINARY OR FISHERIES OR DEMOGRAPHY OR
AGRICULTURAL ECONOMICS POLICY OR STATISTICS PROBABILITY OR
MATHEMATICS APPLIED OR SOCIAL ISSUES OR URBAN STUDIES ) AND RESEARCH
AREAS: ( ENVIRONMENTAL SCIENCES ECOLOGY OR AREA STUDIES OR BUSINESS
ECONOMICS OR GOVERNMENT LAW OR MINING MINERAL PROCESSING OR SOCIAL
SCIENCES OTHER TOPICS OR GEOGRAPHY OR METEOROLOGY ATMOSPHERIC
SCIENCES OR MATHEMATICAL METHODS IN SOCIAL SCIENCES OR SCIENCE
TECHNOLOGY OTHER TOPICS OR WOMEN S STUDIES OR SOCIOLOGY OR
INTERNATIONAL RELATIONS OR FISHERIES OR ENERGY FUELS OR DEMOGRAPHY
OR WATER RESOURCES OR FORESTRY OR AGRICULTURE OR ETHNIC STUDIES OR
SOCIAL ISSUES OR MATHEMATICS OR URBAN STUDIES )
Timespan: 1980-2017. Indexes: SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, BKCI-S,
BKCI-SSH, ESCI, CCR-EXPANDED, IC.
Scopus
Search #1
Results: 691
TITLE ( "distributive justice" OR *equity OR fair* OR justice ) AND TITLE-ABS-KEY (
"Climate change" OR "deforestation" OR "ozone depletion" OR "land degradation" OR
"Reclamation" OR "Mining" OR "air pollution" OR "water pollution" OR "land pollution" OR
"Fishing" OR "Water degradation" ) AND NOT TITLE-ABS-KEY (conservation OR ecology
OR biology OR species ) AND PUBYEAR > 1980 AND ( LIMIT-TO ( DOCTYPE , "ar " )
OR LIMIT-TO ( DOCTYPE , "re " ) ) AND ( LIMIT-TO ( SUBJAREA , "SOCI " ) OR
LIMIT-TO ( SUBJAREA , "ENVI " ) OR LIMIT-TO ( SUBJAREA , "EART " ) OR LIMIT-
TO ( SUBJAREA , "ECON " ) OR LIMIT-TO ( SUBJAREA , "ENER " ) OR LIMIT-TO (
SUBJAREA , "BUSI " ) OR LIMIT-TO ( SUBJAREA , "AGRI " ) OR LIMIT-TO (
SUBJAREA , "MULT " ) OR EXCLUDE ( SUBJAREA , "BUSI " ) ) AND ( LIMIT-TO (
LANGUAGE , "English " ) )
Search #2
Results: 1,519
TITLE ("distributive justice" OR equit* OR inequit* OR *fair* OR justice OR burden OR just OR
*equal* OR ethic*) AND TITLE-ABS-KEY ("Climate change" OR "deforestation" OR "ozone
Page 139
139
depletion" OR "land degradation" OR "Reclamation" OR "Mining" OR "air pollution" OR "water
pollution" OR "land pollution" OR "Fishing" OR "Water degradation") AND NOT TITLE-ABS-
KEY (conservation OR ecology OR biology OR species) AND PUBYEAR > 1980 AND ( LIMIT-
TO ( SUBJAREA,"ENVI" ) OR LIMIT-TO ( SUBJAREA,"SOCI" ) OR LIMIT-TO (
SUBJAREA,"EART" ) OR LIMIT-TO ( SUBJAREA,"ARTS" ) OR LIMIT-TO (
SUBJAREA,"ECON" ) OR LIMIT-TO ( SUBJAREA,"AGRI" ) OR LIMIT-TO (
SUBJAREA,"ENER" ) OR LIMIT-TO ( SUBJAREA,"MATH" ) OR LIMIT-TO (
SUBJAREA,"MULT" ) OR LIMIT-TO ( SUBJAREA,"DECI" ) OR LIMIT-TO (
SUBJAREA,"Undefined" ) ) AND ( LIMIT-TO ( DOCTYPE,"ar" ) ) AND ( LIMIT-TO (
LANGUAGE,"English" ) )
Proquest
Search #1
217 results
ti("distributive justice" OR equity OR fair???? OR justice OR inequity) AND all("Climate change"
OR "deforestation" OR "ozone depletion" OR "land degradation" OR "Reclamation" OR "Mining"
OR "air pollution" OR "water pollution" OR "land pollution" OR "Fishing" OR "Water
degradation") NOT all(conservation OR ecology OR biology OR species)
Applied filters
Publication date: 1980 - 2017
Document type:
Article OR Review
Subject:
climate change OR studies OR environmental justice OR environmental policy OR air pollution OR
social justice OR emissions OR environmental protection OR global warming OR humans OR
socioeconomic factors OR greenhouse effect OR greenhouse gases OR justice OR politics OR
environmental economics OR sustainable development OR energy policy OR industrialized nations
OR developing countries--ldcs OR international law OR environmental health OR environmental
impact OR ethics OR human rights OR community OR economic models OR emissions trading OR
environment OR international agreements OR outdoor air quality OR data collection OR
environmental law OR international relations OR low income groups OR negotiations OR public
policy OR race OR research OR water pollution OR climate OR decision making
Search #2
Results: 763
Page 140
140
ti("distributive justice" OR equit???? OR inequit???? OR fair???? OR unfair???? OR justice OR
burden OR just OR equal????? OR unequal???? OR inequal???? OR ethic????) AND all ("Climate
change" OR "deforestation" OR "ozone depletion" OR "land degradation" OR "Reclamation" OR
"Mining" OR "air pollution" OR "water pollution" OR "land pollution" OR "Fishing" OR "Water
degradation") NOT all (conservation OR ecology OR biology OR species)
Applied filters
Publication date: 1980-2019
Document type: Article
Language: English
Subject: No filter –subjects too granular on Proquest to match WOS or Scopus.
Page 141
141
Additional file 3 – Variable typologies
These typologies were developed as per Methods in main manuscript. Note that we took variable
descriptions in studies at face value. For example, if an article included a variable named ‘Gender’
as a study demographic, we did not further assess if the authors may have defined gender in a
different manner to other articles.
Social demographic variables
This typology was developed using the social stratifications outlined by O'Neill et al. (2014) as a
baseline. We updated this list during the title and abstract screening phases to include any
demographic variables used in screened publications, which were not included in the baseline.
Social demographic variable Description Notes
Geographic
Article relates to, or studies
people who live within a
specific geographic area. E.g.
a particular neighborhood,
city, or nation.
Children
Article relates to, or studies
children.
No age range for definition of
children across articles was
assessed.
Age
Article relates to, or studies
people of specific age range.
May also include children.
Education
Article relates to, or studies
level of education.
Different levels of education
used in articles were not
assessed.
Employment
Article relates to, or studies
type of, or lack of
employment.
Different types of employment
were not assessed.
Farmers\Fishers\Hunters
Article relates to, or studies
Farmers\Fishers\Hunters.
Groups distinct but aggregated,
due to very low rates of
occurrence.
Future generations
Article relates to, or studies
any form of future
generations.
Can also include future
generations of other
demographic groups, such as
Page 142
142
children.
Immigrants
Article relates to, or studies
immigrants from one nation to
another.
Can also include immigrants
from other demographic groups,
such as children.
Indigenous
Article relates to, or studies
people identified by authors as
Indigenous to the place of
study.
Marital status
Article relates to, or studies
the marital status of groups.
Multiple unspecified
Article relates to, or studies
aggregated social groups, but
does not explicitly describe
the groups within the
aggregations.
Other
Article relates to, or studies a
social demographic not
captured within this typology.
Large number of articles fit
under this category due to the
very diverse nature of studies.
Rural/Urban
Article relates to, or studies
groups which are specifically
rural, or rural/urban
differences.
Does not include solely urban
studies.
LGBTQ+
Article relates to, or studies
any LGBTQ+ groups or sub-
groups.
Any sub-LGBTQ+ groups (e.g.
gay or lesbian) aggregated due
to very low rates of occurrence.
Economic status
Article relates to, or studies
economic status of groups.
Can include income, relative or
absolute poverty etc.
Racial/ethnic
Article relates to, or studies
racial or ethnic groups.
Some studies may define
particular religious groups as
belonging to an ethnic group.
These delineations were not
examined.
Religious Article relates to, or studies Some studies may define
Page 143
143
particular religious groups. particular religious groups as
belonging to an ethnic group. If
so, these studies would be
captured under the Racial/ethnic
category.
Health/Disability
Article relates to, or studies
groups based on physical
health, or groups with
disability.
Gender\sex
Article relates to, or studies
particular genders or sexes.
Human well-being outcome variables
This typology was developed using the domains and definitions of human well-being outcomes,
outlined by McKinnon et al. (2016) as a baseline. We updated and modified this list during the title
and abstract screening phases to include any well-being outcome variables used in screened
publications, which were not included in the baseline.
Human well-being
outcome variable
Description Notes
Education
Article relates to, or studies
outcomes that affect education
level, or access to education
institutions, of groups.
Governance
Article relates to, or studies
outcomes that affect access to
structures and processes for
decision making.
Health/Mortality
Article relates to, or studies
outcomes that affect physical
health, mortality, or access to
Page 144
144
health institutions.
Economic livelihood
Article relates to, or studies
outcomes that affect
livelihoods, such as income and
employment.
Living standards
Article relates to, or studies
outcomes that affect living
standards, such as poverty.
Mental health
Article relates to, or studies
outcomes that affect mental
health.
Multiple HWB
outcomes
Article relates to, or studies
multiple outcomes.
Captured separately to identify
studies which may be assessing
the complex nature of multiple
outcomes and their interactions.
Not specified
Article does not relate to, study,
or specify any HWB outcomes.
Natural resource
rights/access
Article relates to, or studies
outcomes that affect the rights
of groups to, or to access,
particular natural resources.
Security/Safety
Article relates to, or studies
outcomes that affect physical
security, or safety.
Social wellness
Article relates to, or studies
outcomes that affect social
Page 145
145
interactions, social capital, or
other social relationships.
Spiritual/Cultural
fulfilment
Article relates to, or studies
outcomes that affect spiritual or
cultural fulfilment.
Environmental exploitation activity variables
This typology was developed using the exploitation activities described in United Nations (2017),
Pearce et al. (2010), and European Environment Agency (2007) as a baseline. We updated this list
during the title and abstract screening phases to include any exploitation variables used in screened
publications, which were not included in the baseline.
Environmental
exploitation activity
variable
Description Notes
Air pollution Article relates to, or studies
air pollution.
Can include a high variety of
sources, such as traffic and
industry.
Climate change Article relates to, or studies
the caused or effects of
climate change.
Climate change is a highly
complex topic and as such its
various causes and effects are
aggregated under the single
term.
Hydrological
modification
Article relates to, or studies
hydrological change, such as
damming or riverine changes.
Vegetation clearing Article relates to, or studies
vegetation clearing, such as
deforestation, riparian
clearing, or the clearing of
remnant vegetation.
Fishing Article relates to, or studies
fisheries.
Page 146
146
Hunting Article relates to, or studies
hunting as an industry or
community activity.
Land degradation other Article relates to, or studies
the degradation of land which
is not related to soil pollution,
such as erosion, or salinity.
Land reclamation Article relates to, or studies
the reclamation of land, such
as creating new land from
ocean, riverbeds, or lake beds
Mining Article relates to, or studies
mining.
Ozone depletion Article relates to, or studies
ozone depletion.
Soil pollution Article relates to, or studies
soil pollution.
Fresh water degradation
other
Article relates to, or studies
non-hydrological
modification impacts to fresh
water systems.
Sea water
degradation/pollution
Article relates to, or studies
the degradation of sea water,
such as acidification, mining,
or pollution.
Fresh water pollution Article relates to, or studies
the pollution of fresh water
systems.
Environmental hazard variables
This typology was developed using hazard variables described by GRID Arendal (2017) as a
baseline. We updated this list and added descriptions during the title and abstract screening phases
to include any hazard variables used in screened publications, which were not included in the
baseline.
Page 147
147
Environmental hazard
variable
Description Notes
Air quality Article relates to, or studies
exploitation activity which
results in a change to air
quality.
Amenity degradation Article relates to, or studies
exploitation activity which
results in the loss of
environmental amenity.
Fire Article relates to, or studies
exploitation activity which
results in/contributes to
unnatural fire event(s).
Often associated with
climate change.
Intensification of natural
hazards
Article relates to, or studies
exploitation activity which
results in natural hazards
(such as extreme weather)
intensifying.
Used where particular
hazards (e.g. fire events) are
not specified, but generic
language such as
‘intensification of natural
disasters’ etc. used.
Extreme storms Article relates to, or studies
exploitation activity which
contributes to extreme
storm events, such as
cyclones, and hurricanes.
Often associated with
climate change.
Deforestation Article relates to, or studies
exploitation activity which
results in loss of forest.
Desertification Article relates to, or studies
exploitation activity which
results in desertification.
Often associated with
climate change.
Disease Article relates to, or studies
exploitation activity which
Page 148
148
results in increased
occurrence or incidence of
disease.
Drought Article relates to, or studies
exploitation activity which
results in or contributes to
drought events.
Often associated with
climate change.
Erosion Article relates to, or studies
exploitation activity which
results in soil erosion.
Flooding Article relates to, or studies
exploitation activity which
results in or contributes to
flood events.
Often associated with
climate change,
deforestation and erosion.
Food quality\supply Article relates to, or studies
exploitation activity which
results in impacts to either
the quality or availability of
food.
Heat stress/wave Article relates to, or studies
exploitation activity which
results in or contributes to
heat wave or heat stress
events.
Often associated with
climate change.
Landslide Article relates to, or studies
exploitation activity which
results in or contributes to
landslides.
Not specified Article does not specify any
natural hazards.
Oil spill Article relates to, or studies
exploitation activity which
results in oil entering the
Page 149
149
environment uncontrolled.
Over grazing Article relates to, or studies
exploitation activity which
results in overgrazing of
land.
Usually also related to soil
erosion/pollution.
Over fishing Article relates to, or studies
exploitation activity which
results in unsustainable,
over exploitation of a
fishery.
Pest invasion Article relates to, or studies
exploitation activity which
results in or contributes to
vectors used by exotic, pest
species.
Soil pollution Article relates to, or studies
exploitation activity which
results in or contributes to
soil pollution.
Sea level rise Article relates to, or studies
exploitation activity which
results in or contributes to
rising sea levels.
Almost exclusively
associated with climate
change.
Water quality\supply Article relates to, or studies
exploitation activity which
results in or contributes to a
loss of water supply or
quality.
Page 150
150
Additional file 3 – Data tables
Tables which contain all data used in figures within the main manuscript.
Table 1 – Methods used. Counts and percent (rounded to two figures) of methods used.
Method Studies Percent
Case studies 11 3.107345
Census analysis 1 0.282486
Desktop study 73 20.62147
Discussion 25 7.062147
Document analysis 48 13.55932
Ethnographic 3 0.847458
Field research 1 0.282486
Focus groups 2 0.564972
Framework 7 1.977401
Interviews 11 3.107345
Mixed methods 12 3.389831
Modelling 25 7.062147
Philosophical argument 54 15.25424
Spatial analysis 67 18.92655
Surveys 8 2.259887
Informed opinion 6 1.694915
Table 2 – First author institute location. Counts and percent (rounded to two figures) of total
authors shown. NA denotes papers where author location was indeterminable.
Country Count Percent
United states 149 42.09
United Kingdom 45 12.71
Australia 24 6.78
NA 24 6.78
Canada 20 5.65
China 8 2.26
Germany 8 2.26
Spain 8 2.26
Page 151
151
France 6 1.69
India 6 1.69
Italy 6 1.69
Netherlands 5 1.41
Norway 5 1.41
Denmark 4 1.13
Austria 3 0.85
New Zealand 3 0.85
Thailand 3 0.85
Brazil 2 0.56
Chile 2 0.56
Finland 2 0.56
Indonesia 2 0.56
Ireland 2 0.56
Malaysia 2 0.56
South Africa 2 0.56
Sweden 2 0.56
Switzerland 2 0.56
Taiwan 2 0.56
Czech republic 1 0.28
Jamaica 1 0.28
Japan 1 0.28
Pakistan 1 0.28
Singapore 1 0.28
Swaziland 1 0.28
Tanzania 1 0.28
Table 3 – Study location. Counts and percent (rounded to two figures) of study location shown.
NA denotes papers where study location was indeterminable or global.
Country Count Percent
NA 155 43.79
United states 89 25.14
United Kingdom 18 5.08
Page 152
152
Canada 12 3.39
Australia 11 3.11
India 10 2.82
China 8 2.26
New Zealand 7 1.98
France 6 1.69
Chile 4 1.13
South Africa 4 1.13
Brazil 3 0.85
Peru 3 0.85
Spain 3 0.85
Ghana 2 0.56
Indonesia 2 0.56
Bangladesh 1 0.28
Czech republic 1 0.28
Guatemala 1 0.28
Italy 1 0.28
Jamaica 1 0.28
Malaysia 1 0.28
Nepal 1 0.28
Netherlands 1 0.28
Panama 1 0.28
Puerto Rico 1 0.28
Russia 1 0.28
Saint Vincent and the grenadines 1 0.28
Somalia 1 0.28
Sudan 1 0.28
Sweden 1 0.28
Taiwan 1 0.28
Thailand 1 0.28
Page 153
153
Table 4 Demographic themes/variables explored. Counts and percent of total for each paper
shown. Multiple unspecified was recorded where a paper used multiple demographic
themes/variable but did not explicitly state which. Other was recorded to capture the very diverse
range of variables used which were not included in our demographic typology. Note that a single
article may explore several themes and as such counts here are higher than total article counts.
Theme Count Percent
Economic Status 147 16.66667
Geographic 112 12.69841
Racial/ethnic 96 10.88435
Other 84 9.52381
Education 64 7.256236
Age 58 6.575964
Employment 41 4.648526
Gender 41 4.648526
Indigenous 40 4.535147
Multiple unspec. 37 4.195011
Children 30 3.401361
Farmers/Fishers/Hunters 28 3.174603
Health/Disability 26 2.947846
Future generations 25 2.834467
Immigrants 24 2.721088
Rural 20 2.267574
Marital status 7 0.793651
LGBTQ+ 1 0.113379
Religious 1 0.113379
Page 154
154
Table 5 Human well-being themes explored. Counts and percent of total for each paper shown.
Note that a single article may explore several themes and as such counts here are higher than total
article counts.
Theme Count Percent
Health/Mortality 214 38.07829
Living standards 78 13.879
Not specified 78 13.879
Livelihood 63 11.20996
Resource access 42 7.47331
Security/Safety 33 5.871886
Social wellness 23 4.092527
Spiritual/Cultural 12 2.135231
Governance 10 1.779359
Mental health 5 0.88968
Education 4 0.711744
Page 155
155
Table 6 Natural resource exploitation themes explored. Counts and percent of total for each
paper shown. Note that a single article may explore several themes and as such counts here are
higher than total article counts.
Theme Count Percent
Climate change 152 35.43124
Air pollution 130 30.30303
Mining 46 10.72261
Water pollution 30 6.993007
Soil pollution 16 3.729604
Veg. clearing 15 3.496503
Land deg. other 14 3.263403
Fishing 11 2.564103
Fresh water deg. other 7 1.631702
Hydro. modification 4 0.932401
Hunting 2 0.4662
Ozone depletion 2 0.4662
Page 156
156
Table 7 Natural hazard themes explored. Counts and percent of total for each paper shown. Note
that a single article may explore several themes and as such counts here are higher than total article
counts.
Theme Count Percent
Air quality 143 24.23729
Not specified 83 14.0678
Water quality/supply 59 10
N.disaster intensity 56 9.491525
Disease 34 5.762712
Flooding 28 4.745763
Food quality/supply 28 4.745763
Heatwave/stress 27 4.576271
Sea level rise 25 4.237288
Soil pollution 24 4.067797
Drought 18 3.050847
Deforestation 17 2.881356
Intense Storm 14 2.372881
Amenity degradation 12 2.033898
Over fishing 6 1.016949
Erosion 5 0.847458
Land slide 4 0.677966
Desertification 4 0.677966
Extreme Fire 2 0.338983
Over grazing 1 0.169492
Page 157
157
Additional file 4 Complete dataset
Complete dataset is hosted at url: https://data.mendeley.com/datasets/g6frzjmnzt/1
Page 158
158
Chapter 3 Supplementary materials
The text appears exactly as published, with the exception of change from Nature referencing
style to Sage Harvard referencing style and placement of Methods section before Results
section.
S1 Overview
This document provides supplementary information not provided in the main text of the article
Global mismatch between greenhouse gas emissions and the burden of climate change. S2 shows
the climate equity Lorenz curve with Gini and Robin hood indices. S3 shows the results of
performing the same analyses provided in the Methods of the main text, but uses per-capita
emissions. S4 presents a table of data used for analyses.
S2 Summary of Lorenz curve
As detailed in the Methods of the main text, equity indices were created to measure climate
inequity. We created a Lorenz curve (Supplementary Fig. S2) which shows the equity of the
distribution of GHG emissions data for the year 2010(Coulter, 1989a). The blue line represents the
hypothetical line of equality where countries would reside if GHG emissions were divided with
perfect equity between all countries. The red line represents the Lorenz curve and denotes actual
GHG emissions equity for the year 2010. The distance between the blue and red lines represents
GHG emissions inequity and is quantified by the Gini (80.9) and Robin Hood indices (64). As such,
the curve and both indices show that the current distribution of GHG emissions is highly
inequitable.
S3 A summary of per capita GHG emissions and national vulnerability
We conducted separate analyses to those reported in the Methods section of the main text, using per
capita GHG emissions. We created a similar set of maps to those in the main text for per capita
emissions by dividing emissions by national populations (World Bank Group, 2015)
(Supplementary Fig. S3). Per capita emissions are shown for years 2010 (Supplementary Fig. S3a)
and 2030 (Supplementary Fig. S3b). Maps generated using ESRI ArcGIS (ESRI ArcGIS,
2011).The patterns we found here are broadly similar to those in the main text (Supplementary Fig.
S3c). For example, Australia, Russia and the United States of America remain as free riders.
However, several populous major emitters (e.g. United Kingdom, China, Brazil) were no longer
categorised as free riders.
S4 Summary of data
A table of all countries, their level of vulnerability at 2010 and 2030, GHG emissions, GDP at 2010
and population at 2010 and 2030 is presented (Supplementary Table S4).
Page 159
159
Supplementary Figure S2
Supplementary Figure S3
a
b
Page 161
161
Supplementary table S4
Country
Climate
Change
Vulnerabili
ty 2010
Climate
change
vulnerabili
ty 2030
GHG
Emissions
(MtCO2e)
2010 GDP 2010
Population
2010
Population
2030
Afghanistan SEVERE SEVERE 26.5382
159367844
37 28397812 43499632
Albania LOW
MODERAT
E 6.8977
119269572
55 3150143 3310564
Algeria
MODERAT
E
MODERAT
E 166.7126
161207304
960 37062820 48561408
Angola SEVERE ACUTE 262.3005
824708948
68 19549124 34783312
Antigua and
Barbuda SEVERE ACUTE 1.0505
113553903
7 87233 104982
Argentina LOW LOW 426.0262
462703793
707 40374224 46859381
Armenia
MODERAT
E
MODERAT
E 7.8514
926028741
6 2963496 2969807
Australia LOW LOW 592.4785
114126776
0188 22404488 28335501
Austria LOW LOW 89.7307
389656071
767 8401924 9005424
Azerbaijan LOW LOW 58.5329
529027033
76 9094718 10474377
Bahamas SEVERE ACUTE 2.7347
791000000
0 360498 447410
Bahrain LOW LOW 27.9484
257135448
25 1251513 1641988
Bangladesh SEVERE ACUTE 157.8669
115279077
465 151125475 185063630
Barbados
MODERAT
E HIGH 3.5481
443370000
0 280396 305709
Belarus MODERAT MODERAT 110.5189 552209326 9491070 8488334
Page 162
162
E E 14
Belgium LOW LOW 129.1643
484404271
608 10941288 11664194
Belize ACUTE ACUTE 13.9414
139711345
0 308595 461277
Benin SEVERE ACUTE 27.9443
656178231
3 9509798 15506762
Bhutan SEVERE ACUTE -7.4517
158539625
6 716939 897761
Bolivia HIGH ACUTE 149.3575
196496313
08 10156601 13665316
Bosnia and
Herzegovin
a
MODERAT
E
MODERAT
E 27.3532
168474930
59 3845929 3700255
Botswana
MODERAT
E
MODERAT
E 20.1171
137467127
11 1969341 2347860
Brazil LOW
MODERAT
E 1392.6399
214306787
1760 195210154 222748294
Brunei
MODERAT
E
MODERAT
E 23.009
123697088
59 400569 499424
Bulgaria LOW
MODERAT
E 47.5596
486690605
12 7389175 6213179
Burkina
Faso SEVERE ACUTE 32.0538
920928838
3 15540284 26564341
Burundi SEVERE ACUTE 41.1171
202686441
4 9232753 16392403
Cambodia SEVERE ACUTE 48.7258
112422663
34 14364931 19143612
Cameroon SEVERE ACUTE 199.5086
236224829
55 20624343 33074215
Canada LOW LOW 841.5982
161407209
3764 34126240 40616997
Cape Verde HIGH SEVERE 0.6499 na 487601 576734
Central ACUTE ACUTE 103.4364 198601475 4349921 6318381
Page 163
163
African
Republic
9
Chad SEVERE ACUTE 52.7923
106577050
72 11720781 20877527
Chile
MODERAT
E
MODERAT
E 86.8362
217501911
334 17150760 19814578
China LOW
MODERAT
E 9387.0095
593050227
0313
135982146
5
145329730
4
Colombia
MODERAT
E
MODERAT
E 216.3056
287018184
638 46444798 57219408
Comoros ACUTE ACUTE 0.5174 516962949 683081 1057197
Costa Rica
MODERAT
E HIGH 7.584
362983276
70 4669685 5759573
Côte
d'Ivoire SEVERE ACUTE 62.1688
248845039
51 18976588 29227188
Croatia
MODERAT
E
MODERAT
E 15.7967
596438181
82 4338027 4015138
Cuba
MODERAT
E SEVERE 45.2579
643282200
00 11281768 10847333
Cyprus LOW LOW 8.7575
215619413
69 1103685 1306312
Czech
Republic LOW LOW 124.4638
231324503
31 10553701 11053125
Democratic
Republic of
the Congo SEVERE ACUTE 317.3795
207016402
026 62191161 103743184
Denmark LOW LOW 60.1913
319812413
597 5550959 6009458
Djibouti HIGH SEVERE 1.1688
112861170
0 834036 1075146
Dominica SEVERE ACUTE 0.2797 492962963 71167 76952
Dominican
Republic HIGH ACUTE 31.442
530429437
31 10016797 12218615
Ecuador MODERAT HIGH 135.5452 695553670 15001072 19648546
Page 164
164
E 00
Egypt LOW LOW 276.3722
218887812
550 78075705 102552797
El Salvador SEVERE ACUTE 14.1367
214183000
00 6218195 6874758
Equatorial
Guinea HIGH SEVERE 25.5787
115829177
90 696167 1138788
Eritrea HIGH SEVERE 6.1925
211703951
1 5741159 9782455
Estonia LOW LOW 29.0265
194790124
23 1298533 1212150
Ethiopia HIGH SEVERE 148.3437
299337903
34 87095281 137669707
Fiji SEVERE ACUTE 1.3405
314050883
6 860559 939469
Finland LOW LOW 83.3728
247799815
768 5367693 5649744
France LOW LOW 488.4332
264683711
1795 63230866 69286370
Gabon SEVERE ACUTE 6.661
145695271
25 1556222 2382369
Gambia ACUTE ACUTE 6.944 951827285 1680640 3056357
Georgia
MODERAT
E HIGH 9.4743
116385368
34 4388674 3953077
Germany LOW LOW 827.0112
341200877
2737 83017404 79551501
Ghana HIGH ACUTE 59.4589
321748397
13 24262901 35264291
Greece LOW LOW 107.6441
299598056
253 11109999 10975530
Grenada SEVERE ACUTE 1.7461 771015875 104677 107433
Guatemala
MODERAT
E HIGH 50.5749
413379582
52 14341576 22566243
Guinea ACUTE ACUTE 32.8177 473595647 10876033 17322136
Page 165
165
6
Guinea-
Bissau ACUTE ACUTE 4.3053 847491367 1586624 2472642
Guyana SEVERE ACUTE 7.4287
225928839
6 786126 852670
Haiti ACUTE ACUTE 7.9624
662254152
9 9896400 12536811
Honduras SEVERE ACUTE 47.4738
158393445
92 7621204 10811004
Hungary LOW LOW 70.5821
129585601
616 10014633 9525243
Iceland
MODERAT
E
MODERAT
E 2.9422
132610355
17 318042 383558
India SEVERE ACUTE 2304.0432
170845887
6830
120562464
8
147637790
3
Indonesia
MODERAT
E HIGH 2033.0459
709190823
320 240676485 293482460
Iran
MODERAT
E
MODERAT
E 695.2992
422567967
405 74462314 91336270
Iraq
MODERAT
E
MODERAT
E 227.1567
138516722
650 30962380 50966609
Ireland LOW LOW 60.163
218435251
789 4467561 5346841
Israel LOW LOW 86.9162
232907996
791 7420368 9632030
Italy LOW LOW 463.1778
212662040
2889 60508978 61211831
Jamaica ACUTE ACUTE 12.0012
132308440
40 2741485 2949838
Japan LOW LOW 1119.975
549538718
2996 127352833 120624738
Jordan LOW LOW 25.8197
264253794
37 6454554 9355173
Kazakhstan LOW LOW 300.8368 148047348 15921127 18572745
Page 166
166
241
Kenya
MODERAT
E HIGH 56.8606
400001388
31 40909194 66306063
Kiribati SEVERE ACUTE 0.0774 150431114 97743 130715
Kuwait LOW LOW 191.8787
115428557
470 2991580 4832793
Kyrgyzstan
MODERAT
E HIGH -4.6703
479435779
5 5334223 6871058
Laos HIGH ACUTE 45.2108
718144115
2 6395713 8806260
Latvia LOW LOW -2.2515
240096804
60 2090519 1855822
Lebanon LOW
MODERAT
E 24.3469
380099502
49 4341092 5171981
Lesotho
MODERAT
E
MODERAT
E 2.6993
217568568
1 2008921 2419217
Liberia ACUTE ACUTE 16.8827
129269647
6 3957990 6395182
Libya LOW
MODERAT
E 145.7536
747552889
17 6040612 7459411
Lithuania LOW LOW 26.7093
367095115
68 3068457 2816749
Luxembour
g LOW LOW 12.1485
521436503
83 507885 636826
Macedonia LOW
MODERAT
E 12.4073
933867407
8 2102216 2068730
Madagascar ACUTE ACUTE 78.5476
872993613
6 21079532 36000163
Malawi SEVERE ACUTE 23.3628
539861698
5 15013694 25959551
Malaysia
MODERAT
E HIGH 428.9105
247533525
518 28275835 36845517
Maldives HIGH SEVERE 1.1599
213410488
4 325694 435873
Page 167
167
Mali ACUTE ACUTE 32.2724
942226726
0 13985961 26034111
Malta LOW LOW 3.012
816384106
0 424738 436792
Mauritania SEVERE ACUTE 9.9108
352694760
9 3609420 5640323
Mauritius HIGH SEVERE 5.9566
971823391
1 1230659 1287944
Mexico
MODERAT
E HIGH 706.0062
105162794
9327 117886404 143662574
Moldova
MODERAT
E HIGH 11.8209
581160405
2 3573024 3066205
Mongolia HIGH HIGH 52.5383
620035707
0 2712738 3387631
Morocco
MODERAT
E HIGH 92.0352
907706714
32 31642360 39190274
Mozambiqu
e ACUTE ACUTE 56.6871
101653535
91 23967265 38875906
Myanmar ACUTE ACUTE 238.0118 na 51931231 58697747
Namibia
MODERAT
E SEVERE 23.5119
112819964
26 2178967 3042197
Nepal HIGH SEVERE 42.8585
159940946
07 26846016 32853228
Netherlands LOW LOW 220.1469
836389937
229 16615243 17268589
New
Zealand LOW LOW 53.0408
143466743
661 4368136 5208035
Nicaragua SEVERE ACUTE 46.5508
893820965
1 5822209 7390914
Niger SEVERE ACUTE 20.259
571858955
0 15893746 34512751
Nigeria SEVERE ACUTE 491.669
369062403
182 159707780 273120384
North SEVERE ACUTE 108.6613 na 24500520 26718625
Page 168
168
Korea
Norway LOW LOW 27.2304
420945705
225 4891251 5837893
Oman
MODERAT
E HIGH 99.0907
586413528
78 2802768 4920265
Pakistan HIGH ACUTE 326.6593
177165635
077 173149306 231743898
Palau SEVERE ACUTE 0.2839 197558700 20470 24836
Panama
MODERAT
E SEVERE 21.1526
288141000
00 3678128 4882047
Papua New
Guinea SEVERE ACUTE 97.6247
948004795
9 6858945 10044486
Paraguay
MODERAT
E HIGH 112.8127
200305297
33 6459721 8693133
Peru
MODERAT
E HIGH 150.7154
148522810
972 29262830 36513996
Philippines
MODERAT
E SEVERE 145.7789
199589447
424 93444322 127797234
Poland LOW LOW 333.2904
476687891
752 38198754 37447642
Portugal LOW LOW 74.1196
238303443
425 10589792 10432816
Qatar LOW LOW 75.2752
125122249
141 1749713 2760329
Republic of
Congo HIGH SEVERE 21.9944
120078800
67 4111715 6753771
Romania LOW LOW 117.5375
164792252
746 21861476 20232088
Russia LOW LOW 2133.8469
152491611
2079 143617913 133556108
Rwanda HIGH SEVERE 3.9692
569854892
3 10836732 17771249
Saint Lucia SEVERE ACUTE 1.1142
124429740
7 177397 201817
Page 169
169
Saint
Vincent and
the
Grenadines
MODERAT
E SEVERE 0.2239 681225963 109316 110012
Samoa HIGH ACUTE 0.3459 643059403 186029 211105
Sao Tome
and
Principe ACUTE ACUTE 0.1718 201037917 178228 278192
Saudi
Arabia LOW LOW 510.1386
526811466
667 27258387 35634201
Senegal SEVERE ACUTE 28.3144
129324277
24 12950564 21855703
Seychelles SEVERE ACUTE 0.7709 973355738 91208 98416
Sierra
Leone ACUTE ACUTE 10.7209
257815949
6 5751976 8057580
Singapore LOW LOW 70.055
236420337
243 5078969 6577884
Slovakia LOW LOW 37.0475
890119192
05 5433437 5395535
Slovenia LOW
MODERAT
E 5.5184
479729887
42 2054232 2086066
Solomon
Islands ACUTE ACUTE 2.2297 681587105 526447 764146
South
Africa
MODERAT
E
MODERAT
E 458.765
365208432
989 51452352 58095501
South
Korea LOW
MODERAT
E 629.5671
109449935
0178 48453931 52190069
Spain LOW LOW 334.3103
143158761
2302 46182038 48235492
Sri Lanka
MODERAT
E SEVERE 48.0389
495675216
70 20758779 23271183
Sudan HIGH ACUTE 186.007
656320812
40 35652002 55077835
Suriname HIGH SEVERE 9.1899 436839804 524960 603805
Page 170
170
8
Swaziland
MODERAT
E HIGH 2.8898
389156347
8 1193148 1515527
Sweden LOW LOW 39.5656
488377689
565 9382297 10690986
Switzerland LOW LOW 51.274
581208562
423 7830534 9477452
Syria LOW LOW 93.866 na 21532647 29933865
Tajikistan HIGH HIGH 10.8228
564217858
0 7627326 11407028
Tanzania HIGH ACUTE 172.5317
229150045
89 44973330 79354326
Thailand
MODERAT
E SEVERE 344.5517
318907930
076 66402316 67554088
Togo SEVERE ACUTE 16.7057
317294550
6 6306014 10014965
Tonga
MODERAT
E ACUTE 0.3755 369212477 104098 120995
Trinidad
and Tobago
MODERAT
E
MODERAT
E 46.049
207581918
58 1328095 1307826
Tunisia LOW
MODERAT
E 34.2352
444260164
87 10631830 12561225
Turkey LOW
MODERAT
E 353.3523
731168051
903 72137546 86825345
Turkmenist
an
MODERAT
E
MODERAT
E 101.005
221480701
75 5041995 6159875
Uganda HIGH SEVERE 45.6373
160309961
79 33987213 63387713
Ukraine
MODERAT
E
MODERAT
E 356.9985
136419300
368 46050220 39841900
United
Arab
Emirates LOW
MODERAT
E 202.3437
286049293
399 8441537 12330367
United LOW LOW 579.3852 240793376 62066350 68630898
Page 171
171
Kingdom 7805
United
States LOW LOW 6253.7159
149644000
00000 312247116 362628830
Uruguay
MODERAT
E
MODERAT
E 13.1671
388811021
00 3371982 3581432
Uzbekistan LOW
MODERAT
E 214.0483
393327709
29 27769270 34146873
Vanuatu ACUTE ACUTE 0.4817 700804286 236299 352225
Venezuela
MODERAT
E HIGH 399.0262
393801556
872 29043283 37172167
Vietnam HIGH ACUTE 250.0867
115931749
905 89047397 101830324
Yemen
MODERAT
E HIGH 34.1512
317437511
69 22763008 33991041
Zambia SEVERE ACUTE 128.6095
202653963
26 13216985 24956509
Zimbabwe HIGH SEVERE 57.8627
945680820
0 13076978 20292380
Page 172
172
Chapter 4 Supplementary materials
The text appears exactly as intended for publication.
Supplementary file 1
#AUS_EMISSIONS_DJ.R
#Analysis of Australian air pollution, social demographics and health data.
#Written by Glenn Althor 2018
#### load packages and write functions ####
library(truncreg) # Truncated regression
library(ggplot2) # plotting
library(reshape2) # melt
library(geoR) # boxcoxfit
library(psych) # describe
library(dplyr) # data wrangling
library(mice) # ad hoc imputation
library(censReg) # censored/tobit modelling
library(viridis) # colour blind friendly colour pallete
library(Publish) # ci.mean
library(Cairo) # print figures
#### create colour palettes ####
two_col <- viridis(2)
#### Prepare emissions data ####
rm(list = ls())
setwd("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/R/AUS_EMISSIONS_DJ")
npi_emissions <- read.table(header=TRUE,
sep=",","C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/NPI/@Working/npi_emissions.csv", stringsAsFactors = F)
colnames(npi_emissions)[1] <- "report_year"
Page 173
173
tox_matched_substances <- read.table(header=TRUE,sep =
',',"C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/NPI/@Working/tox_matched_substances.csv", stringsAsFactors = F)
colnames(tox_matched_substances)[1]<- "substance_name"
# merge datasets, keeping matching toxicity to npi emissions (all.x = T) = join keeping x.
dat <- merge(npi_emissions, tox_matched_substances, by = "substance_name")
rm(npi_emissions)
rm(tox_matched_substances)
# remove entries for non-air/water emissions and 0 values for emissions
dat <- subset(dat, air_total_emission_kg != "NA" |
water_emission_kg != "NA")
dat <- subset(dat, air_total_emission_kg != 0 |
water_emission_kg != 0)
# Convert emissions NA values to 0
dat[c(3:6,10:13)][is.na(dat[c(3:6,10:13)])] <- 0
# remove entries for chemicals with no HTP (may reconsider later)
dat <- subset(dat, HTP_Chemical_name != "NA")
# Create total risk - emission*HTP
dat$weight_cancer_risk_air <- dat$air_total_emission_kg*dat$HTP_Cancer_Air
dat$weight_cancer_risk_water <- dat$water_emission_kg*dat$HTP_Cancer_Water
dat$tot_cancer_risk <- dat$weight_cancer_risk_air+dat$weight_cancer_risk_water
dat$weight_noncancer_risk_air <- dat$air_total_emission_kg*dat$HTP_NonCancer_Air
dat$weight_noncancer_risk_water <- dat$water_emission_kg*dat$HTP_NonCancer_Water
dat$tot_noncancer_risk <- dat$weight_noncancer_risk_air+dat$weight_noncancer_risk_water
dat$tot_risk <- dat$tot_cancer_risk+dat$tot_noncancer_risk
# filterout all non-necessary columns
dat <- dat[c(2,7,8,20)]
dat.2016 <- subset(dat, report_year == "2015/2016")
dat.2011 <- subset(dat, report_year == "2011/2012")
dat.2006 <- subset(dat, report_year == "2006/2007")
Page 174
174
# # Export csv files
write.csv(dat.2016,file = "./outputs/2016_weighted_emissions.csv", row.names = F)
write.csv(dat.2011,file = "./outputs/2011_weighted_emissions.csv", row.names = F)
write.csv(dat.2006,file = "./outputs/2006_weighted_emissions.csv", row.names = F)
#### ARC GIS spatial join process ####
# 1. Convert csv to xls
# 2. Import xls
# 3. Right click - Display XY data
# 4. Export as shape
# 5. Spatial join shape to SA1 data - Join one to one, and sum total risk
# 6. Export new data set as text, and convert to csv "2016_emissions_join.csv"
# Add SA2 and SA3 Codes to dataset
rm(list = ls())
setwd("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/R/AUS_EMISSIONS_DJ")
SA1_2016_emissions <- read.csv2("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3.
Environmental equity in Australia/Data/R/AUS_EMISSIONS_DJ/2016_emissions_join.csv", sep =
',')
colnames(SA1_2016_emissions)[5] <- "SA1_7DIGITCODE_2016"
SA_All <- read.csv2("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/R/AUS_EMISSIONS_DJ/SA1_2016_AUST.csv", sep = ',')
dat <- merge(SA1_2016_emissions, SA_All)
write.csv(dat,file = "./outputs/2016_emissions_join_SA.csv")
#### Impute 2016 ####
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
## TTP 2016 data ##
# TTP 2016 import #
Page 175
175
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
TTP_2016 <- read.table("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/outputs/relative_TTP_2016.txt", sep = ',', header = T)
# Clean data up & Filter out unnecessary columns
dat_2016 <- TTP_2016[c(5,11)]
colnames(dat_2016)[1] <- "SA1_7DIGITCODE_2016"
colnames(dat_2016)[2] <- "TTP_2016"
rm(TTP_2016)
## SEIFA 2016 data ##
SEIFA_2016 <- read.csv2("./SEIFA_2016.csv", sep = ',', header = T, stringsAsFactors = F)
colnames(SEIFA_2016)[1] <- "SA1_7DIGITCODE_2016"
SEIFA_2016 <- SEIFA_2016[c(1:2,4,6,8)]
SEIFA_2016[,] <- as.numeric(as.character(unlist(SEIFA_2016[,])))
dat_2016 <- merge(dat_2016, SEIFA_2016, all.x = T)
colnames(dat_2016) <- c("SA1_7DIGITCODE_2016",
"TTP_2016",
"IRSD_Score",
"IRSAD_Score",
"IEO_Score",
"IER_Score")
dat_2016[is.na(dat_2016)] <- NA
dat_2016 <- na.omit(dat_2016)
rm(SEIFA_2016)
## Indigenous % data 2016##
IND_pct_2016 <- read.csv2("./INDI_SA1_2016.csv", sep = ',', header = T, stringsAsFactors = F)
IND_pct_2016[,] <- as.numeric(as.character(unlist(IND_pct_2016[,])))
dat_2016 <- merge(dat_2016, IND_pct_2016, all.x = T)
dat_2016$pct_ind[is.na(dat_2016$pct_ind)] <- NA
dat_2016$pct_ind <- na.omit(dat_2016$pct_ind)
rm(IND_pct_2016)
Page 176
176
# Summary statistics #
# describe(dat_2016[c(2:7)])
## Adhoc imputation ##
# convert 0 obs to NA #
dat_2016$TTP_2016[dat_2016$TTP_2016 == 0] <- NA
# View missing data #
md_2016 <- md.pattern(dat_2016)
md_2016
# Mulitple imputations - Predictive mean matching #
imp_2016 <- mice(dat_2016[c(2:7)], m = 50, method = "pmm", seed = 1)
# xy plot to check imputed values
TTP_2016_imp_p <- xyplot(imp_2016, TTP_2016 ~ IRSAD_Score | .imp, pch = 20, cex = 1.4,
col=two_col, alpha = 0.5)
## Regressions ##
# Percent Ind regression
IND_2016 <- with(imp_2016, lm(log(TTP_2016) ~ pct_ind))
# summary(IND_2016)
# pool(IND_2016)
# summary(pool(IND_2016))
# pool.r.squared(IND_2016)
# IRSAD regression #
IRSAD_2016 <- with(imp_2016, lm(log(TTP_2016) ~ IRSAD_Score))
# summary(IRSAD_2016)
# pool(IRSAD_2016)
# summary(pool(IRSAD_2016))
# pool.r.squared(IRSAD_2016)
# IRSD regression #
Page 177
177
IRSD_2016 <- with(imp_2016, lm(log(TTP_2016) ~ IRSD_Score + pct_ind))
# summary(IRSD_2016)
# pool(IRSD_2016)
# summary(pool(IRSD_2016))
# pool.r.squared(IRSD_2016)
# IER regression #
IER_2016 <- with(imp_2016, lm(log(TTP_2016) ~ IER_Score))
# summary(IER_2016)
# pool(IER_2016)
# summary(pool(IER_2016))
# pool.r.squared(IER_2016)
# IEO regression #
IEO_2016 <- with(imp_2016, lm(log(TTP_2016) ~ IEO_Score))
# summary(IEO_2016)
# pool(IEO_2016)
# summary(pool(IEO_2016))
# pool.r.squared(IEO_2016)
# Combine all R^2 values #
r_sq_all_2016_imp <- as.data.frame(cbind(c(pool.r.squared(IRSAD_2016)[1,1],
pool.r.squared(IRSD_2016)[1,1],
pool.r.squared(IER_2016)[1,1],
pool.r.squared(IEO_2016)[1,1],
pool.r.squared(IND_2016)[1,1])))
r_sq_all_2016_imp$y <- round(r_sq_all_2016_imp$V1*100, digits = 3)
colnames(r_sq_all_2016_imp) <- c("Estimated r^2","Estimated % variance")
rownames(r_sq_all_2016_imp) <- c("IRSAD","IRSD","IER","IEO","% Indig. Aus.")
#### Impute 2011 ####
# TTP Import #
Page 178
178
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
dat_2011 <- read.table("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/outputs/relative_TTP_2011.txt", sep = ',', header = T)
# Clean data up & Filter out unnecessary columns
dat_2011 <- dat_2011[c(5,11)]
colnames(dat_2011)[1] <- "SA1_7DIGITCODE_2016"
colnames(dat_2011)[2] <- "TTP_2011"
# SEIFA 2011 DATA #
SEIFA_2011 <- read.csv2("./SEIFA_2011.csv", sep = ',', header = T, stringsAsFactors = F)
colnames(SEIFA_2011)[1] <- "SA1_7DIGITCODE_2016"
SEIFA_2011 <- SEIFA_2011[c(1:2,4,6,8)]
SEIFA_2011[,] <- as.numeric(as.character(unlist(SEIFA_2011[,])))
dat_2011 <- merge(dat_2011, SEIFA_2011, all.x = T)
dat_2011[is.na(dat_2011)] <- NA
dat_2011 <- na.omit(dat_2011)
rm(SEIFA_2011)
## Indigenous % Data 2011##
IND_pct_2011 <- read.csv2("./INDI_SA1_2011.csv", sep = ',', header = T, stringsAsFactors = F)
IND_pct_2011[,] <- as.numeric(as.character(unlist(IND_pct_2011[,])))
IND_pct_2011 <- IND_pct_2011[c(1,4)]
colnames(IND_pct_2011)[c(1,2)] <- c("SA1_7DIGITCODE_2016","pct_ind")
dat_2011 <- merge(dat_2011, IND_pct_2011, all.x = T)
dat_2011$pct_ind[is.na(dat_2011$pct_ind)] <- NA
dat_2011$pct_ind <- na.omit(dat_2011$pct_ind)
rm(IND_pct_2011)
# Summary statistics #
describe(dat_2011[c(2:7)])
## Adhoc imputation ##
# convert 0 obs to NA #
Page 179
179
dat_2011$TTP_2011[dat_2011$TTP_2011 == 0] <- NA
# Check for missing data
md_2011 <- md.pattern(dat_2011)
md_2011
# adhoc imputation prediction
imp_2011 <- mice(dat_2011[c(2:7)], m = 50, method = "pmm", seed = 1)
# Check imputed data
TTP_2011_imp_p <- xyplot(imp_2011, TTP_2011 ~ IRSAD_Score | .imp, pch = 20, cex = 1.4,
col=two_col, alpha = 0.5)
## Regressions ##
# Percent Ind regression
IND_2011 <- with(imp_2011, lm(log(TTP_2011) ~ pct_ind))
# summary(IND_2011)
# pool(IND_2011)
# summary(pool(IND_2011))
# pool.r.squared(IND_2011)
# IRSAD regression #
IRSAD_2011 <- with(imp_2011, lm(log(TTP_2011) ~ IRSAD_Score))
# summary(IRSAD_2011)
# pool(IRSAD_2011)
# summary(pool(IRSAD_2011))
# pool.r.squared(IRSAD_2011)
# IRSD regression #
IRSD_2011 <- with(imp_2011, lm(log(TTP_2011) ~ IRSD_Score + pct_ind))
# summary(IRSD_2011)
# pool(IRSD_2011)
# summary(pool(IRSD_2011))
# pool.r.squared(IRSD_2011)
Page 180
180
# IER regression #
IER_2011 <- with(imp_2011, lm(log(TTP_2011) ~ IER_Score))
# summary(IER_2011)
# pool(IER_2011)
# summary(pool(IER_2011))
# pool.r.squared(IER_2011)
# IEO regression #
IEO_2011 <- with(imp_2011, lm(log(TTP_2011) ~ IEO_Score))
# summary(IEO_2011)
# pool(IEO_2011)
# summary(pool(IEO_2011))
# pool.r.squared(IEO_2011)
# Combine all R^2 values #
# 2011 #
r_sq_all_2011_imp <- as.data.frame(cbind(c(pool.r.squared(IRSAD_2011)[1,1],
pool.r.squared(IRSD_2011)[1,1],
pool.r.squared(IER_2016)[1,1],
pool.r.squared(IEO_2016)[1,1],
pool.r.squared(IND_2016)[1,1])))
r_sq_all_2011_imp$y <- round(r_sq_all_2011_imp$V1*100, digits = 3)
colnames(r_sq_all_2011_imp) <- c("Estimated r^2","Estimated % variance")
rownames(r_sq_all_2011_imp) <- c("IRSAD","IRSD","IER","IEO","% Indig. Aus.")
#### Impute 2006 ####
# TTP Import #
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
dat_2006 <- read.table("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/outputs/relative_TTP_2006.txt", sep = ',', header = T)
# Clean data up & Filter out unnecessary columns
dat_2006 <- dat_2006[c(4,5)]
Page 181
181
colnames(dat_2006)[1] <- "CD_CODE06"
colnames(dat_2006)[2] <- "TTP_2006"
# SEIFA 2006 DATA #
SEIFA_2006 <- read.csv2("./SEIFA_2006.csv", sep = ',', header = T, stringsAsFactors = F)
SEIFA_2006 <- SEIFA_2006[c(1:2,4,6,8)]
SEIFA_2006[,] <- as.numeric(as.character(unlist(SEIFA_2006[,])))
dat_2006 <- merge(dat_2006, SEIFA_2006, all.x = T)
dat_2006[is.na(dat_2006)] <- NA
dat_2006 <- na.omit(dat_2006)
rm(SEIFA_2006)
## Indigenous % Data 2006##
IND_pct_2006 <- read.csv2("./INDI_CD_2006.csv", sep = ',', header = T, stringsAsFactors = F)
IND_pct_2006[,] <- as.numeric(as.character(unlist(IND_pct_2006[,])))
IND_pct_2006 <- IND_pct_2006[c(1,4)]
colnames(IND_pct_2006)[c(1,2)] <- c("CD_CODE06","pct_ind")
dat_2006 <- merge(dat_2006, IND_pct_2006, all.x = T)
dat_2006$pct_ind[is.na(dat_2006$pct_ind)] <- NA
dat_2006$pct_ind <- na.omit(dat_2006$pct_ind)
rm(IND_pct_2006)
# Summary statistics #
describe(dat_2006[c(2:7)])
## Adhoc imputation ##
# convert 0 obs to NA #
dat_2006$TTP_2006[dat_2006$TTP_2006 == 0] <- NA
# Check for missing data
md_2006 <- md.pattern(dat_2006)
md_2006
# adhoc imputation prediction
imp_2006 <- mice(dat_2006[c(2:7)], m = 50, method = "pmm", seed = 1)
Page 182
182
# Check imputed data
TTP_2006_imp_p <- xyplot(imp_2006, TTP_2006 ~ IRSAD_Score | .imp, pch = 20, cex = 1.4,
col=two_col, alpha = 0.5)
## Regressions ##
# Percent Ind regression
IND_2006 <- with(imp_2006, lm(log(TTP_2006) ~ pct_ind))
# summary(IND_2006)
# pool(IND_2006)
# summary(pool(IND_2006))
# pool.r.squared(IND_2006)
# IRSAD regression #
IRSAD_2006 <- with(imp_2006, lm(log(TTP_2006) ~ IRSAD_Score))
# summary(IRSAD_2006)
# pool(IRSAD_2006)
# summary(pool(IRSAD_2006))
# pool.r.squared(IRSAD_2006)
# IRSD regression #
IRSD_2006 <- with(imp_2006, lm(log(TTP_2006) ~ IRSD_Score))
summary(IRSD_2006)
# pool(IRSD_2006)
# summary(pool(IRSD_2006))
# pool.r.squared(IRSD_2006)
# IER regression #
IER_2006 <- with(imp_2006, lm(log(TTP_2006) ~ IER_Score))
# summary(IER_2006)
# pool(IER_2006)
# summary(pool(IER_2006))
# pool.r.squared(IER_2006)
# IEO regression #
Page 183
183
IEO_2006 <- with(imp_2006, lm(log(TTP_2006) ~ IEO_Score))
# summary(IEO_2006)
# pool(IEO_2006)
# summary(pool(IEO_2006))
# pool.r.squared(IEO_2006)
# Combine all R^2 values #
r_sq_all_2006_imp <- as.data.frame(cbind(c(pool.r.squared(IRSAD_2006)[1,1],
pool.r.squared(IRSD_2006)[1,1],
pool.r.squared(IER_2016)[1,1],
pool.r.squared(IEO_2016)[1,1],
pool.r.squared(IND_2016)[1,1])))
r_sq_all_2006_imp$y <- round(r_sq_all_2006_imp$V1*100, digits = 3)
colnames(r_sq_all_2006_imp) <- c("Estimated r^2","Estimated % variance")
rownames(r_sq_all_2006_imp) <- c("IRSAD","IRSD","IER","IEO","% Indig. Aus.")
#### Data and plot exports ####
## Data outputs ##
dir.create(file.path(getwd(), "outputs/imputed/"))
# Missing Data #
write.csv(md_2016,file = "./outputs/imputed/missing_data_2016.csv")
write.csv(md_2011,file = "./outputs/imputed/missing_data_2011.csv")
write.csv(md_2006,file = "./outputs/imputed/missing_data_2006.csv")
## Summary statistics ##
# Change NA back to 0 #
dat_2006[c(2)][is.na(dat_2006[c(2)])] <- 0
dat_2011[c(2)][is.na(dat_2011[c(2)])] <- 0
dat_2016[c(2)][is.na(dat_2016[c(2)])] <- 0
write.csv(describe(dat_2016[c(2:7)]),file = "./outputs/2016_summary_stats.csv")
write.csv(describe(dat_2011[c(2:7)]),file = "./outputs/2011_summary_stats.csv")
write.csv(describe(dat_2006[c(2:7)]),file = "./outputs/2006_summary_stats.csv")
Page 184
184
# Pooled regressions #
write.csv(summary(pool(IND_2016)), file = "./outputs/imputed//IND_2016.csv")
write.csv(summary(pool(IRSAD_2016)), file = "./outputs/imputed//IRSAD_2016.csv")
write.csv(summary(pool(IRSD_2016)), file = "./outputs/imputed//IRSD_2016.csv")
write.csv(summary(pool(IER_2016)), file = "./outputs/imputed//IER_2016.csv")
write.csv(summary(pool(IEO_2016)), file = "./outputs/imputed//IEO_2016.csv")
write.csv(summary(pool(IND_2011)), file = "./outputs/imputed//IND_2011.csv")
write.csv(summary(pool(IRSAD_2011)), file = "./outputs/imputed//IRSAD_2011.csv")
write.csv(summary(pool(IRSD_2011)), file = "./outputs/imputed//IRSD_2011.csv")
write.csv(summary(pool(IER_2011)), file = "./outputs/imputed//IER_2011.csv")
write.csv(summary(pool(IEO_2011)), file = "./outputs/imputed//IEO_2011.csv")
write.csv(summary(pool(IND_2006)), file = "./outputs/imputed//IND_2006.csv")
write.csv(summary(pool(IRSAD_2006)), file = "./outputs/imputed//IRSAD_2006.csv")
write.csv(summary(pool(IRSD_2006)), file = "./outputs/imputed//IRSD_2006.csv")
write.csv(summary(pool(IER_2006)), file = "./outputs/imputed//IER_2006.csv")
write.csv(summary(pool(IEO_2006)), file = "./outputs/imputed//IEO_2006.csv")
# pooled R^2 estimates #
# Output #
write.csv(r_sq_all_2006_imp, file = "./outputs/imputed/r_sq_all_2006_imp.csv")
write.csv(r_sq_all_2011_imp, file = "./outputs/imputed/r_sq_all_2011_imp.csv")
write.csv(r_sq_all_2016_imp, file = "./outputs/imputed/r_sq_all_2016_imp.csv")
## Plots ##
# Imputed data distribution plots #
Cairo(3800, 3800, file="./outputs/imputed//plots/TTP_2016_imp_p.png",
type = "png", dpi = 300, bg="white")
TTP_2016_imp_p
dev.off()
Page 185
185
Cairo(3800, 3800, file="./outputs/imputed//plots/TTP_2011_imp_p.png",
type = "png", dpi = 300, bg="white")
TTP_2011_imp_p
dev.off()
Cairo(3800, 3800, file="./outputs/imputed//plots/TTP_2006_imp_p.png",
type = "png", dpi = 300, bg="white")
TTP_2006_imp_p
dev.off()
#### truncated regressions ####
#### trunc 2016 ####
# TTP 2016 Import #
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
TTP_2016 <- read.table("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/outputs/relative_TTP_2016.txt", sep = ',', header = T)
# Clean data up & Filter out unnecessary columns
dat_2016_t <- TTP_2016[c(5,11)]
colnames(dat_2016_t)[1] <- "SA1_7DIGITCODE_2016"
colnames(dat_2016_t)[2] <- "TTP_2016"
rm(TTP_2016)
dat_2016_t <- subset(dat_2016_t, TTP_2016 > 0) # 0 values
# SEIFA 2016 DATA #
SEIFA_2016 <- read.csv2("./SEIFA_2016.csv", sep = ',', header = T, stringsAsFactors = F)
colnames(SEIFA_2016)[1] <- "SA1_7DIGITCODE_2016"
SEIFA_2016 <- SEIFA_2016[c(1:2,4,6,8)]
SEIFA_2016[,] <- as.numeric(as.character(unlist(SEIFA_2016[,])))
dat_2016_t <- merge(dat_2016_t, SEIFA_2016, all.x = T)
dat_2016_t[is.na(dat_2016_t)] <- NA
dat_2016_t <- na.omit(dat_2016_t)
colnames(dat_2016_t) <- c("SA1_7DIGITCODE_2016",
"TTP_2016",
Page 186
186
"IRSD_Score",
"IRSAD_Score",
"IEO_Score",
"IER_Score")
rm(SEIFA_2016)
## Indigenous % data 2016 ##
IND_pct_2016 <- read.csv2("./INDI_SA1_2016.csv", sep = ',', header = T, stringsAsFactors = F)
IND_pct_2016[,] <- as.numeric(as.character(unlist(IND_pct_2016[,])))
dat_2016_t <- merge(dat_2016_t, IND_pct_2016, all.x = T)
dat_2016_t$pct_ind[is.na(dat_2016_t$pct_ind)] <- NA
dat_2016_t$pct_ind <- na.omit(dat_2016_t$pct_ind)
rm(IND_pct_2016)
# IND Truncated regression #
IND_2016_t <- truncreg(TTP_2016~pct_ind, data = dat_2016_t, point = 0, direction = "left")
# summary(IRSAD_2016_t)
# IND 2016 Generate r^2 estimates
dat_2016_t$IND_2016_t_yhat <- fitted(IND_2016_t)
r_IND_2016_t <- with(dat_2016_t, cor(TTP_2016, IND_2016_t_yhat))
r_sq_IND_2016_t <- r_IND_2016_t^2
rm(r_IND_2016_t)
# IRSAD Truncated regression #
IRSAD_2016_t <- truncreg(TTP_2016~IRSAD_Score, data = dat_2016_t, point = 0, direction =
"left")
# summary(IRSAD_2016_t)
# Generate r^2 estimates
dat_2016_t$IRSAD_2016_t_yhat <- fitted(IRSAD_2016_t)
r_IRSAD_2016_t <- with(dat_2016_t, cor(TTP_2016, IRSAD_2016_t_yhat))
r_sq_IRSAD_2016_t <- r_IRSAD_2016_t^2
rm(r_IRSAD_2016_t)
# IRSD Truncated regression #
Page 187
187
IRSD_2016_t <- truncreg(TTP_2016~IRSD_Score, data = dat_2016_t, point = 0, direction = "left")
# summary(IRSD_2016_t)
# Generate r^2 estimates
dat_2016_t$IRSD_2016_t_yhat <- fitted(IRSD_2016_t)
r_IRSD_2016_t <- with(dat_2016_t, cor(TTP_2016, IRSD_2016_t_yhat))
r_sq_IRSD_2016_t <- r_IRSD_2016_t^2
rm(r_IRSD_2016_t)
# IER Truncated regression #
IER_2016_t <- truncreg(TTP_2016~IER_Score, data = dat_2016_t, point = 0, direction = "left")
# summary(IER_2016_t)
# Generate r^2 estimates
dat_2016_t$IER_2016_t_yhat <- fitted(IER_2016_t)
r_IER_2016_t <- with(dat_2016_t, cor(TTP_2016, IER_2016_t_yhat))
r_sq_IER_2016_t <- r_IER_2016_t^2
rm(r_IER_2016_t)
# IEO Truncated regression #
IEO_2016_t <- truncreg(TTP_2016~IEO_Score, data = dat_2016_t, point = 0, direction = "left")
# summary(IEO_2016_t)
# Generate r^2 estimates
dat_2016_t$IEO_2016_t_yhat <- fitted(IEO_2016_t)
r_IEO_2016_t <- with(dat_2016_t, cor(TTP_2016, IEO_2016_t_yhat))
r_sq_IEO_2016_t <- r_IEO_2016_t^2
rm(r_IEO_2016_t)
# Export all r^2 values #
r_sq_all_2016 <- c(r_sq_IRSAD_2016_t,r_sq_IRSD_2016_t,r_sq_IER_2016_t,r_sq_IEO_2016_t,
r_sq_IND_2016_t)
r_sq_all_2016 <- melt(data.frame(r_sq_all_2016))
r_sq_all_2016 <- r_sq_all_2016[c(2)]
r_sq_all_2016$y <- round(r_sq_all_2016$value/1*100, digits = 3)
Page 188
188
colnames(r_sq_all_2016) <- c("Estimated r^2","Estimated % variance")
rownames(r_sq_all_2016) <- c("IRSAD","IRSD","IER","IEO","% Indig. Aus.")
#### trunc 2011 ####
# TTP 2011 Import #
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
TTP_2011 <- read.table("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/outputs/relative_TTP_2011.txt", sep = ',', header = T)
# Clean data up & Filter out unnecessary columns
dat_2011_t <- TTP_2011[c(5,11)]
colnames(dat_2011_t)[1] <- "SA1_7DIGITCODE_2011"
colnames(dat_2011_t)[2] <- "TTP_2011"
rm(TTP_2011)
dat_2011_t <- subset(dat_2011_t, TTP_2011 > 0) # 0 values
# SEIFA 2011 DATA #
SEIFA_2011 <- read.csv2("./SEIFA_2011.csv", sep = ',', header = T, stringsAsFactors = F)
colnames(SEIFA_2011)[1] <- "SA1_7DIGITCODE_2011"
SEIFA_2011 <- SEIFA_2011[c(1:2,4,6,8)]
SEIFA_2011[,] <- as.numeric(as.character(unlist(SEIFA_2011[,])))
dat_2011_t <- merge(dat_2011_t, SEIFA_2011, all.x = T)
dat_2011_t[is.na(dat_2011_t)] <- NA
dat_2011_t <- na.omit(dat_2011_t)
colnames(dat_2011_t) <- c("SA1_7DIGITCODE_2011",
"TTP_2011",
"IRSD_Score",
"IRSAD_Score",
"IEO_Score",
"IER_Score")
rm(SEIFA_2011)
## Indigenous % Data 2011##
Page 189
189
IND_pct_2011 <- read.csv2("./INDI_SA1_2011.csv", sep = ',', header = T, stringsAsFactors = F)
IND_pct_2011[,] <- as.numeric(as.character(unlist(IND_pct_2011[,])))
IND_pct_2011 <- IND_pct_2011[c(1,4)]
colnames(IND_pct_2011)[c(1,2)] <- c("SA1_7DIGITCODE_2011","pct_ind")
dat_2011_t <- merge(dat_2011_t, IND_pct_2011, all.x = T)
dat_2011_t$pct_ind[is.na(dat_2011_t$pct_ind)] <- NA
dat_2011_t$pct_ind <- na.omit(dat_2011_t$pct_ind)
rm(IND_pct_2011)
# IND Truncated regression #
IND_2011_t <- truncreg(TTP_2011~pct_ind, data = dat_2011_t, point = 0, direction = "left")
# summary(IRSAD_2011_t)
# IND 2011 Generate r^2 estimates
dat_2011_t$IND_2011_t_yhat <- fitted(IND_2011_t)
r_IND_2011_t <- with(dat_2011_t, cor(TTP_2011, IND_2011_t_yhat))
r_sq_IND_2011_t <- r_IND_2011_t^2
# IRSAD Truncated regression #
IRSAD_2011_t <- truncreg(TTP_2011~IRSAD_Score, data = dat_2011_t, point = 0, direction =
"left")
# summary(IRSAD_2011_t)
# Generate r^2 estimates
dat_2011_t$IRSAD_2011_t_yhat <- fitted(IRSAD_2011_t)
r_IRSAD_2011_t <- with(dat_2011_t, cor(TTP_2011, IRSAD_2011_t_yhat))
r_sq_IRSAD_2011_t <- r_IRSAD_2011_t^2
# IRSD Truncated regression #
IRSD_2011_t <- truncreg(TTP_2011~IRSD_Score + pct_ind, data = dat_2011_t, point = 0,
direction = "left")
# summary(IRSD_2011_t)
# Generate r^2 estimates
dat_2011_t$IRSD_2011_t_yhat <- fitted(IRSD_2011_t)
r_IRSD_2011_t <- with(dat_2011_t, cor(TTP_2011, IRSD_2011_t_yhat))
r_sq_IRSD_2011_t <- r_IRSD_2011_t^2
Page 190
190
rm(r_IRSD_2011_t)
# IER Truncated regression #
IER_2011_t <- truncreg(TTP_2011~IER_Score, data = dat_2011_t, point = 0, direction = "left")
# summary(IER_2011_t)
# Generate r^2 estimates
dat_2011_t$IER_2011_t_yhat <- fitted(IER_2011_t)
r_IER_2011_t <- with(dat_2011_t, cor(TTP_2011, IER_2011_t_yhat))
r_sq_IER_2011_t <- r_IER_2011_t^2
rm(r_IER_2011_t)
# IEO Truncated regression #
IEO_2011_t <- truncreg(TTP_2011~IEO_Score, data = dat_2011_t, point = 0, direction = "left")
# summary(IEO_2011_t)
# Generate r^2 estimates
dat_2011_t$IEO_2011_t_yhat <- fitted(IEO_2011_t)
r_IEO_2011_t <- with(dat_2011_t, cor(TTP_2011, IEO_2011_t_yhat))
r_sq_IEO_2011_t <- r_IEO_2011_t^2
rm(r_IEO_2011_t)
# Export all r^2 values #
r_sq_all_2011 <- c(r_sq_IRSAD_2011_t,r_sq_IRSD_2011_t,r_sq_IER_2011_t,r_sq_IEO_2011_t,
r_sq_IND_2011_t)
r_sq_all_2011 <- melt(data.frame(r_sq_all_2011))
r_sq_all_2011 <- r_sq_all_2011[c(2)]
r_sq_all_2011$y <- round(r_sq_all_2011$value/1*100, digits = 3)
colnames(r_sq_all_2011) <- c("Estimated r^2","Estimated % variance")
rownames(r_sq_all_2011) <- c("IRSAD","IRSD","IER","IEO","% Indig. Aus.")
#### trunc 2006 ####
# TTP 2006 Import #
Page 191
191
setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
TTP_2006 <- read.table("C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental
equity in Australia/Data/outputs/relative_TTP_2006.txt", sep = ',', header = T)
# Clean data up & Filter out unnecessary columns
dat_2006_t <- TTP_2006[c(4:5)]
colnames(dat_2006_t)[1] <- "CD_CODE06"
colnames(dat_2006_t)[2] <- "TTP_2006"
rm(TTP_2006)
dat_2006_t <- subset(dat_2006_t, TTP_2006 > 0) # 0 values
# SEIFA 2006 DATA #
SEIFA_2006 <- read.csv2("./SEIFA_2006.csv", sep = ',', header = T, stringsAsFactors = F)
colnames(SEIFA_2006)[1] <- "CD_CODE06"
SEIFA_2006 <- SEIFA_2006[c(1:2,4,6,8)]
SEIFA_2006[,] <- as.numeric(as.character(unlist(SEIFA_2006[,])))
dat_2006_t <- merge(dat_2006_t, SEIFA_2006, all.x = T)
dat_2006_t[is.na(dat_2006_t)] <- NA
dat_2006_t <- na.omit(dat_2006_t)
colnames(dat_2006_t) <- c("CD_CODE06",
"TTP_2006",
"IRSD_Score",
"IRSAD_Score",
"IEO_Score",
"IER_Score")
rm(SEIFA_2006)
## Indigenous % Data 2006##
IND_pct_2006 <- read.csv2("./INDI_CD_2006.csv", sep = ',', header = T, stringsAsFactors = F)
IND_pct_2006[,] <- as.numeric(as.character(unlist(IND_pct_2006[,])))
IND_pct_2006 <- IND_pct_2006[c(1,4)]
colnames(IND_pct_2006)[c(1,2)] <- c("CD_CODE06","pct_ind")
dat_2006_t <- merge(dat_2006_t, IND_pct_2006, all.x = T)
dat_2006_t$pct_ind[is.na(dat_2006_t$pct_ind)] <- NA
Page 192
192
dat_2006_t$pct_ind <- na.omit(dat_2006_t$pct_ind)
rm(IND_pct_2006)
# IND Truncated regression #
IND_2006_t <- truncreg(TTP_2006~pct_ind, data = dat_2006_t, point = 0, direction = "left")
# summary(IRSAD_2006_t)
# IND 2006 Generate r^2 estimates
dat_2006_t$IND_2006_t_yhat <- fitted(IND_2006_t)
r_IND_2006_t <- with(dat_2006_t, cor(TTP_2006, IND_2006_t_yhat))
r_sq_IND_2006_t <- r_IND_2006_t^2
# IRSAD Truncated regression #
IRSAD_2006_t <- truncreg(TTP_2006~IRSAD_Score, data = dat_2006_t, point = 0, direction =
"left")
# summary(IRSAD_2006_t)
# Generate r^2 estimates
dat_2006_t$IRSAD_2006_t_yhat <- fitted(IRSAD_2006_t)
r_IRSAD_2006_t <- with(dat_2006_t, cor(TTP_2006, IRSAD_2006_t_yhat))
r_sq_IRSAD_2006_t <- r_IRSAD_2006_t^2
# IRSD Truncated regression #
IRSD_2006_t <- truncreg(TTP_2006~IRSD_Score, data = dat_2006_t, point = 0, direction = "left")
# summary(IRSD_2006_t)
# Generate r^2 estimates
dat_2006_t$IRSD_2006_t_yhat <- fitted(IRSD_2006_t)
r_IRSD_2006_t <- with(dat_2006_t, cor(TTP_2006, IRSD_2006_t_yhat))
r_sq_IRSD_2006_t <- r_IRSD_2006_t^2
rm(r_IRSD_2006_t)
# IER Truncated regression #
IER_2006_t <- truncreg(TTP_2006~IER_Score, data = dat_2006_t, point = 0, direction = "left")
# summary(IER_2006_t)
# Generate r^2 estimates
Page 193
193
dat_2006_t$IER_2006_t_yhat <- fitted(IER_2006_t)
r_IER_2006_t <- with(dat_2006_t, cor(TTP_2006, IER_2006_t_yhat))
r_sq_IER_2006_t <- r_IER_2006_t^2
rm(r_IER_2006_t)
# IEO Truncated regression #
IEO_2006_t <- truncreg(TTP_2006~IEO_Score, data = dat_2006_t, point = 0, direction = "left")
# summary(IEO_2006_t)
# Generate r^2 estimates
dat_2006_t$IEO_2006_t_yhat <- fitted(IEO_2006_t)
r_IEO_2006_t <- with(dat_2006_t, cor(TTP_2006, IEO_2006_t_yhat))
r_sq_IEO_2006_t <- r_IEO_2006_t^2
rm(r_IEO_2006_t)
# Export all r^2 values #
r_sq_all_2006 <- c(r_sq_IRSAD_2006_t,r_sq_IRSD_2006_t,r_sq_IER_2006_t,r_sq_IEO_2006_t,
r_sq_IND_2006_t)
r_sq_all_2006 <- melt(data.frame(r_sq_all_2006))
r_sq_all_2006 <- r_sq_all_2006[c(2)]
r_sq_all_2006$y <- round(r_sq_all_2006$value*100, digits = 3)
colnames(r_sq_all_2006) <- c("Estimated r^2","Estimated % variance")
rownames(r_sq_all_2006) <- c("IRSAD","IRSD","IER","IEO","% Indig. Aus.")
#### Data and plot exports ####
## Data outputs ##
dir.create(file.path(getwd(), "outputs/truncated/"))
# truncated regression coefficients #
write.csv(summary(IND_2016_t)$coefficients, file = "./outputs/truncated/IND_2016_t.csv")
write.csv(summary(IRSAD_2016_t)$coefficients, file = "./outputs/truncated/IRSAD_2016_t.csv")
write.csv(summary(IRSD_2016_t)$coefficients, file = "./outputs/truncated/IRSD_2016_t.csv")
write.csv(summary(IER_2016_t)$coefficients, file = "./outputs/truncated/IER_2016_t.csv")
write.csv(summary(IEO_2016_t)$coefficients, file = "./outputs/truncated/IEO_2016_t.csv")
Page 194
194
write.csv(summary(IND_2011_t)$coefficients, file = "./outputs/truncated/IND_2011_t.csv")
write.csv(summary(IRSAD_2011_t)$coefficients, file = "./outputs/truncated/IRSAD_2011_t.csv")
write.csv(summary(IRSD_2011_t)$coefficients, file = "./outputs/truncated/IRSD_2011_t.csv")
write.csv(summary(IER_2011_t)$coefficients, file = "./outputs/truncated/IER_2011_t.csv")
write.csv(summary(IEO_2011_t)$coefficients, file = "./outputs/truncated/IEO_2011_t.csv")
write.csv(summary(IND_2006_t)$coefficients, file = "./outputs/truncated/IND_2006_t.csv")
write.csv(summary(IRSAD_2006_t)$coefficients, file = "./outputs/truncated/IRSAD_2006_t.csv")
write.csv(summary(IRSD_2006_t)$coefficients, file = "./outputs/truncated/IRSD_2006_t.csv")
write.csv(summary(IER_2006_t)$coefficients, file = "./outputs/truncated/IER_2006_t.csv")
write.csv(summary(IEO_2006_t)$coefficients, file = "./outputs/truncated/IEO_2006_t.csv")
# R^2 estimates #
write.csv(r_sq_all_2016, file = "./outputs/truncated/r_sq_all_2016_t.csv")
write.csv(r_sq_all_2011, file = "./outputs/truncated/r_sq_all_2011_t.csv")
write.csv(r_sq_all_2006, file = "./outputs/truncated/r_sq_all_2006_t.csv")
#### Plots ####
ggplot(dat_2016, aes(TTP_2016)) +
geom_histogram()
# #### Unpublished code ####
#
# ## RA Data 2016 ##
# setwd('C:/Users/glenn/Dropbox/Uni/phd/@Research/Paper - Ch3. Environmental equity in
Australia/Data/Census')
# RA_2016 <- read.table("./RA_2016_AUST.csv", sep = ',', header = T)
# # Clean data up & Filter out unnecessary columns
# RA_2016 <- RA_2016[c(3,5)]
# colnames(RA_2016 )[1] <- "SA1_7DIGITCODE_2016"
# colnames(RA_2016 )[2] <- "RA_2016"
Page 195
195
# # rename values to remove state names
# RA_2016$RA_2016 <- gsub("^\\Migratory - Offshore - Shipping.*", "Migratory - Offshore -
Shipping", RA_2016$RA_2016)
# RA_2016$RA_2016 <- gsub("^\\No usual address.*", "No usual address", RA_2016$RA_2016)
# RA_2016$RA_2016 <- as.factor(RA_2016$RA_2016)
# RA_2016[,] <- as.numeric(as.character(unlist(RA_2016[,])))
# RA_2016 <- RA_2016[!duplicated(RA_2016), ]
# dat_2016 <- merge(dat_2016, RA_2016, all.x = T)
# rm(RA_2016)
#
#
# # # Merge
# # colnames(imp_complete)[2] <- "TTP_2016_imp"
# # dat_2016 <- merge(dat_2016, imp_complete, all.x = T)
#
# # ## Testing model - Adhoc imputation - and mice - vignette:
https://gerkovink.github.io/miceVignettes/Ad_hoc_and_mice/Ad_hoc_methods.html
# # fit2 <- with(data = dat_2016, expr = lm(log(TTP_2016_imp) ~ IRSD_Score + RA_2016))
# # mean(log(dat_2016$TTP_2016), na.rm = TRUE)
# # mean(fit$fitted.values)
# # confint(fit1, level = 0.9)
# # confint(fit2, level = 0.9)
Supplementary file 2
Counts of ‘missing’ data per year
CD_CODE06 2006_IRSAD_Score
2006_IRSD_Score
2006_IER_Score
2006_IEO_Score
2006_pct_ind
TTP_2006
1 1 1 1 1 1 1
1 1 1 1 1 1 0
0 0 0 0 0 0 28851 SA1_7DIGITCODE_2016
2011_IRSAD_Score
2011_IRSD_Score
2011_IER_Score
2011_IEO_Score
2011_pct_ind
TTP_2011
1 1 1 1 1 1 1
1 1 1 1 1 1 0
0 0 0 0 0 0 37479 SA1_7DIGITCODE_2016
2016_IRSD_Score
2016_IRSAD_Score
2016_IEO_Score
2016_IER_Score
2016_pct_ind
TTP_2016
1 1 1 1 1 1 1
Page 196
196
1 1 1 1 1 1 0
0 0 0 0 0 0 44952
Supplementary file 3
Imputed data validity example
Page 199
199
Chapter 5 Supplementary materials
The text appears exactly as published.
Supplementary material S1
Food Security and conservation in the Tonle Sap: Interview guide
Glenn Althor, The University of Queensland, 2015.
This interview guide is written for the exclusive purpose to assist the interviewer while undertaking
interviews. It is broken up into sections which correspond to the project’s research questions. Only
the questions in bold will be directly asked of the participants. Each question is followed by a
checklist of actions and/or potential follow up questions. Follow up questions will only be asked if
the participants answers to open ended questions are insufficient or off topic.
Participant eligibility
How long have you lived in this village?
Must be at least 10 years, if not exit interview.
What is your main occupation?
Must be fisher, if not exit interview.
Read out participant information sheet
Offer participant to ask questions
Introductory comments before recording
Anonymity – data storage and management
Purpose of research, aims, expected outcomes.
Interested in learning from feelings, opinions, and experiences, no right or wrong answers.
Welcome to interrupt, ask for clarification, criticize line of questioning, refuse to answer, end
interview at any point.
Expected duration of 60-90 minutes, so at some points we may move along to adhere to time
requirements. But will take time to answer any questions for the participant.
Consent form.
Start recording.
Fish profile
o What are the main fish types that people in your village catch?
o Write down list of fish on seasonal calendar – try to get at least 6.
o What are the most common fish species eaten? – be insistent only require a common
name (Khmer or English)
Page 200
200
What are the most common fish species that people in your village eat every day?
o Write on the seasonal calendar
What are the most common fish species that people in your village sell?
o Write on seasonal calendar
o Are different fish sold to middle merchants from other villagers?
Seasonal calendar by month
o Fill out fish section of the seasonal calendar
How do people obtain fish?
o Do people catch fish?
o How many people in the village catch fish?
o Do people catch all of the fish they eat or do people also buy fish?
o How many people in the village buy fish?
o Do people farm fish, if yes which species?
o How many people in the village farm fish?
Can you please draw a map of the areas where people in your community fish?
o Give participant pen and paper to draw a map.
o Ask them to draw wet season and dry season fishing areas (dry red, wet blue).
Over the last 10-20 years, do you think fish catch has increased, declined or stayed the same?
o By how much do you think fish catches changed?
o Have any particular fish species abundance changed? - just need a common name (Khmer
or English)
o Have fish changed in size, getting bigger or smaller? Which species?
o If they answer ‘declined’, also ask following question
Why do you think fish have declined?
o Is there overfishing?
o Has the population in the village changed in the last 10 years?
o What fishing gear are people commonly using?
o Has illegal fishing increased?
o Are there less flooded forests (fish habitat loss)?
o Are there weed problems (habitat degradation/eutrophication)?
o Are you aware of any efforts to reverse fish declines?
Impacts of Environmental variability
Is the fish catch between the annual wet and dry seasons, the same or different?
o Catch size in dry seasons, per year for last 5 years?
Page 201
201
o Catch size in wet seasons, per year for last 5 years?
Is this year unusually, very dry (as dry as 2002)?
o Yes/No
Is fish catch different when it is very dry (drought years)?
o Difference in catch during drought years?
o Did the village undertake any new activities to relieve pressures in very dry years?
Do you remember when the flood was big flood in years, 2013, 2011 and 2000?
o Yes/No
Was the fish catch different in these big flood years?
o Difference in catch during flood years?
o Did the village undertake any new activities to relieve pressures in big flood years?
Does your community experience severe storms?
o Yes/No
Is the fish catch impacted by severe storms?
o Are the fishing conditions to poor to fish in?
o Does your fishing gear get damaged?
o Did the village undertake any new activities to relieve pressures in very dry years?
Conservation Area
Are you familiar with the nearby protected areas?
o Yes/No
Do people ever enter the protected areas to gather resources like fish or wood?
o Do people do other activities in the Prek Toal Core Zone such as bird egg harvest?
Some people think the bird sanctuary increases the fish catch in the area surrounding it, do
you think this happens here?
o Is fish catch better near the bird sanctuary?
o Where on the border of the sanctuary is catch better?
o Use participant’s map of area
Is access to the fisheries near the protected areas equal between the commune villages?
o Do other villages get more benefits and fewer costs from the protected areas?
Closing comments
o Are there any additional points the particpant wishes to make?
o Are there any remarks the participant wishes to clarify?
o Does the participant have any questions for the interviewer about interivew, project, etc?
Page 203
203
Supplementary material S2
Page 204
204
Supplementary material S3
Page 205
205
Supplementary material S4
Code index
Code title Definition
0 Trans Translation unclear - data not used.
1.0 Fish Not used.
1.1 Fish - catch Non-specific discussion of people or community catching fish
1.1.1 Fish - catch - freq Not used.
1.1.1.1 Fish - catch - freq - rare Species catch is rare
1.1.1.2 Fish - catch - freq - common Species catch is common
1.1.2 Fish - catch - use Catch utility
1.1.2.1 Fish - catch - use - eat Catch is for use as subsistence food
1.1.2.2 Fish - catch - use - sell Catch is for use as source of income
1.1.2.3 Fish - catch - use - farm feed Catch is for use as food in fish/crocodile farms
1.1.3 Fish - catch - seasonal Catch differences between seasons
1.1.3.1 Fish - catch - seasonal - dry Description of catch in dry season
1.1.3.2 Fish - catch - seasonal - wet Description of catch in wet season
1.1.3.3 Fish - catch - seasonal - drought Description of catch during drought periods
1.1.3.4 Fish - catch - seasonal - flood Description of catch during (extreme) floods
1.1.3.5 Fish - catch - seasonal - storm Description of catch during storms
1.1.3.6 Fish - catch - seasonal - month Monthly seasonality of catch
1.1.4 Fish - catch - area Areas where fishing takes place
1.1.4.1 Fish - catch - area - lake (within) Fishing within Tonle Sap lake
1.1.4.2 Fish - catch - area - lake (shore) Fishing on Tonle Sap shore
1.1.4.3 Fish - catch - area - stream Fishing within rivers/steams (note that lake is often translated as river)
1.1.4.4 Fish - catch - area - protected area (within)
Fishing within protected area
1.1.4.5 Fish - catch - area - protected area (border)
Fishing along protected area border
1.1.4.6 Fish - catch - area - community fishery Fishing in the community fishery (lot 3)
1.1.5 Fish - catch - species names Names of species caught by participant
1.1.4.7 Fish - catch - area - greatest Description of area where catch is greatest
1.1.4.8 Fish - catch - area - poor Description of area where catch is poor
1.1.4.9 Fish - catch - area - average Description of area where catch is average
1.2 Fish - description descriptions of fish species
1.3 Fish - size descriptions of fish sizes
1.3.1 Fish - size - decrease fish size decreases
1.3.1.1 Fish - size - decrease - reason reasons why the fish size has decreased
1.3.3 Fish - size - no change no change in fish sizes
1.3.2 Fish - size - increase fish size increases
1.3.2.1 Fish - size - increase - reason reasons why the fish size has decreased
1.4 Fish - Buy Non-specific discussion of people or community buying fish
1.4.1 Fish - Buy - Eat Fish are bought to eat
1.4.2 Fish - Buy - Farm Fish are bought to farm
1.5 Fish - farm Fish are farmed
1.5.1.1 Fish - farm - use - eat Farmed to eat
1.5.1.2 Fish - farm - use - sell Farmed to sell
Page 206
206
1.5.2 Fish - farm - common Villagers commonly farm fish
1.5.3 Fish - farm - rare Villagers rarely farm fish
1.5.4 Fish - farm - species Descriptions of species that are farmed
1.5.5 Fish - farm - feed Descriptions of feed - not local catch (see 1.1.2.3)
2 Fishery Codes for fishery
2.1 Fishery - change Changes in fishery (over last 20 years)
2.1.1 Fishery - change - decline declines in fishery output
2.1.1.1.1 Fishery - change - decline - reason - other
reasons for declines - not in list below
2.1.1.1.2 Fishery - change - decline - reason - gear
declines due to fishing gear
2.1.1.1.3 Fishery - change - decline - reason - overfish
declines due to over harvesting of fish
2.1.1.1.4 Fishery - change - decline - reason - no jobs
declines due to no other employment opportunities
2.1.1.1.5 Fishery - change - decline - reason - over pop
declines due to too many people needing food
2.1.1.1.6 Fishery - change - decline - reason - conservation
declines due to the restrictions on people from conservation
2.1.1.1.7 Fishery - change - decline - reason - vietnamese
declines due to activities by Vietnamese people
2.1.2 Fishery - change - improve improvements in fishery output
2.1.2.1 Fishery - change - improve - reason reasons for improvements
2.1.1.2 Fishery - change - decline - human response
How have people respond to declines in the fishery
2.2 Fishery - access access to fisheries information
3 Climate change Mention of changing climate not related to fish catch