Download The

BUILDING TOOLS TO MODEL THE EFFECTS OF OCEAN ACIDIFICATION AND

HOW IT SCALES FROM PHYSIOLOGY TO FISHERIES

Travis Christopher Tai

B.Sc., The University of Western Ontario, 2010

M.Sc., The University of Victoria, 2014

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES

(Oceans and Fisheries)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

September 2019

Building tools to model the effects of ocean acidification and how it scales from physiology to fisheries.

submitted by Travis Tai in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Ocean and Fisheries

Examining Committee: Dr. U. Rashid Sumaila Co-supervisor

Dr. William W.L. Cheung Co-supervisor

Dr. Nadja Steiner Supervisory Committee Member

Dr. Marie Auger-Méthé University Examiner

Dr. Colin J. Brauner University Examiner

Dr. Sam Dupont External Examiner Additional Supervisory Committee Members: Dr. Christopher D.G. Harley Supervisory Committee Member

Abstract

Ocean acidification is a direct consequence of elevated atmospheric carbon dioxide caused by anthropogenic fossil fuel burning and is one of multiple climate-related stressors in marine environments. Understanding of how these stressors will interact to affect marine life and fisheries is limited. In this thesis, I used integrated modelling approaches to scale the effects of biophysical drivers from physiology to population dynamics and fisheries. I focused on ocean acidification and how it interacts with other main drivers such as temperature and oxygen. I used a dynamic bioclimatic envelope model (DBEM) to project the effects of global environmental change on fisheries under two contrasting scenarios of climate change—the low optimistic climate change scenario in line with the 2015 Paris Agreement to limit global warming to 1.5˚ C, and the high climate change scenario on par with our current ‘business-as-usual’ trajectory. First,

I developed an ex-vessel fish price database and explored methods using various ocean acidification assumptions. Ex-vessel fish prices are essential for fisheries economic analyses, while model development of ocean acidification effects are important to better understand the uncertainties surrounding acidification and the sensitivity of the model to these uncertainties.

These tools and methods were then used to project the impacts of ocean acidification, in the context of climate change, on global invertebrate fisheries—the species group most sensitive to acidification. My results showed that areas with greater acidification have greater negative responses to climate change, e.g. polar regions. However, ocean warming will likely be a greater driver in species distributions and may overshadow direct effects of acidification. While greater climate change will generally have negative consequences on fisheries, Arctic regions may see increased fisheries catch potential as species shift poleward. Canada’s Arctic remains one of the

iii

most pristine marine regions left in the world and climate-driven increases in fisheries potential will have major implications for biodiversity and local indigenous reliance on marine resources.

In the face of global environmental change, my thesis provides databases, modelling approaches, scenario development, and assessments of global change necessary for adaptation and mitigation of climate-related effects on marine fisheries.

Lay Summary

Marine species are impacted by climate change and human exploitation for fisheries resources.

There is a major gap in understanding how ocean acidification—amongst other climate-change impacts—will affect organisms and how impacts will scale from biology to fisheries. My thesis uses quantitative simulation models to determine how climate-change related stressors and fishing pressure affect the biology, population, and fisheries of commercially valuable marine species. Two main findings emerge:

1. Climate change has profound effects on global and regional fisheries, affecting potential

catch and revenues; and

2. Ocean acidification effects on species are variable across regions, and its impacts may be

secondary to temperature effects.

I used future scenarios of ocean acidification to derive results from various indicators to evaluate impacts at different spatial scales. Scenario development facilitates an understanding of potential climate change outcomes and is used to identify and quantify the various sources of uncertainty.

Preface

I am the primary author of this thesis and the lead author for all chapters. I led the primary design, implementation, and analyses of all research in this thesis. This thesis was written with the guidance and support of my co-supervisors, Dr. Rashid Sumaila and Dr. William Cheung, as well as my other committee members, Dr. Chris Harley and Dr. Nadja Steiner. Furthermore, co- authors provided insightful thought and contributions to ensure the quality of each chapter for publication.

A version of chapter 2 has been published as, “Tai TC, Cashion T, Lam VWY, Swartz W,

Sumaila UR (2017) Ex-vessel Fish Price Database: Disaggregating Prices for Low-Priced

Species from Reduction Fisheries. Frontiers in Marine Science, 4:1–10, DOI:

10.3389/fmars.2017.00363.” I conceived the initial idea for this manuscript with guidance from

URS. The original draft of the manuscript was researched and written by me. I primarily collected the data for this manuscript, with help from TC to identify and collect additional data.

TC also added valuable suggestions and knowledge on the various fisheries end-products (e.g. direct human consumption, fishmeal and fish oil) and kindly provided that dataset to be incorporated into the update of fisheries ex-vessel prices and landed values. All authors contributed to writing and revisions for publication.

A version of chapter 3 has been published as, “Tai TC, Harley CDG, Cheung WWL (2018)

Comparing model parameterizations of the biophysical impacts of ocean acidification to identify limitations and uncertainties. Ecological Modelling, 385:1–11, DOI:

10.1016/j.ecolmodel.2018.07.007.” I conceptualized the initial research questions for this

manuscript. Furthermore, I collected the data, conducted the model simulations and statistical analyses, and wrote the original draft version of the manuscript. The initial model was written by

WWLC and then revised by me to test the main research questions. All authors contributed to writing and revisions for final publication.

A version of chapter 4 has been prepared as a manuscript with co-authors UR Sumaila and WWL

Cheung, with the working title, “Ocean acidification amplifies multi-stressor impacts on global marine invertebrate fisheries.” This manuscript has been submitted for peer review. I conceptualized the research questions for this manuscript with guidance from WWLC, collected the data, conducted the model simulations and analyses, and wrote the original draft version of the manuscript. The initial model was written by WWLC and then revised by me to test the main research questions. All authors contributed to writing and revisions for final publication.

A version of chapter 5 has been prepared as a manuscript and published to Marine Policy as,

“Tai TC, Steiner NS, Hoover C, Cheung WWL, Sumaila UR (2019) Evaluating present and future potential of arctic fisheries in Canada. Marine Policy, 108, 103637, DOI:

10.1016/j.marpol.2019.103637.” I conceptualized the research questions for this manuscript, collected the data, ran model simulations, conducted analyses, and wrote the original draft version of the manuscript. The initial model was written by WWLC and then revised by me to test the main research questions. All authors contributed to writing and revisions for final publication.

vii

Table of Contents

Abstract ...... iii

Lay Summary ...... v

Preface ...... vi

Table of Contents ...... viii

List of Tables ...... xiii

List of Figures ...... xvi

List of Supplementary Material ...... xxiv

List of Abbreviations ...... xxv

Acknowledgements ...... xxvi

Dedication ...... xxvii

Chapter 1: Introduction ...... 1

1.1 ANTHROPOGENIC CLIMATE CHANGE ...... 2

1.2 CLIMATE CHANGE EFFECTS ON OCEANS ...... 2

1.3 MARINE ECOSYSTEM RESPONSES TO GLOBAL CHANGE ...... 5

1.3.1 Physiological responses ...... 5

1.3.2 Population and ecosystem level responses ...... 9

1.3.3 Marine species adaptation ...... 11

1.4 ANTHROPOGENIC PRESSURES ON MARINE ECOSYSTEMS ...... 12

1.5 MARINE CAPTURE FISHERIES ...... 13

1.6 GLOBAL CHANGE EFFECTS ON MARINE FISHERIES ...... 15

1.7 DISSERTATION OBJECTIVES ...... 16

viii

Chapter 2: Ex-vessel fish price database: disaggregating prices for low-priced species from reduction fisheries ...... 18

2.1 INTRODUCTION ...... 18

2.2 METHODS ...... 21

2.2.1 Data collection ...... 21

2.2.2 Price estimation with the country-product-dummy model ...... 23

2.2.3 Landed values ...... 26

2.2.4 Comparing methods, prices, and landed values ...... 27

2.2.5 Model validation ...... 27

2.2.6 Model assumptions ...... 28

2.3 RESULTS ...... 29

2.3.1 Ex-vessel prices and landed values ...... 29

2.3.2 Comparing prices and landed values ...... 34

2.3.3 Model validation ...... 36

2.4 DISCUSSION ...... 37

2.4.1 Limitations of the global price database ...... 41

2.5 CONCLUSION ...... 42

Chapter 3: Comparing model parameterizations of the biophysical impacts of ocean acidification to identify limitations and uncertainties ...... 44

3.1 INTRODUCTION ...... 44

3.2 METHODS ...... 47

3.2.1 Modelling the effects of global change ...... 48

3.2.2 Characterizing uncertainties ...... 52 ix

3.2.3 Modelled species ...... 56

3.3 RESULTS ...... 58

3.3.1 Projected changes in climate stressors ...... 58

3.3.2 Responses to ocean acidification and global change ...... 59

3.3.3 Sensitivity to uncertainty ...... 66

3.4 DISCUSSION ...... 69

Chapter 4: Ocean acidification amplifies multi-stressor impacts on global marine invertebrate fisheries ...... 75

4.1 INTRODUCTION ...... 75

4.2 METHODS ...... 77

4.2.1 Dynamic bioclimatic envelope model ...... 77

4.2.2 Modelling ocean acidification impacts in a multi-stressor framework ...... 79

4.2.3 Modelled species ...... 82

4.2.4 Model uncertainties ...... 83

4.2.5 Analysis ...... 84

4.3 RESULTS ...... 85

4.4 DISCUSSION ...... 93

Chapter 5: Evaluating present and future potential of Arctic fisheries in Canada ...... 97

5.1 INTRODUCTION ...... 97

5.2 METHODS ...... 100

5.2.1 Study areas ...... 100

5.2.1.1 Canada Eastern Arctic and West Greenland – LME #18 ...... 101

5.2.1.2 Beaufort Sea – LME #55 ...... 102 x

5.2.1.3 Hudson Bay complex – LME #63 ...... 103

5.2.1.4 Canada high-Arctic and North Greenland – LME #66 ...... 104

5.2.2 Taking stock of Canada’s Arctic fisheries ...... 105

5.2.3 Modelled species distribution and abundance ...... 106

5.2.4 Fisheries catch, prices and landed value potential ...... 107

5.2.5 Multi-model scenarios and simulations: characterizing uncertainty ...... 109

5.3 RESULTS ...... 110

5.3.1 Current reported fisheries catch ...... 110

5.3.2 Current and future catch potential ...... 111

5.4 DISCUSSION ...... 118

5.5 CONCLUSION ...... 122

Chapter 6: Conclusion – thesis contributions and implications for current and future research ...... 123

6.1 RESEARCH CONTRIBUTIONS ...... 126

6.1.1 Fisheries databases ...... 126

6.1.2 Ocean acidification modelling ...... 127

6.1.3 Scenario development and identifying high impact areas ...... 129

6.2 CAVEATS AND LIMITATIONS ...... 130

6.3 NEXT STEPS ...... 132

6.3.1 Continuing projects ...... 132

6.3.2 Recommendations for future experimental research ...... 133

6.3.3 Recommendations for future modelling approaches ...... 134

6.4 FINAL REMARKS ...... 135 xi

Bibliography ...... 136

Appendices ...... 169

Appendix A – Supplementary material for “Ex-vessel fish price database: disaggregating

prices for low-priced species from reduction fisheries.” ...... 169

Appendix B – Supplementary material for “Comparing model parameterizations of the

biophysical impacts of ocean acidification to identify limitations and uncertainties.” ...... 173

B.1 Dynamic bioclimatic envelope model ...... 173

B.2 Tables and figures ...... 175

Appendix C – Supplementary material for “Ocean acidification amplifies multi-stressor

impacts on global marine invertebrate fisheries.” ...... 179

Appendix D – Supplementary material for “Evaluating present and future potential of Arctic

fisheries in Canada.” ...... 191

D.1 Dynamic bioclimatic envelope model ...... 191

D.2 Tables and figures ...... 195

xii

List of Tables

Table 2.1. Contribution of forage fishes and the top 12 taxa for FMFO landed value production.

...... 33

Table 2.2. Top 5 countries for FMFO landed value production in 2010...... 34

Table 3.1. Effect sizes (with 95% confidence limits in parentheses) of OA impacts on life history traits (modified from Kroeker et al., 2013)...... 55

Table 3.2. Species analyzed for the impacts of OA...... 57

Table 3.3. Magnitude of the output range of percent changes (with minimum and maximum values in parentheses) in abundance due to ocean acidification (OA) with each source of uncertainty tested in the model.† ...... 68

Table 5.1. Average reported historical marine fisheries total catch and landed value for years

2005-2014. Data from www.seaaroundus.org (Pauly & Zeller, 2015)...... 111

Table 5.2. Model projections of annual catch and landed value potential for current and two future climate change scenarios (low and high) across all four arctic large marine ecosystems.†

...... 112

xiii

Table 5.3. Top 5 species for annual projected catch and landed value for years 2005-2014 and

2091-2100 for two representative concentration pathway (RCP) scenarios across all four LMEs.†

...... 117

...... 171

Appendix Table A.3. Chronological order for estimating prices based on certain criteria and the number of ex-vessel prices in our price database that were matched to each category...... 171

Appendix Table A.4. Linear regression model comparison of WBC average ex-vessel prices trends over time between our model and Swartz et al., (2013), including the number of parameters (K), residual degrees of freedom (Residual DF), Akaike Information Criterion (AIC) value, difference in AIC values (ΔAIC), and AIC weights...... 172

Appendix Table B.1. Sensitivity test of maximum body size to environmental change using varying gill-body size scaling coefficients...... 175

xiv

Appendix Table C.1. Effect sizes (with 95% confidence limits in parentheses) of OA impacts on life history traits (modified from Kroeker et al., 2013)...... 189

Appendix Table C.2. Pearson’s correlation coefficients (r)† between changes in abundance, range

st size, and distributional shift of latitudinal centroid in a high CO2 scenario by the end of the 21 century (2091-2100) relative to the 1951-1960 period...... 190

Appendix Table D.1. List of species modelled in the dynamic bioclimate envelope model...... 195

List of Figures

Figure 2.1. Global marine fisheries weighted-by-catch average ex-vessel prices from 1950-2010 as estimated by the model. Weighted-by-catch average ex-vessel prices are further disaggregated into product usage: direct human consumption, fishmeal and fish oil reduction, and other uses. 30

Figure 2.2. Global marine fisheries (a) landed values ± confidence limits (solid line) and landings

(dashed line) from 1950-2010, and (b) the proportion of landed values derived from fisheries catch destined for direct human consumption (DHC), fishmeal and fish oil production (FMFO), and other uses...... 31

Figure 2.3. The effects of estimating prices separately for different product types versus one overall price, showing average ex-vessel prices for forage fishes...... 35

Figure 2.4. Cross validation where half of the reported prices were removed to estimate the remaining half of the reported prices for DHC (direct human consumption) and FMFO/Other

(fishmeal, fish oil and other uses) based on country, taxa, or at random...... 37

Figure 3.1. a) Conceptual diagram illustrating the scenarios of various relationships between [H+] and survival rate explored in this study, under a hypothetical linear change in acidity over time.

Survival rate is used here as an example that can be applied to other biological parameters impacted by ocean acidification (OA). b) Diagram of how I presented the results by isolating the net impacts of OA on abundance by taking the differences between model outputs with OA and no OA effects...... 54

xvi

Figure 3.2. Projected changes in ocean variables used as the main biophysical drivers in the model. Thin lines are projections from each of the three earth system models used (GFDL, IPSL,

MPI) while thick lines are multi-model means. Projections are smoothed using 10-year running means...... 59

Figure 3.3. Effects of changes in temperature and ocean acidity on life history parameters in

American lobster (H. americanus): a) maximum body weight, b) von Bertalanffy growth parameter K, and growth rate for c) small (100 g), d) medium (250 g), and e) large individuals

(1000 g). Black circles represent initial conditions with no change in temperature or acidity. .... 60

Figure 3.4. Projected changes in species abundance (relative to 2000) due to OA impacts on growth, survival, and both combined for two climate change scenarios using GFDL earth system model. Abundances are smoothed by 10-year running means...... 62

Figure 3.5. Projected changes in American lobster (H. americanus) abundance (relative to 2000) due to OA under different assumptions for the relationship between OA and changes in both life history parameters (growth and survival) for two climate change scenarios, low CO2 (a) and high

CO2 (b), using GFDL earth system model. Abundances are presented as 10 year running averages...... 63

Figure 3.6. Projected changes in abundance (relative to 2000) for model simulations with OA

(solid lines) and without OA (dashed lines) impacts on growth and survival under two climate change scenarios. Changes in abundance are presented as multi-model averages across the earth system model used (GFDL, IPSL, MPI) and smoothed using 10 year running averages...... 65

xvii

Figure 3.7. Biogeographic changes in American lobster (H. americanus) abundance by year 2100 in response to a) changes in all climate stressors (i.e. pH, temperature, O2, primary production), and b) isolated impacts of ocean acidification (OA)...... 66

2011), and (c) impacts on fisheries catch potential (Cheung et al., 2016a). Sopt is the optimal temperature scenario at which aerobic scope is at maximum. Other stressors such as hypoxia shrink the overall performance (blue curve) and reduce the overall aerobic scope (S1) (panel a),

leading to reductions in growth rate and maximum attainable size (w∞) (panel b), as well as potential decreases in fisheries catch (panel c). Multi-stressor impacts of temperature, hypoxia, and acidosis will exacerbate physiological limitations and impacts on fisheries catch (red curve;

S2)...... 77

Effects of projected ocean warming and acidification on aerobic scope for growth. (b) Change in the von Bertalanffy growth curve and maximum body size in the 2091-2100 period. (c) Changes in global maximum catch potential (MCP) projected by the dynamic bioclimatic envelope model.

Results shown are relative to the 1951-1960 period and are multi-model averages from the three earth system models used in this study...... 86 xviii

Figure 4.3. Projected multi-stressor impacts on maximum catch potential (MCP) of marine invertebrates across large marine ecosystems. Results shown are for the 2091-2100 period

(relative to 1951-1960) in high CO2 (RCP 8.5) relative to low CO2 scenario (RCP 2.6)...... 87

GFDL – Geophysical Fluid Dynamics Laboratory; MPI – Max Planck Institute; IPSL – Institute

Pierre Simon Laplace. Black lines and grey bands are selected regressions and 95% confidence limits. MCP data are smoothed by a 10-year running mean and relative to 1951-1960...... 88

Figure 4.5. Biogeographical changes in range size, abundance, and distributional shift of latitudinal centroids for 210 invertebrate fisheries species in the high CO2 scenario. (a)

Responses to global change stressors excluding ocean acidification, and (b) responses to ocean acidification separated from other stressors. Values shown are multi-model means for 2091-2100 period relative to 1951-1960. Correlations between variables are shown in Appendix Table C.1.

Note log scales...... 90

Figure 4.6. Testing model variability for the projected impacts on maximum catch potential

(MCP) due to ocean acidification (OA). (a) Projected changes using the upper and lower bounds of parameter, model, and scenario uncertainty; and (b) the proportion of total uncertainty allocated to each source of uncertainty. Default conditions held constant when testing each source of uncertainty were: 1) parameter = mean OA effect size; 2) model = GFDL earth system model; and 3) scenario = RCP 8.5. Results are smoothed by 10-year running means and relative to the 1951-1960 average...... 93 xix

Figure 5.1. Large marine ecosystem boundaries of Canada’s Arctic and Canada’s Exclusive

Economic Zone (EEZ). Large marine ecosystems: BS = Beaufort Sea; CEA-WG = Canada East

Arctic and West Greenland; HAA = High Arctic Archipelago (Canada high Arctic and North

Greenland); HB = Hudson Bay Complex...... 101

Figure 5.2. Projected maximum sustainable catch and landed value potential in each of Canada’s

Arctic Large Marine Ecosystems. Thin lines represent each model simulation using the different earth systems models, while bold lines are multi models means. Blue lines are projections of the low climate change scenario (RCP 2.6) and red lines are the high climate change scenario (RCP

8.5). Data are smoothed using a 10-year running mean...... 115

Figure 6.1. Thesis contributions and research directions...... 125

Appendix Figure A.1. Average ex-vessel prices over time for various taxon groups using new methods with separate price estimates for different product types (i.e. direct human consumption, fishmeal and fish oil, and other uses), and previous methods with single prices estimates for all product types...... 169

Appendix Figure A.2. Average ex-vessel prices and landed values over time for each country, using new methods with separate price estimates for different product types (i.e. direct human consumption, fishmeal and fish oil, and other uses), and previous methods with single prices estimates for all product types. Linear trends show the value of estimating prices separately in specific countries...... 169

Appendix Figure B.1. Conceptual diagram of the dynamic bioclimatic envelope model...... 176

Appendix Figure B.2. Initial species distribution maps and the abundance (tonnes) for the ten invertebrate species (Table 3.2) used in modeling simulations...... 176

Appendix Figure B.3. Projected changes in ocean variables used as the main biophysical drivers in our model for Northwest Atlantic FAO major fishing area (FAO area 21). Thin lines are projections from each of the three earth system models used (GFDL, IPSL, MPI) while thick lines are multi-model means. Data presented here are smoothed using a 10-year running mean.

...... 177

Appendix Figure B.4. Projected changes in ocean variables used as the main biophysical drivers in our model for Northeast Pacific FAO major fishing area (FAO area 67). Thin lines are projections from each of the three earth system models used (GFDL, IPSL, MPI) while thick lines are multi-model means. Data presented here are smoothed using a 10-year running mean.

...... 177

Appendix Figure B.5. Projected changes in abundance (relative to 2000) for all ten species due to

OA under different assumptions for the relationship between OA and changes in both life history parameters (growth and survival): linear, exponential, and shifting baseline. Results shown are for low and high CO2 scenarios (RCP 2.6 and RCP 8.5) using GFDL earth system model and abundances are presented as 10 year running means...... 178

c, e) show projected environmental changes by year 2100 (relative to 1950) in a high CO2 scenario (RCP 8.5) and show multi-model means from model simulations. Panels on the left (b, d, e) show projected environmental changes over time (relative to 1950) in a high (RCP 8.5) and low (RCP 2.6) CO2 scenario where bold lines represent multi-model means and faint lines are results from individual earth system model projections simulations: GFDL – Geophysical Fluid

Dynamics Laboratory; MPI – Max Planck Institute; IPSL – Institute Pierre Simon Laplace. ... 179

Appendix Figure C.2. Projected environmental changes to surface conditions in a high (RCP 8.5) and low (RCP 2.6) CO2 scenarios. Bold lines represent multi-model means and faint lines represent the individual earth system model projections from: GFDL – Geophysical Fluid

Dynamics Laboratory; MPI – Max Planck Institute; IPSL – Institute Pierre Simon Laplace. ... 180

Appendix Figure C.3. Projected ocean acidification (OA) impacts on the maximum catch potential (MCP) for invertebrate species within different taxonomic groups in addition to other climate change stressors. Outputs are changes in MCP for the 2091-2100 period relative to the

1951-1960...... 181

Appendix Figure C.4. Projected changes in maximum catch potential (MCP) and ocean surface pH in each individual large marine ecosystem. Impacts to MCP were calculated by finding the difference between model runs with and without modelled impacts ocean acidification. Multi- model mean projections are shown for the high (RCP 8.5) and low (RCP 2.6) CO2 scenario and smoothed by a 10-year running mean...... 189

Appendix Figure D.1. Projected maximum sustainable catch and landed value potential in all four of Canada’s Arctic Large Marine Ecosystems tested: Canada Eastern Arctic – West

xxii

Greenland; Beaufort Sea; Hudson Bay complex; and Canada high-Arctic – North Greenland.

Thin lines represent each model simulation using the different earth systems models, while bold lines are multi models means. Data are smoothed using a 10-year running mean...... 197

Appendix Figure D.2. Model and scenario uncertainty for projections of catch potential across all four large marine ecosystems combined and separately. The left column shows the range of results for each source of uncertainty, while the right column shows the proportion of total uncertainty attributed to each source of uncertainty. Model uncertainty addresses the different earth system models, and scenario uncertainty addresses the different representative concentration pathway scenarios (RCP 2.6 and RCP 8.5)...... 198

xxiii

List of Supplementary Material

Appendix Figure A.1.

Appendix Figure A.2.

Appendix Figure B.2.

Appendix Figure B.5.

*Supplementary Material can be accessed from the online UBC cIRcle repository.

xxiv

List of Abbreviations

CC climate change

DBEM dynamic bioclimatic envelope model

EEZ exclusive economic zone

ESM earth system model

FAO United Nations Food and Agriculture Organization

GFDL Geophysical Fluid Dynamics Laboratory (USA)

GOL gill-oxygen limitation

IPCC United Nations Intergovernmental Panel on Climate Change

IPSL Institute Pierre Simon Laplace (France)

MCP maximum catch potential

MPI Max Planck Institute (Germany)

OA ocean acidification

OCLTT oxygen- and capacity- limited thermal tolerance

RCP representative concentration pathway

SAU Sea Around Us

UN United Nations

xxv

Acknowledgements

I would like to acknowledge the numerous people who have helped me along this path of research and the pursuit of a doctorate degree. Foremost, I would like to thank my supervisors,

Rashid and William, and committee members, Chris and Nadja, for their guidance and inspiring me to accomplish this thesis with little to no hiccups and obstacles along the way—it was smooth sailing to say the least! Thanks to my two labs, CORU and FERU, for accepting me even though

I had divided allegiances to both groups. IOF students and students affiliated in other programs made this an enjoyable few years of my life. Inclusive of all those mentioned above, thanks to my family and friends for the moral, educational, and financial support that has let me continue to be a student for three decades and the formative part of my life thus far.

xxvi

Dedication

To life between mountain summits and the ocean abyss.

xxvii

Chapter 1: Introduction

Marine environments have continued to provide fisheries resources to society for much of human history, yet are recently confronted with greater challenges with increased demands and exploitation and rapidly changing environments caused by anthropogenic climate change

(Barange et al., 2014; Pörtner et al., 2014). Understanding how multiple stressors—such as ocean acidification, warming, and deoxygenation—will interact with fishing to affect marine ecosystems is essential to ensure the long-term sustainability of fisheries into the future.

Improving our knowledge of these processes that alter the flow of resources from marine environments to human society will allow us to better prepare adaptation and mitigation strategies into an uncertain future.

This thesis addresses the impacts of ocean acidification (OA) and climate change on fisheries by using tools and models to connect biophysical responses and downstream socioeconomic implications. My motivation for this thesis and major objective is to shed light on mechanisms for large-scale effects of climate change on fisheries to identify further research gaps, highlight the species and geographic regions at greatest risk, and guide research at smaller-scales

(temporal and spatial) that can subsequently alter local scale policy. Below, I provide a brief introduction on the background information essential to conduct my research: the drivers of climate change and ocean acidification, its effects on marine ecosystems and how it may interact with concurrent anthropogenic fishing pressures, and the subsequent effects on food and income security of fisheries-dependent nations and communities.

1.1 ANTHROPOGENIC CLIMATE CHANGE

Accelerated population and economic growth and development of human society in recent decades have increased greenhouse gas emissions and atmospheric concentrations since the

Industrial Revolution in the early 19th century. Rapid environmental change has occurred as a result of the widespread use of fossil fuels (Bindoff et al., 2013). Of the significant greenhouse gases that contribute to radiative forcing for climate change, carbon dioxide (CO2), nitrous oxide

(N2O), and methane (CH4), were reported to have increased by 40%, 20%, and 150%, respectively, since pre-industrial periods (Hartmann et al., 2013). Current atmospheric concentrations of greenhouse gases are reported to be at their highest level in 3 million years and at its highest rate of increase in at least 22,000 years (Masson-Delmotte et al., 2013; Corinne Le

Quéré et al., 2018; Willeit et al., 2019). CO2 alone contributes over 80% of the radiative forcing

(Hartmann et al., 2013) and current studies estimate 9.4 billion tonnes of CO2 emitted annually in the last decade (Corinne Le Quéré et al., 2018). Atmospheric greenhouse gases trap solar energy and regulate global temperature by absorbing thermal radiation, but recent increases of greenhouse gases in the past 200 years have resulted in unprecedented rates of global change and significant atmospheric and ocean warming. Global mean surface temperatures have already increased by an estimated 0.72˚C since 1950, with a large body of scientific evidence suggesting that this warming is due to anthropogenic emissions (Bindoff et al., 2013; Hartmann et al.,

2013).

1.2 CLIMATE CHANGE EFFECTS ON OCEANS

Increases in atmospheric CO2 concentrations and temperatures have translated to significant physical and chemical changes in the ocean, with gases and thermal energy crossing the air-sea 2

interface and circulating through the entire ocean basin (AMAP, 2013). A primary response to increasing greenhouse gas concentrations is ocean warming. Since the 1970s, ocean warming has accounted for over 93% of the increase in earth’s energy inventory as global ocean surface temperatures have increased by 0.11˚C decade-1 between 1971 and 2010 (Rhein et al., 2013).

Furthermore, global sea surface temperatures are estimated to increase by an additional 1.8-4.0

˚C by the end of the century (Solomon et al., 2009; Doney et al., 2012).

The ocean absorbs a significant amount of the anthropogenic carbon emissions from fossil fuel burning as a major global carbon sink (Sabine et al., 2004; Le Quéré et al., 2016)—recent numbers were estimated to be ~25% and up to 40% in the past decade (DeVries et al., 2019).

However, increased CO2 alters seawater chemistry. Dissolved CO2 reacts with water to produce

+ - carbonic acid (H2CO3), which then dissociates into H and HCO3 , increasing hydrogen

2- concentration and acidity and lowering pH. Carbonate (CO3 ) ions act as a buffer and bind to

- some of the newly liberated hydrogen ions, forming bicarbonate (HCO3 ). This reduces the concentration of available carbonate ions, critical for many organisms—as discussed in the next section. Seawater pH has already decreased by 0.1 units from pre-industrial levels, an increase of

26% in acidity (Caldeira & Wickett, 2003; Feely et al., 2009; Lovenduski et al., 2016). Increased rates of atmospheric carbon accumulation could further accelerate climate change and its impacts.

Increased absorption of energy by the ocean triggers multiple other changes including decreases in sea-ice extent, sea level rise, changes in salinity, stratification of the water column, and changes in patterns of ocean circulation (Rhein et al., 2013). Arctic year-round sea-ice extent has decreased at alarming rates, shrinking by over 10% decade-1 (Vaughan et al., 2013).

Atmospheric warming also accelerates further warming for other portions of the cryosphere. For example, between 2002 and 2011, the Greenland and Antarctic ice sheets have lost ice at 215 and

147 Gt year-1, equivalent to global sea-level rises of 0.59 and 0.40 mm year-1, respectively

(Vaughan et al., 2013). In these regions, the bright surfaces of sea-ice and ice sheets reflect solar energy back into the space. Decreased ice coverage will increase the amount of solar energy absorbed by Earth’s surface. Additionally, melting ice sheets in combination with warming ocean have contributed to sea level rise of 3.2 mm year -1 in the past 20 years and a total of 190 mm since 1901 (Vaughan et al., 2013; Nerem et al., 2018).

Warming temperatures also have drastic effects on ocean mixing, which has major biogeochemical implications. Increased temperatures have further increased salinity in mid- latitudinal areas through increases in net evaporation rates, while salinity has further decreased in tropical and polar regions with increased precipitation (Rhein et al., 2013). Changes in surface water density as a result of increased temperatures and salinity changes has likely influenced the long-term trends of ocean surface dynamics (Rhein et al., 2013). Ocean warming has also been linked to many other changes, including increased stratification and decreasing nutrient supply to surface waters (Bopp et al., 2001; Capotondi et al., 2012), and decreased dissolved oxygen concentration which has led to greater numbers of hypoxic dead zones and a vertical expansion of existing hypoxic oxygen minimum zones (Schmidtko et al., 2017; Breitburg et al., 2018).

Conversely, salinity is a stronger driver of mixing and stratification in colder regions, where low salinity-low density water from sea ice melt creates strong haloclines and reducing mixing between deep and shallow layers.

Climate change impacts on weather and increase variability of environmental conditions on short time scales. Increasing variability has led to an increase in extreme weather events, such as marine heatwaves (Hartmann et al., 2013; Hoegh-Guldberg et al., 2014). Increased atmospheric carbon concentrations will continue to drive long-term climate change but also increase short- term environmental variability, and both with have differing implications for marine life.

1.3 MARINE ECOSYSTEM RESPONSES TO GLOBAL CHANGE

Marine life has largely persisted through the stability of the marine environment. With new challenges presented as a result of climate change, we can expect these impacts to manifest as changes from genes to ecosystems (e.g. Pörtner et al., 2014). Tolerance to environmental change is largely determined by physiological performance (Doney et al., 2012), and organisms respond to changes through shifts in physiology and behaviour shaped by their evolutionary history

(Somero, 2012). Specific responses to climate change will vary by organism, but general categories of environmental change triggers similar responses within and sometimes across taxonomic groups. In this section, I describe the physiological responses to environmental change as principle determinants to subsequent impacts on population and ecosystem structure.

The major known biophysical drivers—ocean warming, acidification, and deoxygenation—are described in detail, while other drivers are briefly discussed.

1.3.1 Physiological responses

Environmental change that brings species outside of their optimal range stimulates physiological responses that may affect the biological performance of marine ectotherms such as growth, mortality and reproduction. Physiological changes, whether beneficial or detrimental, potentially

lead to changes in life history traits (e.g. survival, growth, reproduction), and even population dynamics and ecosystem structure. Marine species may respond to environmental variability through acclimatization in which the organism adjusts to the environmental change through physiological mechanisms (Doney et al., 2012; Boyd et al., 2016). Environmental change outside of the tolerable range may lead to a decrease in performance and increase in mortality rate, consequently leading to decrease in population abundance and productivity. This may then result in spatial and temporal reorganization in the form of shifts in distribution (migration to tolerable habitats) or changes in phenology (timing of seasonal events) (Parmesan, 2006; Doney et al., 2012; Cheung & Pauly, 2016). Alternatively, some species or populations may benefit from environmental change within the tolerable range. Increased food and resources, reduced physiological demands for maintenance (e.g. respiration, acid-base balance), increased competitiveness for resources, and increased ability to avoid predation (Doney et al., 2012;

Pörtner et al., 2014; Sunday et al., 2017) may increase the capacity for individual and population growth.

Ocean warming, one of the most evident changes, has been linked to a number of various organism, population, and ecosystem level responses (e.g. Pörtner et al., 2001, 2014; O’Connor et al., 2007; Pörtner & Knust, 2007; Dulvy et al., 2008; Gooding et al., 2009; Hein et al., 2012).

Biochemical reactions that drive organism’s metabolism scale positively with temperature and the relationship generally follow the Arrhenius equation (Bruno et al., 2015). However, other factors such as organism’s thermal tolerance, nutrient and oxygen availability may limit growth and production and shape their relationship with temperature (Doney et al., 2012). Changes in temperature beyond the organism’s tolerance range will compromise biological functions (e.g.

Pörtner & Knust, 2007). Furthermore, warmer temperatures can increase energy demand to the point where it exceeds energy intake, compromising growth, reproduction, and spawning success

(e.g. Farrell et al., 2008).

OA and the resulting biogeochemical changes may have profound effects on marine life. It reduces carbonate ion concentration and increases carbonate solubility, likely increasing the cost of calcification for many shell-forming organisms (e.g. mussels, corals, oysters) (Feely et al.,

2004; Fabry et al., 2008). Aragonite saturation state has quickly become a commonly used measure of calcium carbonate solubility and is a measure of the thermodynamic potential for the mineral to form or dissolve (Mathis et al., 2015). Aragonite is one of the more soluble forms of calcium carbonate and is widely used by marine calcifying organisms. Current literature has identified an aragonite saturation state of 3 as a threshold for susceptibility of calcifiers to OA

(e.g. Ricke et al., 2013; Gattuso et al., 2015). Furthermore, if the saturation state falls below 1.0, seawater is corrosive to calcium carbonate and can result in deformed shells (Watson et al.,

2009; Bechmann et al., 2011; Crim et al., 2011; Kroeker et al., 2013). For calcifying organisms,

OA results in decreased rates of calcification, growth and survival (Kroeker et al., 2013). In fish the impacts of OA are less known and highly variable. A few physiological responses have been identified, such as effects on olfactory sensory abilities affecting homing abilities and recruitment (Munday et al., 2009a, 2010; Ellis et al., 2016; Pistevos et al., 2016), otolith development (Checkley et al., 2009), and basal metabolic rates (Munday et al., 2009b). Acidified waters can also disrupt acid-base balance for many organisms, affecting other critical metabolic processes (Pörtner, 2010). Some higher-level organisms are well adapted to maintain internal pH, while other organisms may be more susceptible to ocean acidification (Doney et al., 2012).

Not all organisms will experience negative effects due to elevated CO2. Some organisms have shown increased photosynthesis and growth with increased CO2, but this is not ubiquitous across photosynthetic organisms (Kroeker et al., 2013). Furthermore, there are instances of minimal to negligible responses to ocean acidification across a variety of organisms, but our understanding of interactions between multiple stressors remain uncertain (Kroeker et al., 2017).

Climate change has contributed to the vertical expansion and the increase in the number of oxygen minimum zones (Diaz & Rosenberg, 2008; Stramma et al., 2008), inducing hypoxic conditions where dissolved oxygen concentrations fall below 60-120 µmol kg-1 (Stramma et al.,

2008). Most organisms rely on aerobic metabolism as their primary source of energy, and prolonged hypoxic or anoxic conditions can only be sustained temporarily, and tolerance generally favours larger animals (Pörtner et al., 2014). However, microbes are able to continue to consume ambient oxygen to very low levels, sustaining oxygen minimum zones and further depleting oxygen (Pörtner et al., 2014). Organisms may cope with hypoxia by reducing activity to conserve oxygen, although this strategy is rarely found in fish and more common for invertebrates (Pörtner & Farrell, 2008; Doney et al., 2012). Reduced activity may lead to reduced feeding and consequently growth, reproduction, and survival (Doney et al., 2012).

Responses to multiple stressors—primarily temperature, salinity, pH—show synergistic interaction effects in majority of studies (Przeslawski et al., 2015). However there has yet to be a unifying theory on the physiological mechanisms to environmental change (Lefevre, 2016).

Nonetheless, global change is altering many abiotic factors at once and these different stressors will affect marine ecosystems simultaneously.

1.3.2 Population and ecosystem level responses

Climate change impacts directly affect physiological processes with downstream impacts at population and community levels. Ocean warming has already shown considerable impacts, leading to shifts in distribution and abundance (Doney et al., 2012; Pörtner et al., 2014). Shifts in species distributions to deeper and higher latitudinal waters have been observed (Perry et al.,

2005; Dulvy et al., 2008; Poloczanska et al., 2016). These shifts in distribution are an effective strategy for some organisms, but species are limited by three main factors: dispersal capacity, abiotic conditions, and biotic factors (Doney et al., 2012). The ability of a species to move into new habitats is limited by its dispersal capacity as determined by its evolutionary history. For many sessile demersal invertebrates, dispersal is limited to broadcast spawning and its pelagic larval duration (O’Connor et al., 2007). Abiotic conditions must be within the physiological tolerance of a species and thus its fundamental niche (Doney et al., 2012). Lastly, biotic factors such as competition and predation may restrict dispersal and colonization of new habitat in response to climate change.

Another ecological consequence of ocean warming is shifts in phenology. There have been large shifts in the timing of zooplankton biomass formation (e.g. Mackas et al., 1998; Schlüter et al.,

2010), shifts in salmon migration (Kovach et al., 2012), and widespread ecosystem shifts in the timing of biological events of all major taxonomic groups (Thackeray et al., 2010). Decoupling of phenological events between species poses significant implications for trophic interactions, especially for higher trophic level organisms that depend on the annual impulse of planktonic biomass formation (Edwards & Richardson, 2004). Changes in upwelling dynamics also has implications for primary and secondary producers and trophic interactions (Harley et al., 2006).

High rates of offshore transport of herbivorous zooplankton tends to favour primary producers in near-shore systems (Bakun, 1990), altering species composition and potentially ecosystem function.

Ecosystem impacts of ocean acidification will likely be defined by large-scale changes to key, ecologically important species. However, the difficulty of distinguishing ocean acidification impacts from other large ecosystem drivers such as temperature remains a significant challenge.

Ecosystem impacts of ocean acidification are highly variable and not well understood. Observed declines in coral calcification have been attributed to temperature spikes, but may also include impacts due to ocean acidification (Pörtner et al., 2014). On the other hand, in some systems with increasingly acidified waters, no changes have been observed, likely attributed to pre- adapted species to natural fluctuations of highly acidified waters (e.g. upwelling) or the difficulty of detecting such small changes (Pörtner et al., 2014). Despite the lack of resolution of ocean acidification impacts on ecosystems, it is certain that ecosystems built from calcified structures

(i.e. warm-water coral reefs and their cold-water equivalent) will be at most risk to dissolution due to ocean acidification (Pörtner et al., 2014; Sunday et al., 2017). Thus, there is a need for studies that isolate the impacts of ocean acidification from other large ecosystem drivers.

Expansion and increase of oxygen minimum zones and areas characterized by low dissolved oxygen concentrations may lead to habitat compressions and a reduction in the available habitat for hypoxia-sensitive taxa (Stramma et al., 2010, 2011; Breitburg et al., 2018). This may ultimately lead to a loss in biodiversity and changes in ecosystem structure (Stramma et al.,

2011). Distributions of major zooplankton and nekton species will be further constrained by expanding oxygen minimum zones (Ekau et al., 2010). In one of the most productive marine

systems off the coast of Peru, dissolved oxygen contributes to shaping the alternating fluctuations in distribution and abundance between anchovies and sardines (Bertrand et al.,

2010). In general, hypoxia threatens population survival (Ekau et al., 2010), constrains fish stocks and large pelagics (Stramma et al., 2010), and reduces biodiversity (Stramma et al.,

2011), amongst other impacts (Pörtner et al., 2014).

1.3.3 Marine species adaptation

Global environmental change will drive changes to organism biology and potentially lead to adaptation. However, the capacity to adapt to change will vary across species. Species with long generation times, i.e. years to decades, may be limited in their adaptive capacity and any adaptive selection is likely to occur from existing genotypes and phenotypes (Denman, 2017).

Alternatively, species with short generation times (i.e. days to weeks, or less) will have plenty of capacity to adapt through existing genotypes and phenotypes, as well as through mutations that result in favourable traits. Therefore, adaptive capacity manifests differently across organisms and may correspond to body size. We can expect larger species (e.g. fishes) with generally long generation times to have limited capacity for adaptation and smaller species (e.g. phytoplankton) with short generation times to have more capacity for adaptation. Furthermore, species with long generation times will need to respond to the challenges in long-term trends (i.e. climate change- related stressors) as well as increased variability in extreme events (i.e. weather), while species with short generation times will mostly need to only respond to short-term variability and extremes.

1.4 ANTHROPOGENIC PRESSURES ON MARINE ECOSYSTEMS

Marine environments have been important throughout human civilization as estuarine and coastal seas have been central for population settlement (Small & Nicholls, 2003; Lotze et al.,

2006). Threats to marine environments are a direct result of human dependence on marine ecosystem goods and services including physical products (food, fuel, biochemical products), maritime transport, and cultural services (Pörtner et al., 2014). Among those, some of the most critical have been identified as overexploitation, habitat loss/degradation, and pollution (Crain et al., 2009), and these anthropogenic pressures are likely to exacerbate the impacts of climate change on marine environments (IPCC, 2014).

Humans have relied on marine resources for much of their history. However, this has led to overexploitation and overfishing of many marine fish stocks, and destructive fishing practices

(e.g. bottom trawling) have had considerable impacts on habitat (Halpern et al., 2008; McCauley et al., 2015). It is estimated that over one-third of commercial fish stocks are overexploited

(Srinivasan et al., 2010, 2012; FAO, 2018). Over-exploitation is considered to be the main drivers of extinction risk of marine species (Dulvy et al., 2003; Kappel, 2005; IPBES, 2019).

Degradation and direct removal of coastal habitats around the world are considered to be the largest threats to marine environments as it physically alters or completely removes ecosystem structure (Lotze et al., 2006; Crain et al., 2009). Coastal engineering has been one of the largest contributors (Crain et al., 2009). Current estimates of coastal habitat loss are 65% of wetlands,

65% of seagrass beds, and 48% of other submerged aquatic vegetation due to human settlement

(Lotze et al., 2006). Habitat degradation, even as patchy distributions, can drastically alter ecosystem structure and has led to population-wide collapses (e.g. Noonburg and Byers, 2016). 12

Human habitation across coastal margins has inherently increased pollution in marine environments. Inputs from synthetic fertilizers (e.g. nitrogen, phosphorus) and fossil fuel burning has led to eutrophication and hypoxia in coastal waters (Crain et al., 2008; Howarth, 2008).

Eutrophication can lead to shifts in producer assemblages, shifting competition and increasing vulnerability to species invasions, as well as elevate microbial aerobic metabolism creating hypoxic conditions (Williams & Smith, 2007). Low dissolved oxygen waters from coastal estuarine inlets can then extend to open marine waters (Crain et al., 2009). In addition, marine ecosystems are subject to many other pollutants produced by human societies. Other major pollutants from human-made products include persistent organic pollutants (e.g. DDT, PCBs), oil pollution, heavy metals, and plastics. These pollutants can accumulate in organisms and increase in concentration at higher trophic levels (biomagnification), with potentially detrimental consequences to these organisms and even for humans as consumers of many larger species of fish (Islam & Tanaka, 2004; Crain et al., 2009). Additionally, these pollutants can lead to physical deformities, blindness, cancer, and generally increased mortality (Islam & Tanaka,

2004; Crain et al., 2009). While these anthropogenic pressures will likely have considerable effects on marine ecosystems, climate-related stressors (temperature, acidification, and deoxygenation) and capture fisheries are expected to be the main drivers of large-scale global change (Pörtner et al., 2014; IPBES, 2019).

1.5 MARINE CAPTURE FISHERIES

Human society’s dependence on marine ecosystems has been etched in our history, none being more historically important than food derived from marine environments. Records of global marine fisheries estimate catch amounts to be around 110 million tonnes annually in the past

decade (2005-2014; seaaroundus.org) (Pauly & Zeller, 2015). However, marine capture fisheries have declined since it peaked in the 1990s at ~130 million tonnes (Pauly & Zeller, 2016).

Declines in catch are an indication of a decrease in the global supply of fish despite the expansion of industrial fisheries to distant waters and to developing countries (Swartz et al.,

2010). While it may be too early to sound the alarms from the observed trends of declining fish catch, it is important to note the significance of fisheries for human society and the global economy (Sumaila et al., 2012).

Fisheries and aquaculture currently provide about 17% of the world’s animal protein supply for humans and has continued to grow, and in some developing nations it provides over 50% of their animal protein intake (FAO, 2014a, 2018). Approximately 1 billion people in Southeast Asia rely primarily on seafood for animal protein. Marine capture fisheries are especially important for coastal indigenous communities, where they consume an average of 15 times more per capita than non-indigenous communities (Cisneros-Montemayor et al., 2016).

The annual economic value of marine capture fisheries at the first point of sale has been previously estimated to be between $80-85 billion USD, and estimated to have a wider indirect economic impact almost three times greater at $225-240 billion USD (Sumaila et al., 2007; Dyck

& Sumaila, 2010). Marine fisheries are crucial in terms of jobs and income as ~50 million people are employed directly as fishers either in small-scale or industrial sectors; 22 million people are employed in the small-scale sector (Teh & Sumaila, 2013). A total of 260 million people are employed in marine capture fisheries including jobs indirectly related to fishing (e.g. processing, distribution) (Teh & Sumaila, 2013). Marine capture fisheries are especially important to the economy and society of developing regions. People from developing regions make up 78% of

the direct employment as fishers (Teh & Sumaila, 2013), and small-scale fisheries in developing regions contribute over 75% of domestic fish supply (FAO, 2014a). Impacts to fisheries will have major implications globally, but especially in fisheries-dependent communities in developing nations.

1.6 GLOBAL CHANGE EFFECTS ON MARINE FISHERIES

Climate change and its effects on the physical conditions (e.g. temperature), chemistry (e.g. acidity), and biology of ocean environments are creating new challenges for fisheries. Ocean warming has led to changes in the composition of fisheries catch toward a higher proportion of warm water species (Cheung et al., 2013a). Documented OA effects on fisheries are limited to a few examples, and assessments of its potential effects are mostly extrapolated from laboratory experiments. We know that species that form calcium carbonate shells (e.g. oysters, mussels, corals) show the greatest sensitivity to OA (Nagelkerken & Connell, 2015; Kroeker et al., 2017) and thus these fisheries are likely at greatest risk to OA. During an upwelling event in the

Northeast Pacific Ocean, massive die-offs of larval oyster being reared at an aquaculture farm were correlated to the aragonite saturation state of the inflowing water (Barton et al., 2015).

Potential effects on fisheries in future levels of OA have largely focused on shellfish fisheries and identified these to be at greatest risk (e.g. Cooley et al., 2015; Ekstrom et al., 2015; Haigh et al., 2015; Mathis et al., 2015). Furthermore, OA could have far-reaching indirect impacts on fisheries (Le Quesne & Pinnegar, 2012), yet our ability to isolate cause-and-effect relationships remain limited.

Marine fisheries are a significant source of food, income, and culture, and the wellbeing of dependent communities and nations will likely be threatened by climate change (Pörtner et al., 15

2014). Global change will directly affect marine fisheries catch, yet it is important to understand the downstream effects such as impacts on the economy. Quantifying the value of marine fisheries resources and how it is affected by global change is important to society and local, national, and international governing bodies to develop mitigation and adaptation strategies

(Gattuso et al., 2018).

Global projections of climate change effects on fisheries are important developments to identify areas at greatest risk and potential patterns of change. Large shifts in species distributions and catch potential are expected with greater climate change (Cheung et al., 2010). Changes in catch are projected to be minimal if countries meet the target of limiting atmospheric warming to 1.5

˚C (Cheung et al., 2016a), set at the 2015 meeting in Paris for the United Nations Framework

Convention on Climate Change. Furthermore, meeting this target would minimize downstream effects on society, reducing impacts on revenues, incomes, and consumer fish prices (Sumaila et al., 2019)

1.7 DISSERTATION OBJECTIVES

This work aims to assess the impacts of OA and climate change on the future of fisheries. First, I present the development of the various tools and databases used to address this overarching issue. Impacts of OA and climate change on fisheries will vary across time and space, and we expect effects on biological systems to trickle down to society and the economy. Identifying interdisciplinary methods to incorporate these factors is an important development. As global scale changes to climate differ in type and magnitude to that of local and regional scale changes, the resulting impacts will differ in ways we must be able to predict and understand. My research addresses the following general questions: 16

1. How can the value of fisheries be accurately estimated and used in economic analyses,

and how are these likely to change in the face of increasing ocean acidification?

2. What are the biophysical and socioeconomic impacts of ocean acidification on the future

of marine fisheries?

3. How can we use different indicators and methodologies to assess the impacts of ocean

acidification at various spatial scales?

Chapter 2 identifies revised methods to increase the accuracy for estimates of ex-vessel marine fish prices. Prices are separately estimated for the various fisheries products: direct human consumption, fishmeal and fish oil, and other products. This lays the foundation for the subsequent effects of ocean acidification on global fisheries economies. In Chapter 3, I explored various parameterizations and assumptions of how OA will affect species, and how that scales up to abundance and distribution. Chapter 4 looked at the impacts of OA on global invertebrate fisheries, the species groups likely to be most sensitive to OA. In chapter 5, I assess the current and future potential of fisheries catch and landed values in Canada’s Arctic. Lastly in chapter 6, I summarize the theme of my thesis and how it contributes to the scientific community and public, as well as where this contribution will inform future research in this interdisciplinary field.

Chapter 2: Ex-vessel fish price database: disaggregating prices for low-priced species from reduction fisheries

2.1 INTRODUCTION

Initial construction of a global marine ex-vessel fish price database (hereafter referred to as the

Price DB) by Sumaila et al. (2007) was to address the lack of appropriate information required to sustainably manage natural resources, as prices paid to fishers are an integral piece of information as a main determinant of fishing behaviour. Ex-vessel prices are the prices that fishers receive directly for their catch, or the price at which the catch is sold when it first enters the supply chain. Therefore, in order to effectively manage the sustainable use of fisheries resources, managers and policymakers need reliable information on ex-vessel prices. The purpose of such a database is to provide ex-vessel prices for fisheries scientists to conduct socioeconomic analyses at various spatial scales—i.e., global, regional, and national-scale analyses. Here, I present an updated version of the database with major improvements by incorporating new estimates for low-priced species caught by reduction fisheries.

The first construction of the Price DB provided a complementary list of fish and market specific ex-vessel prices for each recorded catch in the Sea Around Us (SAU) catch database (Watson et al., 2004). It involved collecting data from widely scattered national and regional statistical reports from published and grey literature (e.g., governmental agencies, websites). Sumaila et al.

(2007) amassed over 30,000 records of reported ex-vessel prices, but could only directly assign a price to 18% of total tonnage landed from the SAU catch database. They devised a rule-based approach to estimate missing prices, using a combination of various assumptions to relate fish

prices across taxa, countries, and years. Such an extensive database, as they first noted, requires continuous updates over time for both the input data (e.g., increasing records of reported ex- vessel prices and increasing the diversity and evenness of the country sources for the data) and price estimation methodologies (i.e., revising the underlying assumptions to provide more accurate estimates).

The Price DB was next updated by Swartz et al. (2013) to address various limitations with the methodologies for estimating ex-vessel prices. For example, the rule-based approach could match prices from one species to a profoundly different species (e.g., from herring to tuna) in countries with few reported ex-vessel prices. Additionally, price estimates were also derived from data across multiple years, failing to account for inter-annual differences in market prices.

To address these concerns, Swartz et al. (2013) devised a methodology to estimate ex-vessel prices using the country-product-dummy model (Summers, 1973), a multilateral method used by the International Comparison Programme to deal with incomplete matrices and estimate price level and missing commodity prices. The country-product-dummy model addresses year-specific differences in market prices and prioritizes taxonomy for price estimation, where estimates are derived from price data sourced from other countries.

The Price DB has contributed to recent global and regional fisheries economic analyses (e.g.,

Börger et al., 2014; Nunoo et al., 2014; Teh and Sumaila, 2015), which underscores the importance of providing updated and accurate ex-vessel prices for research related to fisheries management and policy. First, I address a major concern identified by Swartz et al. (2013) where prices for low-value fishes destined for reduction were not distinguished from fishes for human consumption and thus were likely overestimated. I constructed a separate database of ex-vessel

prices of fisheries catches destined for purposes other than human consumption. These price estimates are a result of the work of a recent effort to disaggregate catches by product usage: direct human consumption (DHC), fishmeal and fish oil (FMFO) production, and other uses including bait, direct feeding, and industrial uses (Cashion et al., 2017). Together, fish destined for FMFO and other uses are some of the largest fisheries globally in terms of catch in weight, such as the Peruvian anchoveta fishery, and account for approximately 20 million tonnes of annual landings presently (Pauly & Zeller, 2016). While the ten largest taxa for FMFO account for 77% of landings destined for this purpose historically, there is a growing diversity of species used for FMFO and especially for direct feed in aquaculture and thus accounting for this diversity of end uses and prices is an important development (Cashion et al., 2017).

Aquaculture remains one of the largest consumers of global fishmeal (68%) and fish oil (89%) products (Tacon & Metian, 2015). However, there is a growing trend to use alternative sources of feed such as soy protein (Naylor et al., 2009; Salze et al., 2010), and the proportion of catch to

FMFO production has decreased in some of the major reduction fisheries (e.g. Christensen et al.,

2014). With global fish stocks in decline (Myers & Worm, 2003; Srinivasan et al., 2010; Pauly

& Zeller, 2016), we may expect changes in supply and demand of reduction fisheries and thus an effect on prices and other substitutable products. Further, rising costs of FMFO feed may support the shift to more sustainable feeds (Tacon & Metian, 2008; Naylor et al., 2009; Pikitch et al.,

2014).

I further updated the Price DB by providing uncertainty estimates of 95% confidence limits for prices, a valuable addition for researchers looking to use the price database. I was able to revise the use of the country-product-dummy model to provide uncertainty estimates, an important

application overlooked by Swartz et al. (2013). Other updates included increasing the number of input reported ex-vessel price records and extending the database to 2010 to match the SAU catch database. In addition to the extension of catch records from 2007-2010, the SAU catch database has gone through major reconstruction efforts and includes 2.5 times more records of unique taxon-country-year marine fisheries catch than the previous version of the Price DB

(Pauly & Zeller, 2016).

2.2 METHODS

2.2.1 Data collection

The first step in my data collection effort was to identify additional price data, and obtain updated versions from many of the same sources used by Sumaila et al. (2007) and Swartz et al.

(2013). I sourced data from governmental agencies, web sites, published and grey literature, and contacted partners around the world who helped locate data in their particular region (Appendix

Table A.1). I collected an additional ~20,000 reported prices for this price database for a total of over 60,000 reported prices spanning the years from 1950 to 2010 (Appendix Table A.2). I filtered out the top 0.5% of data to remove any extremely high reported prices (exceeding 38,000 real 2010 USD tonne-1) from the database that would then interfere with the estimation model.

The top 0.5% was removed as many of these prices exceeded consumer market prices, especially those exceeding 100,000 USD tonne-1. Many of these prices were also calculated from landings that were very small (i.e. <0.1 tonne) and therefore would have very little weight on average prices at the global scale and may reflect overinflated prices if regional stocks were low. While such prices are interesting when evaluating regional economic trends for individual fisheries, the database is tailored to large-scale global analyses and thus I removed these outliers. I also 21

removed prices that were calculated to be $0 tonne-1—in many instances landed value was recorded as $0 and thus a price of $0 tonne-1. These landed values were likely recorded as zero if they did not receive a price at the point of sale or the landed value amounts were too small to match the resolution of the units (data were presented as thousand USD). In total, less than 500 reported price records were removed.

The same sources used to collect DHC prices were used to collect reported ex-vessel prices of fish destined for non-DHC uses, as well as reports specifically on FMFO production (Appendix

Table A.1). Reported prices were explicitly listed as for ‘reduction to fishmeal and fish oil’. I attempted to find reported prices covering the years from 1950 to 2010 on major reduction taxa and taxa used for direct feed in aquaculture. I estimated prices of taxa by applying the reported price of ‘trash fish’ to known taxa used for reduction purposes in the reporting country (see

Cashion, 2016). I collected a total of over 2,600 prices of fish for non-DHC uses providing adequate regional, temporal, and taxonomic coverage of major reduction species and taxa used for other non-DHC uses (Appendix Table A.2). Again, I filtered out the top 0.5% of data to remove extremely high reported prices (exceeding 3,280 real 2010 USD tonne-1) and any prices calculated to be $0 tonne-1. Fourteen reported price records were removed from the analysis.

The SAU catch database (http://www.seaaroundus.org/; Pauly and Zeller, 2015, 2016) was used to create a list of prices to be estimated. My goal was to provide an estimated ex-vessel price for each listed catch in the SAU catch database. This newly reconstructed version of the catch database provides a detailed list of catches from 1950-2010.

I determined market exchange rates and purchasing power parities from the Penn World Tables,

Version 7.1 (http://www.rug.nl/research/ggdc/data/pwt/; Heston et al., 2012). Purchasing power 22

parity (PPP) is a measure of the relative price level in a particular country, and accounts for the fact that the relative cost of goods in a country may not be fully reflected in market exchange rates. I used PPP to convert domestic prices into ‘real’ prices for a direct international comparison of value, a method used by the International Comparison Program (Rao, 2004). For years where a country’s PPP was unavailable (i.e., earlier years where price level data was scarce for developing countries), I used the price level index—equal to the PPP divided by the market exchange rate—from the nearest year (see Swartz et al., 2013). Regional averages of the price level were calculated for countries with no PPP data (Swartz et al., 2013). I accounted for inflation by standardizing all ex-vessel prices to a reference year using the United States (US)

Consumer Price Index (CPI) prepared by the US Bureau of Labour Statistics

(http://www.bls.gov/cpi/).

2.2.2 Price estimation with the country-product-dummy model

Reported ex-vessel prices were assigned to match taxon-country-year specific catch from the

SAU database where possible. However, the majority of catch records in the SAU catch database were not recorded with an associated price or landed value. As in the previous version of the

Price DB I estimated these ‘missing’ prices using the country-product-dummy model (for details on the country-product-dummy model, see Swartz et al., 2013). Here I describe a revised application for the use of the country-product-dummy model to estimate ‘missing’ prices and

95% confidence limits (Rao, 2004; Silver, 2009); confidence limits were not estimated in the previous version. Confidence limits were derived from applying a regression model to the data, described below.

Ex-vessel prices (denoted by pijt) of each unique commodity-country-year combination were estimated based on three factors: the price level of the i-th commodity (fish taxa) in year t relative to other commodities, also referred to as the international price (denoted by IPit); the general price level of the j-th country in year t relative to other countries, also referred to as the purchasing power parity (denoted by PPPjt); and a random variable with a lognormal distribution

(denoted by uijt). Using these determinants, the country-product-dummy model can be stated as

(Summers, 1973):

�!"# = ��!" ∙ ��!" ∙ �!"# (2.1)

I estimated an international price for each taxa i and year t by taking the natural logarithm of model (1) yielding the dummy model:

�!"# = �!" + �!�!! + �!�!! + ⋯ + �!�!" + �!"# (2.2)

where n = the number of countries (observations), j = 1,2,…, n, and yijt = the natural log of the reported price. The known coefficient x of the dummy variable Bjt is the natural log of the country effect on price (PPP), and the estimated coefficient ait = natural log of the international price. The error term εijt in the dummy model is normally distributed with mean 0 and constant variance σ2 and related to u in equation (2.1) such that ε = ln (u).

To estimate ‘missing’ ex-vessel prices I first matched all raw reported price data (pijt) from various countries within year t to the taxonomic category of interest (i.e. matched to the species level). Next, I matched known purchasing power parity (PPPjt) values to reported ex-vessel prices from each j-th country. I then fit the dummy model using equation (2.2) to estimate a and

from this I calculated international prices (IPit)—expressed in a common currency, US dollars— for each i-th taxon. The relationship (i.e. slope) between PPP and price in equation (2.1) is equal to the exponent of the coefficient a, also equal to the international price. I ran separate regression models for each year to maintain intraspecific year-to-year variability and eliminate any effects of price data from other years. I used linear least squares regression function in R statistical program to estimate coefficient a, as well as the 95% confidence limits (R Core Team, 2018).

Next, I calculated ‘missing’ ex-vessel prices (pijt) for each i-th commodity in country j in year t with equation (2.1) using the estimated international price (i.e. coefficient ait) and PPPjt. I used equation (2.1) to estimate 95% confidence limits for ‘missing’ ex-vessel prices (pijt) from estimates of 95% confidence limits for international prices and the PPPjt.

Where there was no specific reported price data for a particular taxon, I matched raw reported price data based on higher taxonomic classifications (e.g. genus, family), functional groupings

(e.g. large pelagic), and habitat types (e.g. benthic) (as in Sumaila et al. (2007) and Swartz et al.

(2013)) (see Appendix Table A.3 for stepwise schematic). For cases with no matching input data for a taxon-year combination, I estimated prices by finding the average international price of the taxa for all years where it was estimated, accounting for inflation using the US CPI. I then used the CPI to back-convert the average price to the year of interest. Finally, the last estimation step was to use the median for all international prices estimated within the year of interest. My estimation methods prioritize retaining inter-annual variability in market-specific prices.

Where my methods differ from previous versions of the Price DB is the price estimation for various fisheries end-products. Separate ex-vessel prices were estimated for the proportion of catches destined for DHC and for purposes other than DHC (i.e., FMFO and other uses) using

separate input data sets. Ex-vessel prices for purposes other than DHC were assigned to the proportions of catch destined for FMFO and “other” uses. I assumed that prices between catch destined for FMFO and other uses are similar and therefore applied the same input reported price data set to both. “Other” uses represent a small proportion of landings and their price data are often not published. I estimated an international price for over 87,000 and 77,000 unique taxa- year combinations for catches destined for DHC and FMFO/other uses, respectively.

2.2.3 Landed values

The parallel construction of this Price DB with the global SAU catch database (Pauly & Zeller,

2015) allowed me to calculate landed values for each catch using the constructed Price DB.

Catches are broken down by product usage and landed values were calculated using ex-vessel prices for DHC, FMFO, and other uses (Cashion, 2016). Each catch is designated to a fishing entity and taxonomic group. Therefore, I was able to quantify the landed values by the top fishing nations and major taxonomic groups from 1950-2010. This allowed me to determine the distribution of the value of global marine fisheries resources. I compared the landed value by destination type for the specific taxonomic groups that comprise the majority of revenues for uses other than DHC. Additionally, I compared landed values for the top fishing nations for

FMFO production value and its relative contribution to the country’s economy. Values of 95% confidence limits for price estimates were carried through to estimate confidence limits for landed values.

2.2.4 Comparing methods, prices, and landed values

First, I compared ex-vessel price estimates and landed values between methods with and without separate input prices for fisheries destined for FMFO and other uses. Second, I compared the percent differences in the landed values for these top 12 species using my methods (separate prices for non-DHC purposes) versus previous methods (one price for all purposes). Lastly, I applied a linear regression model to compare the average price and landed value trends over time between the two methods for each country.

I used the exchange rates from the Penn World Tables to convert prices from domestic currencies into USD. I used market exchange rates instead of PPP to compare market prices across currencies, and not the “real international” value of ex-vessel prices (Swartz et al., 2013).

I converted prices to real 2010 values using the US CPI to account for inflation, and I assumed that the relative PPP values capture country-specific inflation. These conversions allowed for comparisons to be made across countries and over time.

When comparing average ex-vessel prices, I used weighted-by-catch means instead of normal mean calculations. Weighted-by-catch means were calculated by taking reported catch and multiplying it by price to obtain the landed value, then dividing the sum of all landed values by the sum of all reported catches.

2.2.5 Model validation

My model was validated using a k-fold cross validation by separating half of the reported prices

(training data) to estimate the remaining half of the reported prices (test data). I generated three subsamples, each representing ~50% of the data to measure how well my model estimated prices 27

across countries, taxa, and overall. I removed price data from 35 randomly selected countries,

724 randomly selected taxa, and a random selection of half the data as training data for DHC prices. I did the same for FMFO/Other prices and removed price data from 13 randomly selected countries, 122 randomly selected taxa, and a random selection of half the data. I used Pearson’s correlation coefficient to test how well my model was able to estimate the corresponding reported prices.

2.2.6 Model assumptions

It should be noted that constructing such an extensive database relies on many assumptions, which creates uncertainties in the estimations. One of the main assumptions I applied is using input prices within the same year rather than prices with a more similar taxonomic match from other years. While this retains year-specific market prices, it assumes that prices can be transferred across higher taxonomic classifications. However, I assume that raw ex-vessel price data were available for the major fisheries of the world, and any fisheries catch data without reported prices were simply ‘substitutes’ of their related taxa (Swartz et al., 2013). My methods can also produce conversion errors. Some countries have gone through multiple currency changes (e.g., Chile), and their exchange rates and PPPs may be vastly different at the beginning of the year compared to the end of the year. Therefore, depending on when these prices were recorded, the reported ex-vessel prices may be over- or undervalued when converted to an international price, which can then be carried over to price estimates for other countries.

In using the country-product-dummy regression model to estimate missing prices and confidence intervals, I assume that the random variable term uijt in equation (2.1) has a lognormal

distribution and thus the error term ε in equation (2.2) is normally distributed. This assumption allows me to use the Student’s t-distribution table to estimate confidence intervals using equation

(2.2) (Summers, 1973). With lognormal distribution of errors, the estimates of international prices and confidence intervals can only take on positive real numbers and will be right-skewed, resulting in a larger upper 95% confidence limit relative to the lower confidence limit.

Nonetheless, filling incomplete prices using the country-product dummy model has been shown to be superior to other methods commonly used (Rao, 2004).

All models and analyses were run using statistical programming software R (R Core Team,

2018) and the ‘dplyr’ R package (Wickham & Francois, 2016).

2.3 RESULTS

2.3.1 Ex-vessel prices and landed values

I estimated ex-vessel prices for over 667,000 and 364,000 unique taxon-country-year records of catch destined for DHC and FMFO/other uses, respectively, from the SAU reconstructed reported marine fisheries catch database (Pauly & Zeller, 2015). Approximately 31,000 catch records were matched directly to the raw reported ex-vessel price data, and ~100,000 records were estimated at the species level (Appendix Table A.3). Fisheries catch records span the years

1950-2010, include 197 countries and entities, and over 1,700 different taxon groups.

Ex-vessel prices have generally increased over time. Since the 1950s, prices for DHC have increased by ~54%, while prices for FMFO and other products have increased by 60% and 10%, respectively (Figure 2.1). Total average ex-vessel price decreased in the 1960s but was much greater than the decrease observed for DHC prices (Figure 2.1), indicating that seafood was not 29

necessarily getting cheaper. Instead, catches of low-value small pelagic fish, often destined for reduction purposes, increased substantially in the 1960s (Pauly & Zeller, 2016). Low average ex- vessel prices of fish destined for FMFO and “other uses” are reflected in the total landed values for each destination type.

Global landed values in 2010 were estimated to be almost $150 billion, greater than a five-fold increase since 1950 largely due to the increase in catch over this time (Figure 2.2a). The 95% confidence limits for price estimates put the range for 2010 global landed values between $80 and 245 billion. Global landed values increased consistently with landings until peak catch rates in the mid-1990s, where landed values continued to increase while landings have decreased.

Global landed values have continued to climb, currently (2009-2010) at an all-time high. From

1950-2010, global landed values for FMFO have shown an increase from $640 million to over

$6.3 billion, and have averaged ~5% in its proportion of global landed values over that period

(Figure 2.2b). In 2010, fish for FMFO production was nearly 18% of global catch (Cashion et al., 2017) but only 5% of global landed value.

Figure 2.2. Global marine fisheries (a) landed values ± confidence limits (solid line) and landings (dashed line) from 1950-2010, and (b) the proportion of landed values derived from fisheries catch destined for direct human consumption (DHC), fishmeal and fish oil production (FMFO), and other uses.

Historically from 1950-2010, forage fishes—which include herring, sardines, and anchovies— accounted for over 70% of the tonnage for FMFO production (Cashion, 2016) and over 62% of the landed value of fish for FMFO (Table 2.1). In 2010 alone, forage fishes accounted for $11.8 billion and 61% of landed value of catch destined for FMFO production. However, the

proportion of total forage fish landed value (including DHC, FMFO, and other uses) destined for

FMFO production averaged 8%. Forage fishes are thus important to both the DHC and FMFO sector. The top 12 species used for FMFO production have historically accounted for over 80% of reduction fisheries landed value. Two species, anchoveta and the Pacific sardine, have each accounted for 20% of reduction fisheries landed values from the period 1950-2010. However, the percentage of landed values in 2010 for anchoveta was larger than its historical average at 25%, while they were much smaller for Pacific sardine.

Table 2.1. Contribution of forage fishes and the top 12 taxa for FMFO landed value production. % of total taxon landed % of global FMFO value destined for 2010 global production landed value FMFO production landed value

Taxon name 1950-2010 2010 1950-2010 2010 ($ billion) Forage fishes (e.g. herrings, 62.3 61.2 8.0 7.9 11.8 sardines, anchovies)

Anchoveta (Engraulis ringens) 19.6 25.4 98.1 91.6 2.40 Pacific sardine (Sardinops 19.5 3.5 73.0 44.0 0.69 sagax) Chilean jack mackerel 10.0 7.6 75.3 79.6 0.83 (Trachurus murphyi) Gulf menhaden (Brevoortia 7.1 4.6 95.7 99.9 0.40 patronus) Capelin (Mallotus villosus) 6.2 1.3 84.4 41.9 0.27 Atlantic herring (Clupea 5.4 3.6 32.6 23.3 1.35 harengus) Sand eels (Ammodytes spp.) 4.4 4.3 98.0 99.8 0.37 Blue whiting (Micromesistius 3.3 2.4 70.1 55.0 0.38 poutassou) Araucanian herring (Clupea 2.3 9.8 84.2 100.0 0.85 bentincki) Norway pout (Trisopterus 1.8 1.3 99.8 100.0 0.11 esmarkii) Chub mackerel (Scomber 1.8 2.5 23.1 32.6 0.66 japonicus) Japanese anchovy (Engraulis 1.5 3.7 25.3 38.1 0.83 japonicus)

Total (top 12 species) 82.9 69.9 9.1

The top five countries for FMFO landed value production in 2010 were Chile, Peru, China,

Norway, and Denmark, totalling over 5.9 billion real 2010 USD and 68% of global landed value of fish destined for FMFO production (Table 2.2). Proportion of each country’s total fisheries revenues from catch destined for reduction fisheries were highest in Peru, Denmark, Chile, and other countries including Sweden, Georgia, and Finland, accounting for 29-51% of their total

fisheries landed value in 2010. For China and Norway, the proportion of FMFO landed value represents less than 11% of their total fisheries landed values indicating their fisheries economy is not as dependent on FMFO fisheries.

Table 2.2. Top 5 countries for FMFO landed value production in 2010. % of global FMFO % of country landed FMFO landed value Country landed value value to FMFO ($ billion) Chile 31.0 35.7 2.7 Peru 16.2 51.0 1.4 China 10.2 4.3 0.9 Norway 5.8 11.0 0.5 Denmark 5.1 40.4 0.4 Total 68.3 5.9

2.3.2 Comparing prices and landed values

When comparing this third version of the Price DB with the second version (Swartz et al., 2013),

I found that trends of estimated ex-vessel prices over time were better explained when DHC and

FMFO prices were separated (Appendix Table A.4; Appendix Figure A.2). Model outputs suggest that global weighted average prices were significantly lower using my methods and that prices were overestimated by an average of $93.6 US tonne-1 in the previous version of the price database (Appendix Figure A.2) (K = 2; DFResidual = 111; AICweight = 0.55). Accounting for differences in slope trends of prices over time between the two estimation methods in the model suggested that the appreciation of price trends were higher by $2.12 US tonne-1 year-1 using previous version of the price database (Appendix Figure A.2) (K = 3; DFResidual = 111; ∆AIC =

0.44; AICweight = 0.55). Separating prices by the end-product usage is important as it increases

specificity of prices and significantly affects trends of prices and landed values over time at the global and country level (Appendix Figure A.2).

The differences in average ex-vessel prices between the methods were most pronounced in forage fishes (Figure 2.3; for other taxonomic groups see Appendix Figure A.1). Average prices for forage fish were much lower when using my methods to estimate separate prices for non-

DHC purposes, and were much more consistent with FAO prices. When using only one price for all end-products, average prices for forage fishes increased to over $1000 tonne-1 (Figure 2.3).

Figure 2.3. The effects of estimating prices separately for different product types versus one overall price, showing average ex-vessel prices for forage fishes.

Countries with a historically larger proportion of landed values destined for non-DHC purposes were likely to have prices and thus landed values over estimated when prices were not estimated separately for non-DHC purposes. In Peru for example, 2010 landed values were estimated at

$3.8 billion when using separate prices for different products versus $5.8 billion using previous methods (Appendix Figure A.2). Trends in price and landed values also differed for some

countries when comparing the two methods (Table 2.3; Appendix Figure A.2). Globally, price trends were overvalued, and landed value trends were over-appreciated when using the previous methods.

Table 2.3. Linear regression analysis showing the countries, and globally, where price and landed value trends over time significantly differed in either their slope† or intercept between the two estimation methods (Appendix Figure A.2). Consequences of failing to use separate prices for low- value fisheries Prices Landed values Global Chile El Salvador El Salvador Georgia Georgia Over-appreciation (slope)† Peru Pakistan Thailand Panama Turkey Peru Thailand Turkey

Global Chile China Denmark Overvalued (intercept) Denmark Pakistan South Africa Panama South Africa

†Countries with significantly different slopes between the two methods also had significantly different intercepts.

2.3.3 Model validation

A cross validation showed that my model did well in predicting reported prices using a subset of reported price data. My model was able to estimate prices for DHC across country and taxa, with

R2 values greater than 0.43 (Figure 2.4). Price estimates for FMFO/Other uses across countries had a low R2 value of 0.13. This highlights the importance of having adequate coverage of reported prices across countries (Supplementary material). Additionally, the random models 36

performed better than country and taxa models suggesting that the accuracy of price estimates increases with increased diversity of price data across countries and taxa. However, the slopes between estimated and reported price are less than 1, suggesting that price estimates are underestimated for higher prices and overestimated for lower prices.

Figure 2.4. Cross validation where half of the reported prices were removed to estimate the remaining half of the reported prices for DHC (direct human consumption) and FMFO/Other (fishmeal, fish oil and other uses) based on country, taxa, or at random.

2.4 DISCUSSION

Average ex-vessel prices for FMFO decreased in the 1960s (Figure 2.1), likely due to the growing production of FMFO and the increased proportion of low-value species such as the

Peruvian anchoveta which rapidly expanded during this period (Bell et al., 1970; Naylor et al.,

2000). Following, average ex-vessel prices for FMFO (and across product types) rapidly increased in the 1970s and 1980s, likely due to a combination of multiple global events (Figure

2.1). First, the cost of fishing likely increased due to the 1973 and 1979 oil crises which increased fuel costs (Barsky & Kilian, 2002), although its effect on fish prices are also dependent

on other factors such as management decisions and inputs of subsidies. Swartz et al. (2013) also attributes price increases during this time to the establishment of the Third United Nations

Convention on the Law Of the Sea, which extended maritime jurisdictions to a 200 nautical mile exclusive economic zone and thus increased the distance travelled and costs. Peruvian anchoveta collapsed in 1972 due to El Niño and overfishing (Lluch-Belda et al., 1989; Pauly et al., 2002), which likely contributed to the rise in prices due to a combination of decreased supply of catch for FMFO production and the growing demand of FMFO for aquaculture production. Production of Peruvian anchoveta remained low until the mid-1990s (Alheit & Niquen, 2004), yet ex-vessel prices for reduction fisheries decreased in the 1980s (Figure 2.1). This may be attributed to the persistent growth of FMFO production and the growing diversity in species used for FMFO production throughout this period (Cashion et al., 2017), despite low anchoveta numbers.

Another possible explanation is the 1980s “oil glut” (Ramcharran, 2002), where a surplus of oil and a drop in fuel costs likely reduced the cost of fishing and thus ex-vessel prices. Reduction fisheries prices steadily decreased into the 1990s (Figure 2.1), as Peruvian anchoveta has regained its substantial contribution to FMFO production (Alheit & Niquen, 2004). Since decreasing in the 1980s and 90s, prices for fish for FMFO have steadily increased through the

2000s, likely attributed to the combination of concurrent events, notably a decrease in global marine fisheries catch and supply (Pauly and Zeller, 2016), increased demand from aquaculture expansion (Shepherd & Jackson, 2013), and increasing fuel costs.

Aquaculture has expanded rapidly in the last few decades while forage fisheries production has recently decreased (Shepherd & Jackson, 2013). Stricter fisheries management controls (e.g.,

Aranda, 2009), increased DHC processing of fish species formerly used as FMFO (Shepherd &

Jackson, 2013; Cashion et al., 2017), and alternative sources (e.g. terrestrial) of feed (Naylor et al., 2009) are some of the reasons why reduction fisheries production has not kept up with aquaculture production. While we may expect changes in demand or supply of FMFO to have an effect on ex-vessel prices, there may be other market factors restricting any drastic changes in price. Similar to other land-based farmers, feed will be substituted for other feeds if prices increase and quality is not compromised (Asche & Bjorndal, 1999; Asche & Tveterås, 2004).

Soymeal, a close substitute for fishmeal in terms of crude protein, formerly prevented large changes in price for fishmeal despite the increase in demand (Asche & Tveterås, 2004; Asche et al., 2013). However, this price dynamic between fishmeal and soymeal has broken down in recent years (Asche et al., 2013), as fishmeal is substitutable only to a point in aquaculture diets.

This recent development (post-2004) can be seen in the uptick in ex-vessel prices of fish destined for FMFO production.

The growing demands for FMFO for aquaculture feed is concerning due to its ecological and socioeconomic impacts. FMFO products are mostly derived from forage fishes (Table 2.1;

Cashion et al., 2017), and many of these stocks are overexploited (Srinivasan et al., 2010; Cao et al., 2015). While many of these fish have favourable traits for exploitation (e.g. high growth rates, fecundity, turnover), their populations are vulnerable to collapse. Forage fishes serve an important ecological role as a primary source of prey, and are more valuable to the greater economy in the water than as FMFO (Pikitch et al., 2014). Further, about 90% of non-DHC fisheries catch (~18 million tonnes annually) are food-grade quality (Cashion et al., 2017), and would alleviate global food insecurities.

Global demands for FMFO are mainly driven by aquaculture in China, and FMFO use in aquaculture feed has increased overall despite the decrease in the inclusion of FMFO in aquaculture feed (Naylor et al., 2009). China’s aquaculture industry has rapidly expanded and is currently the largest importer of FMFO products (Cao et al., 2015). There is an attempt to shift to more sustainable plant-based feeds, but many companies have already secured future rights to high quality FMFO products (Cao et al., 2015), limiting the capacity for change. However, the change in aquaculture formulations is driven by price (Hardy, 2010), and this will continue to incentivize the use of alternative sources of feed to substitute FMFO products. It is possible that any depleted reduction fisheries stocks and the higher prices associated with them will drive a decrease in demand for FMFO products as alternative sources of feed are sought. The result of this may be more forage fish for DHC.

There is currently a growing proportion of fish by-catch, often mislabelled as ‘trash fish’, used for fishmeal production (Cao et al., 2015). The FMFO products from this mixed fish catch are of a lower quality and sell for lower prices to be used for low-value aquaculture species (Chiu et al.,

2013). While at first the use of by-catch as aquaculture feed may seem resourceful, it may put further strain on wild fish stocks and ecosystems (Cao et al., 2015). Current production remains relatively low but a rise in price due to a growing demand may increase fishing for these non- targeted species (Cao et al., 2015). What is a concern is that the group of species categorized as

‘trash fish’ are variable and the growth of non-targeted fisheries will be inherently difficult to manage.

This chapter provides valuable information and tools for researchers to value fisheries catches destined for DHC, FMFO, and other uses. Distinguishing prices between product types showed

to have effects on highly aggregated price and landed value trends over time (Table 2.3;

Appendix A). This Price DB is comparable to other efforts in constructing an ex-vessel marine fish price database. Specifically, Melnychuk et al. (2016) used export prices to estimate ex- vessel prices using publicly available FAO data (FAO, 2014b). They were also able to estimate prices for products destined for FMFO. The authors have strong arguments for their approach, such as the consistency of FAO data and its public availability. However, recent studies have shown that global catch may be much higher than reported by FAO (Pauly & Zeller, 2016). My

Price DB has greater temporal coverage (1950-2010) and provides price estimates for the disaggregated catch records from SAU. I also provided price estimates for the various species used for FMFO production and other uses, whereas Melnychuk et al. (2016) only provided an aggregate estimate of price for all FMFO catches.

2.4.1 Limitations of the global price database

Application of this database is best directed towards large-scale regional and global analyses. I caution the use of the database for economic analyses on local and individual fisheries, and instead I suggest that raw data be collected from the fisheries themselves for such applications.

Many of the ex-vessel prices were estimated from related taxa or prices from other countries, and thus may not reflect actual economic trends of the specific fisheries stocks of interest.

Price estimates are assigned to each unique taxon-country-year catch, regardless of fishing sector. The SAU catch database disaggregates catch by sector (industrial, artisanal, subsistence) and consequentially, prices and landed values may be underestimated for artisanal fisheries.

Artisanal fisheries often have a shortened supply chain, where catches can be transferred directly from fisher to consumer, which can affect ex-vessel prices. Therefore, these estimates may better 41

reflect industrial prices depending on where the data were sourced. The next steps for an update of the Price DB would be to break down prices and values by fishing sector.

2.5 CONCLUSION

Future research derived from this database will improve the understanding of the economic contributions and future potential of the different products derived from marine fisheries resources. As global aquaculture has rapidly expanded since the 1990s (Naylor et al., 2000), there are increasing demands for aquaculture feed inputs such as FMFO. However, supply fish destined for FMFO production remained relatively constant (Alder & Pauly, 2006) and has more recently declined to a lower proportion of global landings (Cashion et al., 2017). Prices for

FMFO products have generally remained relatively elastic to growing demand due to the availability of substitutes such as soybean meal (Asche & Tveterås, 2004; Asche et al., 2013).

However, prices have increased in the past few decades (Asche et al., 2013), alluding to changes in feed composition, demand and supply, and the structure of product ownership and future rights. Understanding price trends in response to market factors will facilitate in managing global fish stocks and the demand for FMFO. Further, this price database contributes to developing a better understanding of the price fluctuations of ex-vessel prices for estimating value chains for reduction fisheries (e.g., Christensen et al., 2014). Additionally, this Price DB will contribute to the growing literature on socioeconomic scenario development and analyses in fisheries science

(e.g. Cheung et al., 2016; Lam et al., 2016). For my thesis, this Price DB is a valuable building block for subsequent chapters. It is used in analyses of various climate change scenarios and its effects on fisheries catch and landed value.

An update of the Price DB was required as the global SAU catch database has undergone major reconstruction and is now more detailed and refined. However, the production of such large- scale databases does not come without its caveats. Therefore, application of this price database to finer scales of data poor fisheries, such as species- or community-specific analyses, must be exercised with caution. Nonetheless, the Price DB is a valuable resource for assessing large-scale trends. Landed values calculated from these prices can be found on the SAU website

(www.seaaroundus.org).

Chapter 3: Comparing model parameterizations of the biophysical impacts of ocean acidification to identify limitations and uncertainties

3.1 INTRODUCTION

Carbon dioxide (CO2) emissions from human activities such as the burning of fossil fuels largely contributes to the rapid rate of ocean acidification (OA) since the industrial revolution (IPCC,

2013). OA is the chemical process driven by elevated atmospheric CO2 that results in reduced pH and increased acidity. Global sea surface pH has already decreased by 0.1 units since the pre- industrial average of 8.17, a 26% increase in acidity (Caldeira & Wickett, 2003; Feely et al.,

2009; Pörtner et al., 2014). Under our current emissions trajectory, sea surface pH is projected to decrease by an additional 0.3 units by the end of the 21st century (Ciais et al., 2013; IPCC, 2013).

Some areas are experiencing much larger changes in pH. For example, the Northeast Pacific

Ocean has naturally fluctuating pH levels due to upwelling, and the uptake of anthropogenic CO2 is elevating acidification across these areas (Feely et al., 2014; Haigh et al., 2015).

Ocean acidification is expected to impact marine organisms, communities and ecosystems

(Guinotte & Fabry, 2008; Cooley et al., 2009; Doney et al., 2012; Le Quesne & Pinnegar, 2012;

Branch et al., 2013; Haigh et al., 2015; Mathis et al., 2015), with variations in sensitivity across populations, taxonomic groups and ecosystem types (Kroeker et al., 2013; Heuer & Grosell,

2014; Nagelkerken & Connell, 2015). Most notably, OA compromises the ability of organisms to efficiently build and retain calcium carbonate structures (e.g. coral reefs, oyster and mussel shells, coccolithophore exoskeletons) due to the under-saturation of calcium carbonate (Feely et al., 2004; Kleypas et al., 2006; Fabry et al., 2008; Ries et al., 2009; Nienhuis et al., 2010).

Beyond calcification, OA affects a wide range of physiological processes such as acid-base balance, basal metabolic rates, aerobic scope, oxygen consumption, thermal tolerance, fertilization rates, and development, among others (detailed in Le Quesne and Pinnegar, 2012).

Direct impacts of OA on changes in species abundance will result in important changes to competitive, facilitative, and trophic relationships (Trenkel et al., 2005; Dutkiewicz et al., 2015;

Queirós et al., 2015; Sunday et al., 2017). Overall, changes in physiology and behaviour lead to changes in growth and abundance, and when considered across multiple interacting species, results in important changes in community structure and ecosystem function (Kroeker et al.,

2013; Nagelkerken & Connell, 2015).

OA coincides with other anthropogenic CO2 driven stressors including ocean warming and decreases in dissolved oxygen concentration. Increases in temperature affect physiological processes such as metabolism, increasing the demand for oxygen and reducing aerobic scope

(Pörtner & Lannig, 2009). Decreases in oxygen content further exacerbates this effect, and are projected to lower the metabolic capacity of marine habitats potentially leading to decreased body size (Cheung et al., 2013b; Deutsch et al., 2015; Pauly & Cheung, 2017). Biogeographic responses to ocean warming and decreased oxygen content have been observed as shifts in distributions to deeper and higher latitudinal waters (Perry et al., 2005; Dulvy et al., 2008;

Cheung et al., 2013a). While isolating the effects of OA is important for understanding the mechanisms in which OA effects operate, integrating OA with other stressors provides a more real-world application of the effects of anthropogenic-influenced global changes on species distribution and abundance.

The inclusion of OA in assessing impacts of anthropogenic CO2 emissions is important in developing scenarios of future global change for marine systems. Published syntheses and meta- analyses are extremely useful for providing baseline parameters for modelling and assessing the biological impacts of OA (Kroeker et al., 2013; Nagelkerken & Connell, 2015; Sunday et al.,

2017). They also provide a basis for linking complex physiological responses to life history traits that have direct implications on population dynamics. For example, Cheung et al. (2011) incorporated physiological models into a dynamic bioclimatic envelope model (Cheung et al.,

2008a) to assess climate change effects on species distribution and abundance. This model was then applied to a socioeconomic analysis of climate change and OA impacts in the Arctic Ocean

(Lam et al., 2014). Empirical models of OA effects have also been used to estimate changes in the growth rates of mollusc species and thereby impacts on US mollusc fisheries (Ries et al.,

2009; Moore, 2015). Another promising approach incorporates information about marine food webs by using an ecosystem model to model the impacts of OA on functional groups that include harvested taxa (Ainsworth et al., 2011).

Projection models provide valuable insight to potential future scenarios but are subject to various sources of uncertainty. Uncertainty when modelling ocean acidification arises from the choice of model parameterizations, which produce a range of possible impact pathways. I defined three sources of uncertainty when modelling OA impacts: 1) structural, 2) parameter, and 3) scenario uncertainty (Hawkins & Sutton, 2009). Structural uncertainty refers to the underlying construction of the model, such as the mathematical formulation of a model to represent ecological relationships, or the processes modelled using correlative versus mechanistic approaches (e.g. Pauly et al., 2000). Parameter uncertainty stems from the inherent variability

and our limited ability to accurately and precisely measure biological processes and relationships

(e.g. Kremer, 1983). Scenario uncertainty results from the different possible future pathways due to many socioeconomic factors (e.g. governmental policies, technological development) that affect biophysical drivers. This includes the various greenhouse gas concentration trajectories used to drive climate and biophysical models (IPCC, 2013). The combined uncertainties produce the full range of future trajectories, providing valuable insight to the sensitivities of modelling

OA impacts.

In this study, I explored various ways to incorporate OA impacts into a multi-stressor dynamic bioclimatic envelope model to project changes in the biogeography of ten commercially exploited invertebrate species. I examined the structural, parameterization and scenario uncertainties in modelling OA effects. To eventually improve our confidence in forecasting future scenarios, I explored the variability of model outputs and discussed the utilities and limitations of different ways to incorporate OA impacts in spatial biogeographic models.

3.2 METHODS

I incorporated the impacts of OA into a previously developed dynamic bioclimatic envelope model (DBEM) (Cheung et al., 2008a, 2011, 2016b) to estimate changes in species distribution and abundance. The DBEM uses earth system models (ESM) as inputs (e.g. Dunne et al., 2013) and links species distribution models (Jones et al., 2012), advection-diffusion movement models

(Sibert et al., 1999), growth models (Pauly, 1980a), physiological models (Pauly, 1981), and population dynamics models (Pauly, 1980a; Hilborn & Walters, 1992; O’Connor et al., 2007) to predict how species will move geographically across time (annual time step) and space (on a 0.5˚ latitude x 0.5˚ longitude grid) with climate change. I outline the specifics of modelling the effects 47

of OA and how it interacts with effects from other stressors (i.e. temperature and oxygen) below, while other details on the DBEM can be found in the Supplementary Material (Appendix Figure

B.1).

3.2.1 Modelling the effects of global change

Global change effects on organisms and populations include changes in temperature, oxygen and pH. I integrated the biological impacts of OA on exploited populations through the effects on somatic growth and mortality rates. I define the effects on somatic growth as a mechanistic process, and the effects on survival as a correlative process. First, the model uses the von

Bertalanffy growth function (von Bertalanffy, 1951) to simulate changes in growth in response to ocean warming, decreases in dissolved oxygen concentration, and ocean acidification (Cheung et al., 2011). Growth rate (change in biomass, B, as a function of time, t) is determined with the derived equation from a growth function:

!" = ��! − ��! (3.1) !" where H and k represent the coefficients for anabolism and catabolism, respectively. Anabolism scales with body weight (W) to the exponent d < 1, catabolism scales linearly with (W), i.e. b = 1, and their difference determines the growth rate of species biomass (B). Solving for dB/dt = 0

(!!!) when asymptotic weight (W∞) is reached, I obtained � = ��! . Thus, growth rate is dependent on the available oxygen (anabolism) and oxygen demand for maintenance metabolism

(catabolism).

Integrating equation (3.1) into a generalized von Bertalanffy growth function:

!! !!!! !/(!!!) �! = �![1 − � ] (3.2) where K is the von Bertalanffy growth parameter where � = �(1 − �). The von Bertalanffy growth parameter K represents the rate at which maximum body size is reached. I assume d =

0.7, although values typically range from 0.5 and 0.95 across invertebrate species (Hughes, 1983;

Johnson & Rees, 1988; Jones et al., 1992). Sensitivity of maximum body size to changes in temperature and acidity show that low values of d (< 0.7) results in slight decreases in sensitivity, while larger values of d (>0.7) results in major increases in sensitivity (Appendix Table B.1)

(Pauly & Cheung, 2017). Effects of multiple stressors show an antagonistic interaction for the effects on body size. Therefore, the use of 0.7 for all species considered here are conservative as smaller values of d do not considerably change temperature and acidity effects on maximum body size, while larger values of d only increase sensitivity.

The effects of temperature were modelled to affect metabolism—described in equations (3.3) and (3.4)—through the H and k coefficients following the Arrhenius equation, �!!/!, where

� = �!/�, with �! and � equal to the activation energy and Boltzmann constant, respectively.

Furthermore, oxygen availability affects aerobic scope (i.e. oxygen supply) while acidification affects maintenance metabolism (i.e. oxygen demand). I modelled the impacts of decreases in oxygen and ocean acidification following the current working oxygen- and capacity- limited thermal tolerance (OCLTT) hypothesis, which assumes that deviations from optimal environmental conditions (e.g. temperature, acidity) decreases aerobic scope and thus the energy available for growth (Pörtner, 2008; Pörtner & Farrell, 2008). These effects can be modelled in the form:

!!!/! � = �[O!]� (3.3) and

� = ℎ[H!]�!!!/! (3.4)

The constants �! and �! are equal to Ea/R where Ea (for anabolism and catabolism, respectively) and R are the activation energy and Boltzmann constant, respectively, while T is the absolute temperature (in Kelvin) (Cheung et al., 2011). Changes in the concentration of oxygen [O2] and hydrogen ions [H+] relative to initial conditions thus change H and k, respectively. The coefficients g and h from equations (3.3) and (3.4), respectively, were derived for each species from the average W∞, K, and environmental temperature �! reported in the literature (Cheung et al., 2011):

(!!!) !! ! � = !! /! (3.5) [!!]! ! and

ℎ = !/(!!!) (3.6) [!!]!!!!/!

The model predicts changes in life history parameters due to changes in temperature, oxygen availability, and pH for asymptotic weight (W∞) and von Bertalanffy growth parameter K:

!/(!!!) � = ! (3.7) ! ! and

� = �(1 − �) (3.8)

Other parameters that scale with weight can also be predicted, including asymptotic length and the length at maturity (Beverton & Holt, 1959).

Natural population mortality rates (M) were estimated from the empirical equation (Pauly,

1980a):

� = −0.4851 − 0.0824 log �! + 0.6757log (�) + 0.4687 log (�′) (3.9) where �′ is the average water temperature of a species range in degrees Celsius—other parameters are defined above. Pauly's (1980) model was used here because of data availability and its widespread use for fish stocks and in fisheries assessments; I address the scope of using alternative empirical equations in the discussion section. Thus, my model incorporates trade-offs between basal metabolism and aerobic scope, ultimately affecting life history traits of growth, maximum body size, and mortality rate. Additionally, at the population level, changes in growth parameters affect mean body size and fecundity.

The construction of my model uses a mixed approach and incorporates both mechanistic and correlative models. OA effects on growth (mechanistic model) operate at multiple levels throughout the DBEM, and have downstream effects on maximum weight (equation 3.7), growth rate (equation 3.8), and mortality rate (equation 3.9). OA effects on survival (correlative model) were modelled to directly affect population mortality for both larval and adult stages, such that changes in [H+] results in changes in the population mortality rate (M) (equation 3.9).

Growth and survival rates change accordingly with changes in pH, which I measure here as hydrogen ion concentration, [H+]. I used parameters from a meta-analysis for the effects of OA on life history traits (Kroeker et al., 2013). I modelled the effects of OA under the assumption that the parameter values represent a percent change in growth or survivorship with a doubling of

[H+].

3.2.2 Characterizing uncertainties

I explored three possible sources of uncertainty in modelling the effects of OA: structural, parameter, and scenario uncertainty. First, I focus on structural uncertainty as the inaccuracy in the modelled relationship between [H+] and life history rates (i.e. somatic growth rate and survival rate). To establish relationships between [H+] and life history traits, I used parameter values from Kroeker et al. (2013), which are point estimates for a given minimum change in pH.

Unfortunately, no information on the relationship between these point estimates was available and previous studies have shown variation across species (Ries et al., 2009). I considered three possible ways to represent the structural relationship between [H+] and life history rates and tested the sensitivity of the model projections to each underlying relationship. Figure 3.1 illustrates the various options for which I make assumptions about the relationship. For option A

I scaled changes in [H+] linearly with life history rates relative to initial conditions, reflecting a scenario in which a species’ sensitivity to OA is similar at low and high levels of acidity. This represents the baseline model, i.e.

! ! ! ! ��! = ��!"!# ∗ 1 + �� ∗ ! − 1 (3.10) ! !"!#

Surv is the survival rate per year and used here as an example but can be applied to other biological characteristics affected by OA (e.g. growth, reproduction, calcification). Survival rate in year t is derived from the initial (init) survival rate and the relative change in [H+] between year t and initial [H+] conditions. Per represents the value used from Kroeker et al. (2013) for the percent change in survival rate with a doubling of [H+] (Table 3.1). The exponent w is equal to 1 in the linear model. For option B I used an exponential model, which reflects a scenario in which a species’ sensitivity to OA increases with increasing [H+]. Using equation (3.10), the exponent w is set to 2. Changes in [H+] in these two scenarios are expressed relative to initial conditions, defined as the average [H+] between years 1971 and 2000. For option C, I explored an

‘adaptation’ scenario, where species are assumed to acclimatize or adapt to recent changes in pH.

This scenario is represented by expressing changes in pH relative to a moving window of the

+ + previous year, i.e. where in equation (3.10) [H ]init is substituted with [H ]t – 1 and survival rate in year t is calculated from the relative change in [H+] between year t and the previous year t – 1.

The adaptation scenario assumes a linear relationship between [H+] and the biological traits affected by OA, and further assumes that acclimatization and/or adaptation has a fixed time lag but that the capacity for adaptation is limitless.

a b ] + [H No OA

OA (OA - No OA)

Time ∆abundance(%) Time Time

A. Linear B. Exponential C. 'Adaptation'

Baseline conditions Survival rate

Time

Figure 3.1. a) Conceptual diagram illustrating the scenarios of various relationships between [H+] and survival rate explored in this study, under a hypothetical linear change in acidity over time. Survival rate is used here as an example that can be applied to other biological parameters impacted by ocean acidification (OA). b) Diagram of how I presented the results by isolating the net impacts of OA on abundance by taking the differences between model outputs with OA and no OA effects.

I compare differences between the results from mechanistic and correlative models for OA effects (described above) as an additional source of structural uncertainty. Furthermore, I included simulations using different ESMs, which have different structural characteristics such as grid resolution and marine biogeochemical components (Bopp et al., 2013). I used three

ESMs: NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL-ESM); Max Planck Institute for Meteorology (MPI-ESM); and Institute Pierre Simon Laplace Climate Modelling Centre

(IPSL-ESM). These models were chosen as they provide high-resolution (1˚ longitude x 1˚ latitude) sea surface and bottom data, as well as all the environmental data required by the

DBEM (Cheung et al., 2016a).

Table 3.1. Effect sizes (with 95% confidence limits in parentheses) of OA impacts on life history traits (modified from Kroeker et al., 2013).

Growth† Survival† Mean Mean (Lower, Upper) (Lower, Upper) Crustaceans -16% -13% (-33%, 6%) (-25%, 0%)

Molluscs -17% -35% (-26%, -10%) (-56%, -10%)

†Effect sizes represent the percent change with a doubling of hydrogen ion concentration.

To characterize parameter uncertainty, I generalized OA effects based on taxonomic group and simulated the impacts of OA with the high and low 95% confidence limit values provided in

Kroeker et al. (2013) (Table 3.1). While species-specific parameters do exist (Ries et al., 2009), I chose to use generalized values to make comparisons of the sensitivity of my results to this source of uncertainty.

Lastly, I characterized scenario uncertainty by using the different representative concentration pathways (RCP), which uses atmospheric greenhouse gas concentration scenarios to drive other environmental variables. There are four RCP scenarios commonly used: RCP 2.6, RCP 4.5, RCP

6.0, and RCP 8.5. The numbers represent radiative forcing values in the year 2100 under specific

’ ‘ ’ emissions scenarios. I labelled the pathways as ‘Low CO2 and High CO2 scenarios to predict

OA impacts in the future representing RCP 2.6 and RCP 8.5, respectively. RCP 2.6 characterizes the optimistic scenario in which immediate action is taken and annual global GHG emissions peaks within a decade (year ~2025) but is then drastically reduced. While this pathway of strong mitigation of GHG emissions is unlikely given the current state of progress, it is the closest amongst the RCP scenarios to simulate conditions if we were to achieve targets set in the recent 55

‘Paris Agreement’ at the 2015 United Nations Framework Convention on Climate Change

Conference of the Parties (COP 21). RCP 8.5 characterizes the current emissions trajectory where nations continue to develop their economy and the industrial sector using fossil fuels as the primary source of energy. This pathway is often referred to as the “business-as-usual” scenario, where no shift is made to reduce carbon emissions and invest in sustainable and renewable energy sources. Since this study focuses on exploring the sensitivity of multi-climatic stressors impacts, fishing is assumed to be at level required to achieve maximum sustainable yield for each species in all the scenarios.

I acknowledge that OA can have effects on other biochemical (e.g. ATP production) and physiological processes (e.g. acid-base regulation) which can have downstream effects not directly linked to life history traits (e.g. bioenergetics) (Le Quesne & Pinnegar, 2012;

Waldbusser et al., 2016). However, I constructed the model to link OA effects to physiological processes that can then be directly tied to life history traits that affect population dynamics.

While OA effects on organism biology are generally complex, my model uses more generalized models in order to make comparisons about uncertainties.

3.2.3 Modelled species

I chose 10 commercially exploited invertebrate species distributed in either the Northwest

Atlantic or the Northeast Pacific (Table 3.2) (initial species distributions can be found in

Appendix Figure B.2). Some species were distributed across both regions. This allowed me to compare results across and within ocean basins. Invertebrate species were chosen as current knowledge suggests they are more sensitive to changes in pH than finfish (Kroeker et al., 2013).

I chose 5 mollusc species and 5 crustacean species, spanning various taxonomic groups 56

including: crabs, lobsters, prawns, oysters, mussels, clams, and cephalopods. Bottom environmental data was used for all demersal species, while surface environmental data was used for the only pelagic species included in my analysis—D. pealeii.

Table 3.2. Species analyzed for the impacts of OA. Ocean basin Common name Species name distribution Taxon group American cupped oyster Crassostrea virginica Atlantic Molluscs American lobster Homarus americanus Atlantic Crustaceans American sea scallop Placopecten magellanicus Atlantic Molluscs Blue mussel Mytilus edulis Atlantic Molluscs Dungeness crab Metacarcinus magister Pacific Crustaceans Longfin inshore squid Doryteuthis pealeii Atlantic Molluscs Northern prawn Pandalus borealis Arctic, Atlantic, Pacific Crustaceans Pacific geoduck Panopea generosa Pacific Molluscs Snow crab Chionectes opilio Atlantic, Pacific Crustaceans Coonstripe shrimp Pandalopsis dispar Pacific Crustaceans

First, I present my results for the impacts of OA on species abundance from a multi-stressor model by including results from models run with and without OA impacts. In other words, I isolate the impacts of OA from other climate change stressors (i.e. temperature and oxygen) by taking the difference in abundance between model results with OA and without OA (hereon labelled OA and No OA scenarios, respectively) impacts for each year (Figure 3.1b). I then present full model results to put those OA effects into context of the overall effects of expected global environmental change.

3.3 RESULTS

3.3.1 Projected changes in climate stressors

The ESMs ensemble project changes to both sea surface and bottom environmental conditions and at a greater rate under high CO2 emissions. Globally, average sea surface pH is projected to decrease by as much as 0.4 units, an increase in acidity of 140% by 2100 relative to 1950 (Figure

3.2a). Average change in sea bottom pH is projected to change by only as much as 0.06 units, yet still presents an increase of 14% in acidity (Figure 3.2e). The magnitude of acidification will vary across ocean basins. The three other main drivers (temperature, oxygen, and primary production) of species biomass production in the model are projected to drastically change in high CO2 scenarios (Figure 3.2). For example, sea surface temperature is projected to increase by up to 4˚C by 2100, while bottom temperature is projected to increase by up to 0.4˚C. Change in net primary production was most variable across models and projections show increases and decreases across different ocean basins (Figure 3.2d and Appendix Figure B.3 and B.4). In low

CO2 scenarios, the rate of change for many of these biophysical drivers is projected to decline initially and level off by 2050, with the exception of the sea bottom that will continue to change as surface waters slowly circulate and mix with deeper layers.

a b c d 0.0 0.00 4 0.4 C) C)

° 3 −0.1 −0.02 ° 0.3 ace f ace pH f 2 0.2 −0.2 −0.04 sea sur sea bottom sea bottom pH sea sur 1 ∆ −0.3 ∆ 0.1 temperature ( ∆ ∆ −0.06 temperature (

−0.4 0 0.0 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100 e f g 0 0.0 0.0 ) ) ) 1 1 1 − − − −2.5 −5 −0.1 ace O2 ear f litre litre y y production ⋅ ⋅ r ⋅ Low CO −5.0 −0.2 2 ima mol mol

−10 r µ µ sea bottom O2 sea sur ( ( (Pg C −0.3 High CO2 ∆ ∆ −7.5 net p

−15 ∆ −0.4 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100 Year

Figure 3.2. Projected changes in ocean variables used as the main biophysical drivers in the model. Thin lines are projections from each of the three earth system models used (GFDL, IPSL, MPI) while thick lines are multi-model means. Projections are smoothed using 10-year running means.

3.3.2 Responses to ocean acidification and global change

I used American lobster (H. americanus) to demonstrate how changes in temperature and ocean acidity affect life history parameters. Increases in either temperature or ocean acidity decreases the maximum body size and increases the von Bertalanffy growth parameter K (Figure 3.3a and

3.3b). Compounded effects of temperature and acidity further reduce maximum body size (e.g.

Sheridan and Bickford, 2011), but show diminishing rates of change as it approaches the asymptotic initial body size at time 0. Temperature and acidity combine to then increase the growth parameter K as von Bertalanffy’s theory predicts that the rate of approaching the maximum body size of smaller individuals (and individuals with a smaller maximum body size) is faster (thus higher K) (von Bertalanffy, 1957). Additionally, temperature and acidity increase the growth rate with the trade-off of a reduction in body size (Atkinson, 1994; Sheridan &

Bickford, 2011). Temperature effects on mass-specific growth rate (dB/dt from equation 3.1) show that smaller individuals initially grow much faster (Figure 3.3c) while larger individuals will grow much slower with increased temperature (Figure 3.3d and 3.3e). The rate of increased growth due to temperature effects on smaller individuals diminishes with greater change in stressors (temperature and acidity), while there is a negative synergistic effect on growth rate from multiple stressors for larger individuals.

a b ● ● 2500 ● Change in acidity (%) ● ● ●

) 0.16 ● ● ● 0 ● ● -1 ● (g) 2000 ● ● ● ● ● 10 ∞ ● ● 0.14 ● ● ●

W ● ● ● 20 ● ● K (year ● 1500 ● ● ● ● ● 50 ● ● 0.12 ● ● ● ● ● ● ● 100 1000 ● ● 0.10 ● 0 1 2 3 4 0 1 2 3 4 c d e ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 85 ● ● ●

) ) ● ) 60.0 ● ● ● ● ● ●

-1 -1 ● ● -1 90 ● ● ● ● ● ● ● ● 80 ● ● 57.5 ● ● ● ● ● year ● year year ● ● ● ● ● ● 60 ● ● ● ● (g• (g• ● 55.0 ● ● ● ● 75 ● ● ● ● ● ● ● 30 ● ● ● 52.5 ● dB/dT (g• ● dB/dT 70 ● dB/dT ● 50.0 ● ● 0 ● 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 ∆ temperature (°C)

Figure 3.3. Effects of changes in temperature and ocean acidity on life history parameters in American lobster (H. americanus): a) maximum body weight, b) von Bertalanffy growth parameter K, and growth rate for c) small (100 g), d) medium (250 g), and e) large individuals (1000 g). Black circles represent initial conditions with no change in temperature or acidity.

Impacts of OA on both growth and survival were projected to reduce species’ abundance under

st the high CO2 scenario by the end of the 21 century (Figure 3.4). Specifically, the combined effects of OA on growth and survival are projected to generally decrease species’ abundance by as much as 10% (American cupped oyster) by year 2091-2100 relative to 1996-2005, while under the low CO2 scenario there are negligible (change of <1%) effects on abundance for most 60

species. Pacific geoduck abundance showed the most sensitivity to OA under the low CO2 scenario, decreasing by ~4% (Figure 3.4h). Interestingly, Dungeness crab abundance initially decreased due to OA but effects diminished by the end of the simulation (Figure 3.4e). This pattern is likely due to a greater net increase of suitable habitat relative to the loss of habitat, leading to a spatial expansion of range size.

Changes in abundance over time when only OA impacts on growth were included closely follows the abundance for the combined effect of OA on growth and survival (Figure 3.4). When impacts to both survival and growth were considered, changes in abundance were amplified. The magnitudes of OA impacts on abundance were variable but most species show a negative response to OA. My results suggest that species abundance is more sensitive to the effects of OA on growth (mechanistic model) than the effects of OA on survival (correlative model).

Trends of changes in abundance with the different structural OA relationship models (Figure

3.1a) were consistent across species, and differences between the relationship models were more pronounced under the high CO2 scenario (Figure 3.5; Appendix Figure B.4). For most species

(with the exception of Pacific geoduck), changes in abundance were small (<1%) under the low

CO2 scenario across all relationship curves (Figure 3.5a; Appendix Figure B.4). In the example using American lobster in the high CO2 scenario, abundance decreased by <1% with the linear relationship, while the exponential relationship showed slow declines in abundance initially, but reached a ‘tipping point’ beyond which abundance declined more rapidly to >2% (Figure 3.5b).

Changes in abundance under the ‘adaptation’ scenario were generally small and did not elicit a significant response.

a American cupped oyster b American lobster

0.0 0.00

−2.5 −0.25

−5.0 −0.50

−7.5 −0.75

2000 2050 2100 2000 2050 2100

c American sea scallop d Blue mussel

0.5 0.05 0.0 0.00 −0.5 −0.05 −1.0 −1.5 −0.10 −2.0 −0.15

2000 2050 2100 2000 2050 2100

e Dungeness crab f Longfn inshore squid eto2000) v Combined 0 0.0

undance between Growth b Survival −1 −0.5 Low CO −1.0 2

A scenarios (relati −2 High CO2 O 2000 2050 2100 2000 2050 2100

g Northernprawn h Pacifc geoduck Aandno erence in % change in a f O

Dif 0.5 0 0.0 −2 −0.5 −4 −1.0 −1.5 −6

2000 2050 2100 2000 2050 2100

i Snow crab j Coonstripe shrimp

0.0 0.0

−2.5 −0.5 −5.0

−1.0 −7.5

−10.0 2000 2050 2100 2000 2050 2100

Year

Overall impacts to abundance due to all climate change stressors included in the model (e.g. OA, temperature, oxygen, primary production) showed high variability across species, with decreases 62

greater than 40% (e.g. longfin inshore squid) and increases reaching 100% (e.g. northern prawn) by 2100 (Figure 3.6). Only four of the ten species showed decreased abundance while the other six species showed increased abundance in the high CO2 scenario. Half of the species showed greater than 5% change in abundance in the low CO2 scenario, suggesting that these species may be sensitive to even small changes in environmental conditions. OA showed consistently negative impacts on abundance across most species with greater differences between OA and no

OA models in the high CO2 scenario. Mollusc species were more sensitive to OA than crustacean species, with the exception of blue mussels and coonstripe shrimp.

0.2

0.1

0.0

−0.1

−0.2

eto2000) Linear v −0.3 Exponential undance between

b Adaptation b

Low CO2

0.0 High CO2 A scenarios (relati O −0.5

−1.0 Aandno erence in % change in a f O Dif −1.5

−2.0

2000 2025 2050 2075 2100 Year

(growth and survival) for two climate change scenarios, low CO2 (a) and high CO2 (b), using GFDL earth system model. Abundances are presented as 10 year running averages.

I used American lobster once again as an example to show the biogeographic changes under the high CO2 scenario. The distribution of American lobster shows negative impacts to its southern range (Figure 3.7a). Areas in the Gulf of St. Lawrence and areas farther offshore show increased abundance. However, the net effect of the changes in the geographic distribution is an overall decrease in abundance (Figure 3.6b). OA effects appear to be greatest in more northern and periphery areas of their geographic range (Figure 3.7b). However, areas with greater OA effects correspond with areas of increased abundance, suggesting that OA may be limiting population growth and range expansion in response to other climate change drivers (i.e. temperature).

a American cupped oyster b American lobster

0 0 −5

−3 −10 −15 −6 −20 −9 2000 2050 2100 2000 2050 2100

c American sea scallop d Blue mussel

0 20

−5 15 10 −10 5 −15 0 −20 2000 2050 2100 2000 2050 2100

e Dungeness crab f Longfn inshore squid

eto2000) 0 v 60 −10 OA No OA 40 −20 20 −30 Low CO2 undance (relati b High CO2 0 −40 2000 2050 2100 2000 2050 2100

g Northernprawn h Pacifc geoduck % change in a 100 80

75 60

50 40

25 20 0 0 2000 2050 2100 2000 2050 2100

i Snow crab j Coonstripe shrimp

0 40

−10 20

−20 0 2000 2050 2100 2000 2050 2100

Year

Figure 3.6. Projected changes in abundance (relative to 2000) for model simulations with OA (solid lines) and without OA (dashed lines) impacts on growth and survival under two climate change scenarios. Changes in abundance are presented as multi-model averages across the earth system model used (GFDL, IPSL, MPI) and smoothed using 10 year running averages.

a b

Difference in % change in abundance between % change in abundance (relative to 2000) OA and No OA scenarios (relative to 2000) <-50 -25 -10 -5 +5 +10 +25 >+50 <-25 -10 -5 -2.5 -1 -0.5 >+0.5

3.3.3 Sensitivity to uncertainty

Results from the DBEM showed varying levels of sensitivity to each source of uncertainty for each species. I quantified the range of projected changes in abundance for each source of uncertainty and found that changes in abundance were most sensitive to the uncertainty in the structural relationship between [H+] and life history traits (i.e. linear, exponential, adaptation)

(Table 3.3). I expected scenario uncertainty (RCP 2.6 and 8.5) to be greatest as it essentially indicates diverging pathways for global CO2 concentrations, and this had the second highest average range of uncertainty (Table 3.3).

With other sources of structural uncertainty, my results were moderately sensitive to the choice of modeling approach, i.e. mechanistic effects on growth versus correlative effects on survival.

My results were relatively robust to the various ESMs. Sensitivity to parameter uncertainty was the smallest, suggesting my results are relatively robust to the magnitude of OA effects. This is an important finding as I used parameters for OA effects that were derived from meta-analyses for broad taxonomic groups. However, American cupped oyster showed very high uncertainty when different OA effect sizes were used.

The sensitivity to each level of model uncertainty was highly variable across species. Coonstripe shrimp showed the greatest range of uncertainty for all categories except parameter uncertainty

(Table 3.3) where they were affected by OA equally with the different effect sizes used (Table

3.1). This was surprising as they also showed a significantly large negative response to OA

(Figure 3.4). While results from other species suggest that my model is robust to OA parameter uncertainty, it is not true for all species. American cupped oyster abundance was most sensitive to parameter uncertainty with a range of 13.1% between outputs, but overall trends were consistently negative (Table 3.3). Furthermore, my results suggest that species most vulnerable to OA may be the ones that show greater negative effects of OA but are less sensitive to parameter perturbations, such as Pacific geoduck and coonstripe shrimp.

Structural i Parameter Scenario

ii. OA iii. Earth Mean range i. Mechanistic relationship system iv. OA v. RCP 2.6 of species vs. correlative form models effect size vs. 8.5 uncertainty

American cupped 4.6 6.1 1.9 13.1 8.1 6.8 oyster (-4.7, -0.1) (-14, -7.9) (-9.8, -7.9) (-21, -7.9) (-7.9, 0.2)

American lobster 0.5 1.4 0.3 0.5 0.7 0.7 (-0.6, -0.1) (-2.1, -0.7) (-0.7, -0.4) (-1.2, -0.7) (-0.7, 0.0)

American sea 1.0 4.0 1.5 2.7 2.3 2.3 scallop (-1.3, -0.3) (-6.1, -2.1) (-2.1, -0.6) (-4.8, -2.1) (-2.1, 0.2)

Blue mussel 0.1 0.2 0.0 0.1 0.1 0.1 (-0.1, 0.0) (-0.3, -0.1) (-0.1, -0.1) (-0.2, -0.1) (-0.1, 0.0)

Dungeness crab 0.2 6.7 2.2 0.5 0.2 2.0 (-0.3, -0.1) (-7.0, -0.3) (-2.5, -0.3) (-0.8, -0.3) (-0.5, -0.3)

Longfin inshore 0.6 1.7 0.5 0.5 1.3 0.9 squid (-0.9, -0.3) (-2.8, -1.1) (-1.1, -0.6) (-1.6, -1.1) (-1.1, 0.2)

Northern prawn 1.1 4.0 0.8 1.9 2.0 2.0 (-1.3, -0.2) (-5.6, -1.6) (-1.6, -0.8) (-3.5, -1.6) (-1.6, 0.4)

Pacific geoduck 6.2 3.5 5.8 0.2 2.7 3.7 (-6.6, -0.4) (-9.7, -6.2) (-6.2, -0.4) (-6.4, -6.2) (-6.2, -3.5)

Snow crab 1.0 0.8 0.3 1.5 1.1 0.9 (-1.2, -0.2) (-2.2, -1.4) (-1.4, -1.1) (-2.9, -1.4) (-1.4, -0.3)

Coonstripe shrimp 7.6 9.9 9.5 0.1 9.7 7.4 (-8.3, -0.7) (-19.2, -9.3) (-9.3, 0.2) (-9.3, -9.2) (-9.3, 0.4)

Mean range of 2.3 3.8 2.3 2.1 2.8 model uncertainty

†Values shown are percent changes for the annual 2091-2100 average relative to 2000. The default scenarios/parameters held constant when testing each source of uncertainty were: i) including both mechanistic and correlative OA impacts; ii) baseline linear relationship between OA and effect size (Figure 3.1a); iii) NOAA’s

GFDL earth system model; iv) mean effect size for OA impacts on growth and survival (Table 3.1); and v) high CO2 scenario (RCP 8.5).

3.4 DISCUSSION

My results present an analysis of the assumptions and uncertainties in modelling the biophysical impacts of OA using a spatially explicit model. Results of changes in abundance show sensitivity to all levels of uncertainty tested, although this varied with species and the type of uncertainty. In modelling OA effects, my results were sensitive to important assumptions of impacts on different life history rates, structural relationship between [H+] and life history rates, and parameter estimates. OA impacts on abundance were consistent across species and generally showed a negative relationship with increased acidity (Figure 3.4). Any measured increase in abundance due to OA was associated with the low CO2 scenario and generally small (Table 3.3). However, some species increased in overall abundance in response to all global change stressors (Figure

3.6). This trend is a result of the ability of a species to disperse and recolonize habitats that become more favourable with environmental change.

Temperature-driven geographical shifts and range expansion could overshadow any small negative effects of OA. Temperature effects can outweigh effects of OA on performance such as on growth and calcification (e.g. McNeil et al., 2004; Paul et al., 2015). Organism responses to temperature and OA interactions are complex across species, and previous studies have revealed synergistic, additive, and antagonistic interaction effects (Munday et al., 2009b; Harvey et al.,

2013; Kroeker et al., 2014). However, there have yet to be any empirical studies that show that temperature-driven geographical range shifts can outweigh any negative impacts of OA. My results show overall increases in abundance for some species due to temperature-driven range shifts despite the negative effects of OA (Figure 3.6). OA effects on Dungeness crab abundance appear to diminish with greater increases in abundance after year ~2065, suggesting that range

shift and expansion outweighs the effects of OA (Figure 3.4e and 3.6e). Furthermore, such increases in abundance are not ubiquitous across the species range and may only apply to certain areas (e.g. poleward limits of geographical range), whereas other areas may see significant decreases (e.g. equatorward limits of geographical range) (Figure 3.7).

Variability in my model results shows the importance of parameter selection, uncertainty, and the various underlying assumptions when integrating OA effects in species distribution models.

Previous studies that have incorporated OA usually address uncertainty by using a range of effect sizes, typically derived from meta-analyses and empirical studies (Ainsworth et al., 2011;

Cheung et al., 2011). My results suggest that the model is generally robust to OA parameter uncertainty, but some species were highly sensitive to parameter perturbations.

Using generalized parameters across taxonomic groups is sufficient for broad-scale analyses or comparisons (e.g. Lam et al., 2014), but integrating species-specific data into models is essential when evaluating individual species for management purposes—especially for species that are highly sensitive to sources of uncertainty. With the growing literature on impacts of OA on marine life, projection models can utilize updated empirical data as input parameters to provide more accurate representations of OA effects (Ries et al., 2009).

Incorporating OA effects on different life history traits and the downstream effects on abundance provides novel insight to the alternative approaches that must be considered when constructing models (Kroeker et al., 2017). A mechanistic approach requires an understanding of the drivers and underlying biophysical processes and is therefore more difficult to incorporate, as more information is typically required. However, with a greater understanding of the mechanisms at play and the specific information needed, the responses to OA may be more realistic and 70

accurate. Alternatively, correlative models offer a more straightforward approach, and are often easier to interpret. Correlative models may provide equally accurate results without the additional assumptions needed to incorporate mechanistic processes. My model results indicate that the mechanistic approach has greater effects on abundance, even when the parameters used for survival were much greater (i.e. molluscs; Table 3.1). The correlative approach provides a more direct effect of OA on population dynamics, but assumes that underlying OA effects on physiological processes translate to changes in survival.

Using a mechanistic approach relates OA to changes in aerobic scope and the subsequent trade- offs with other life history parameters (e.g. growth rate, maximum body size, survival) (Pörtner,

2008). For example, natural mortality was modelled to be dependent on growth traits and temperature (equation 3.9). Indirect effects of OA on other traits—such as fecundity—also emerge as we model them as a function of body size. The mechanistic approach allows us to address some of the complexities of OA effects and how it scales from physiology to population biogeography. However, OA affects many other physiological processes and life histories and are highly variable across species; current knowledge of the exact mechanisms and trade-offs between life histories are still not fully understood and there are other competing models (Pörtner et al., 2006; Ries et al., 2009; Kroeker et al., 2013). Pauly's (1980) model for natural mortality was chosen over other alternative empirical equations (e.g. Gislason et al., 2010; Then et al.,

2015) because of its simplicity and the availability of life history parameters for the invertebrate species considered here. Physiological models used here were chosen because they provide a parsimonious representation, despite not being ubiquitous across all species. While there are other working hypotheses for modelling physiological responses to environmental change (e.g.

Clark and Sandblom, 2013; Lefevre et al., 2017) these tend to be much more complex, requiring a deeper knowledge of the mechanistic processes. Indeed, future studies should test the sensitivity of the DBEM to various equations of natural mortality and physiological processes with more experimental evidence. Using both correlative and mechanistic models allows the model to address generalized OA effects on population dynamics, while incorporating effects on organism biology that can have profound effects at the population level.

Few studies have considered the relationship between changes in OA and organism responses when modelling OA impacts on biological systems. The rule of parsimony favours a linear relationship, yet there is evidence of varying relationship curves (e.g. Anthony et al., 2011;

Kleypas et al., 2006; Ries et al., 2009). This is especially important with greater changes in OA.

For example, using the exponential relationship as a ‘threshold’ or ‘tipping point’ scenario, responses are much greater with high acidification but the onset of responses only occurs with a certain degree of acidification (Figure 3.5). Tipping points for OA impacts can have drastic consequences to marine ecosystems and can manifest in ways such as direct impacts on physiological processes (Monaco & Helmuth, 2011) to indirect impacts on habitat availability

(Sunday et al., 2017). Furthermore, experimental studies often focus on acute changes in pH, yet natural OA is a combination of gradual declines in the average pH and dynamic, more extreme temporal fluctuations in pH. Gradual changes in pH may favour species with high evolutionary potential (Sunday et al., 2011; Lohbeck et al., 2012) and more extreme fluctuations in pH may favour species with wide pH range tolerance (Haigh et al., 2015; Ellis et al., 2016). With knowledge of the evolutionary potential and OA tolerance of a species, future studies could

apply a combination of an ‘adaptation’ model and alternative relationship curves to better represent long-term effects of OA on species distribution and abundance.

Species distribution models are valuable tools for projections of future global change, for both natural environments and the societies that depend on ecosystem goods and services. In addition to being some of the most susceptible organisms to OA, I evaluate marine invertebrates as they are ecologically important as key intermediate links to ecosystem structure and function (Reed,

2002; Steneck et al., 2002; Fagerli et al., 2013), and important economic resources as some of the most valuable fisheries species (Cooley & Doney, 2009a; FAO, 2016). Biophysical projection models can then be coupled with socioeconomic scenarios that present possible management regimes for fisheries resources (e.g. Costello et al., 2016). The model does not incorporate feedbacks of fisheries responses (e.g. profitability, management, activity) to changes in stock biomass, but it is an essential component to accurately model the dynamic interplay between environmental and anthropogenic pressures and responses of fisheries stocks and ecosystems. Results from projection models can also help inform future decisions and link to policies and agreements—e.g. projections of climate change impacts on global marine fisheries have shown the importance of meeting international targets to reduce emissions for both fisheries catch (Cheung et al., 2016a) and revenues (Lam et al., 2016). There are a number of studies that provide projections of the potential OA impacts to marine fisheries through extensive literature reviews and risk analyses frameworks (Le Quesne & Pinnegar, 2012; Branch et al., 2013;

Ekstrom et al., 2015; Mathis et al., 2015). I envision the use of my results to develop a better understanding and potentially reducing the uncertainty with projecting OA impacts on living marine resources.

My results of scaling OA effects from physiology to population biogeography provide useful insight to the underlying uncertainties and sensitivities. Constructing projection models such as the one here requires sound theoretical and empirical information. Further progress in reducing uncertainties can be made with more integration across disciplines and communication between the inputs and outputs needed by experimental biologists, ecologists, and modellers. For example, data from ocean acidification and warming experiments were used in a projection model to estimate potential impacts to species distribution and abundance, all in one synthesized paper (Queirós et al., 2015). With accelerated OA and climate change leading to the emergence of more downstream consequences, it is essential to continue to reduce uncertainties of projection models to help inform mitigation and adaptation strategies.

Chapter 4: Ocean acidification amplifies multi-stressor impacts on global marine invertebrate fisheries

4.1 INTRODUCTION

A direct consequence of elevated atmospheric CO2 concentrations is the rapid rate of ocean acidification (OA) (IPCC, 2013), causing changes to the biogeochemical composition of our world’s ocean and affecting marine ecosystem goods and services. Global sea surface pH has already decreased by 0.1 units since the industrial revolution, and is projected to decrease by a total of 0.41 units by the year 2100 (IPCC, 2013), which is a 140% increase in acidity. How organisms will respond to OA effects will vary across populations, taxonomic groups, and ecosystem types (Kroeker et al., 2013; Nagelkerken & Connell, 2015). Particularly, calcareous species (e.g. mussels, oysters, coccolithophore plankton, corals) are most vulnerable to OA as

CO2 interferes with the formation of calcium carbonate (CaCO3) structures (Fabry et al., 2008;

Ries et al., 2009). While much less understood, OA also affects various physiological processes such as acid-base regulation, metabolism, aerobic scope, physiological limits, sensory abilities, reproduction and development (Le Quesne & Pinnegar, 2012). These effects can then lead to changes in population level dynamics such as growth, survival, and fecundity, and ultimately affect marine resources (Kroeker et al., 2013).

Interactions between OA and other concurrent climate change stressors (e.g. ocean warming, decrease in oxygen content) could exacerbate their impacts on marine species and ecosystems. In ectotherms, oxygen demand increases with temperature to maintain basal metabolic rates

(Pörtner & Lannig, 2009). This reduces the aerobic scope and the relative supply of oxygen put

towards other aerobic processes such as growth (Figure 4.1a). We see similar effects on aerobic scope with a reduction in dissolved oxygen concentration (Pörtner & Lannig, 2009). OA is proposed to operate in a similar manner, reducing the overall aerobic scope profile (Pörtner &

Farrell, 2008) (Figure 4.1a), which can then have effects on life history traits (Pauly & Cheung,

2017) (e.g. growth rate, maximum body size) and affect large-scale population dynamics

(Cheung et al., 2011) (Figure 4.1). However, this mechanism is not ubiquitous and lacks general applicability (Clark & Sandblom, 2013; Lefevre et al., 2017; Jutfelt et al., 2018), yet simpler alternative mechanisms for scaling physiological effects to life history traits have yet to be proposed or identified. Studies that examined the effects of multiple climate change stressors on marine organisms are mostly limited to laboratory or small-scale field-based mesocosm experiments (Fabry et al., 2008; Ries et al., 2009), but are extremely useful for informing projection models.

Understanding the implications of the potential multi-stressor interactions for biological communities and fisheries resources at regional and global scales are important for informing climate change mitigation and adaptation policies. Here, I used a DBEM to project direct physiological impacts of changes in pH, temperature, and oxygen content on the spatial distribution of commercially exploited marine invertebrate populations (Pörtner & Lannig, 2009;

Cheung et al., 2011; Pauly & Cheung, 2017)—the species group known to be most sensitive to

OA. My model integrates the OCLTT model (Pörtner & Lannig, 2009) and the gill-oxygen limitation (GOL) model (Pauly & Cheung, 2017) to determine biological responses to environmental stressors and the downstream effects on fisheries catch potential (Cheung et al.,

2011, 2016a) (Figure 4.1). I focus on using these specific hypotheses for how OA may interact

with other stressors in impacting marine organisms—other potential mechanisms are discussed below.

a Oxygen- and capacity- limited b von Bertalanffy growth c Dynamic bioclimatic thermal tolerance model model envelope model

Sopt

w∞ Sopt Sopt ? S1 S1

S1 Hypoxia S S2 2 Weight (g)

Aerobic scope Hypoxia + CO2 ? S2 ∆ maximum catch potential (%)

Temperature (˚C) Age (years) Time (years)

Figure 4.1. Modelling the pathway of the impacts of ocean acidification from organism to population level in a multi-stressor framework. OA impacts were modelled as (a) physiological impacts (Pörtner & Lannig, 2009), then translated to (b) impacts on growth (Cheung et al., 2011), and (c) impacts on fisheries catch potential (Cheung et al., 2016a). Sopt is the optimal temperature scenario at which aerobic scope is at maximum. Other stressors such as hypoxia shrink the overall performance (blue curve) and reduce the overall aerobic scope (S1) (panel a), leading to reductions in growth rate and maximum attainable size (w∞) (panel b), as well as potential decreases in fisheries catch (panel c). Multi-stressor impacts of temperature, hypoxia, and acidosis will exacerbate physiological limitations and impacts on fisheries catch (red curve; S2).

4.2 METHODS

4.2.1 Dynamic bioclimatic envelope model

Changes in catch potential of commercially exploited species were estimated using a DBEM

(Cheung et al., 2008a, 2011, 2016c). The DBEM predicts how species abundance will change in space (on a 0.5˚ longitude by 0.5˚ latitude grid) and time (annual) using an integrative approach by linking species distribution models (Jones et al., 2012), growth models (Pauly, 1980a), physiological models (Pauly, 1981), population dynamics models (Pauly, 1980a; Hilborn & 77

Walters, 1992; O’Connor et al., 2007), and macroecological models (Cheung et al., 2008b).

Changes in species abundance and catch potential are driven by environmental conditions (e.g. temperature, dissolved oxygen concentration, pH) and habitat (e.g. substrate). I used environmental data from ESM projections (Dunne et al., 2013) as inputs to drive changes in species abundance. I describe the necessary components of the DBEM below; for a thorough description of the model, see Cheung et al. (2008a, 2011, 2016c).

Initial species distributions were obtained from the SAU database (Close et al., 2006; Jones et al., 2012; Pauly & Zeller, 2015). Their methods use a sequence of criteria to identify and restrict the habitat of each species on a 0.5˚ longitude by 0.5˚ latitude geospatial grid (for a detailed description see http://www.seaaroundus.org/). First, species habitats were restricted based on recorded presences in FAO area(s). Next, species habitats were restricted to a north-south latitudinal range within the FAO area(s). An expert reviewed range-limiting polygon using observation data and primary literature was overlaid over the map generated from the previous step (Close et al., 2006). Finally, species-specific parameters (primarily collected from

SeaLifeBase (Palomares & Pauly, 2017)) such as depth range, habitat preference, and equatorial submergence were used to determine the final distribution map.

Species-specific habitat suitability was characterized by overlaying environmental variables such as temperature, salinity, depth, sea-ice, and dissolved oxygen concentration over the initial species distribution maps. Habitat preference was also incorporated to characterize a bioclimatic envelope and habitat preference profile for each species (Cheung et al., 2008a). Carrying capacity is determined using the initial species distribution and is positively correlated with habitat suitability (Cheung et al., 2008a, 2016c). A major assumption of the DBEM is that each

cell from the initial species distribution is at carrying capacity. As habitat within each cell changes and becomes more suitable, the carrying capacity will increase.

Individual growth is represented by the von Bertalanffy growth model with species-specific parameters and are constrained by ecophysiological conditions such as oxygen and temperature

(Cheung et al., 2011). Population growth is represented by the logistic growth function (Hilborn

& Walters, 1992) and population mortality rates are calculated from an empirical equation from

Pauly (1980b).

Movement of the organisms at different life stages were represented in the DBEM. Larvae disperse using advection-diffusion models (Sibert et al., 1999; Cheung et al., 2008a), while net adult diffusion rate is determined by a fuzzy logic model. Animals actively move based on the distance between two geographic cells and their relative dispersal rate (e.g. large-bodied pelagic species, small reef-dwelling demersal species). Emigration rates are greater if neighbouring cells have a more favourable habitat, while immigration rates are greater if the present cell is preferable to surrounding neighbour cells.

Annual fisheries maximum catch potential (MCP) for each species was estimated by summing the maximum sustainable yield for each occupied spatial cell. Maximum sustainable yield is assumed to be equal to Ki/2, where Ki = carrying capacity of a spatial cell i.

4.2.2 Modelling ocean acidification impacts in a multi-stressor framework

I modelled the impacts of OA on species abundance in a multi-stressor framework through somatic growth and mortality rates by combining the OCLTT and the GOL hypothesis (Tai et

al., 2018). First, I modelled growth rate, ��/��, as a function of oxygen supply (anabolism) and oxygen demand for maintenance metabolism (catabolism) (Cheung et al., 2011):

!" = ��! − ��! (4.1) !" where B is species biomass, and H and k represent the coefficients for anabolism and catabolism, respectively. Anabolism scales with body weight (W) to the exponent d < 1, while catabolism scales linearly with (W) i.e., b = 1. Values of d typically range from 0.5 to 0.95 across species

(Pauly, 1981; Pauly & Cheung, 2017; Tai et al., 2018), but I assume a mid-range value of d =

0.7. Higher values of d leads to increased sensitivity of growth to temperature, while lower values lead to marginal decreases in sensitivity (Pauly & Cheung, 2017; Tai et al., 2018).

Therefore, d = 0.7 is a conservative measure to use as a scaling coefficient across all species.

Effects of environmental stressors (i.e. temperature, oxygen concentration, and pH) are modelled to affect both growth rate parameters H and k. First, temperature affects both oxygen supply and demand for metabolism (Pauly & Cheung, 2017) following the Arrhenius equation (�!!/!).

Next, oxygen availability affects overall aerobic scope, while other physiological stressors (i.e. acidification) affects maintenance metabolism (Cheung et al., 2011):

!!!/! � = �[O!]� (4.2) and

� = ℎ[H!]�!!!/! (4.3)

The Arrhenius equation constants �! and �! (for anabolism and catabolism, respectively) are equal to Ea/R, where Ea and R are the activation energy and Boltzmann constant, respectively. T

+ is the absolute temperature (in Kelvin). Relative changes in oxygen [O2] and hydrogen ion [H ] concentration thus change H and k, respectively. These impacts on growth rate can first be depicted as impacts to aerobic scope (Figure 4.1a). Coefficients g and h were derived for each species from the maximum weight, von Bertalanffy growth rate parameter, and average environmental temperature reported in the literature (Cheung et al., 2011; Palomares & Pauly,

2017).

With changes in aerobic scope due to environmental stressors, the model predicts changes in life history parameters including asymptotic weight (W∞) and the von Bertalanffy growth rate parameter K (Figure 4.1b):

!/(!!!) � = ! (4.4) ! ! and

� = �(1 − �) (4.5)

Pauly’s empirical model (1980a) was used to estimate natural population mortality rates (M):

� = −0.4851 − 0.0824 log �! + 0.6757log (�) + 0.4687 log (�′) (4.6) using maximum body size, growth rate, and the average water temperature of a species range in degrees Celsius (�′). This model was chosen as it is widely used and life history data is readily available for all of the invertebrate species tested here (Tai et al., 2018).

Second, I modelled OA impacts on survival rates using a correlative approach for both larvae and adults. I measure changes in acidity and its impacts on growth and survival rates as hydrogen concentration [H+]. I model changes in life histories based on the model:

! ! ! ! ��! = ��!"!# ∗ 1 + �� ∗ ! − 1 (4.7) ! !"!#

Surv is the survival rate per year and used here as an example but is also applied to growth.

Survival rate in year t is equal to the initial (init) survival rate and the relative change in [H+] between year t and initial [H+] conditions. I define Per as the value of the OA effect size (from

Kroeker et al., 2013) for the percent change in survival rate with a doubling of [H+] (Appendix

Table C.1). I assume a linear relationship where the exponent w is equal to 1 because it is the most parsimonious assumption considering the highly variable responses across species (Ries et al., 2009; Tai et al., 2018). OA effect sizes for both growth and survival were assigned based on taxonomic groups (Appendix Table C.1). Thus, environmental changes will lead to changes in growth rate, maximum body size, and survival rates. I used the DBEM to scale up and determine how the physiological responses to OA and other stressors affects species distributions and fisheries catch potential (Figure 4.1).

4.2.3 Modelled species

I modelled the impacts of OA and climate change on 210 commercially exploited marine invertebrate species. Invertebrate species tend to be more sensitive to OA and changes in pH

(Kroeker et al., 2013), while the effects on finfish species are generally less sensitive and show greater variation. I included species from the major shellfish fisheries groups including crabs,

lobsters, shrimp and prawns, oysters, scallops, and squid. Most of these species are distributed throughout coastal waters.

4.2.4 Model uncertainties

I quantified the sensitivity of the DBEM results (Cheung et al., 2016b) to three levels of uncertainty: parameter, structural, and scenario uncertainty (see also Chapter 3). Sensitivity to parameter uncertainty was examined by using the upper and lower 95% confidence limits for OA effect sizes on growth and survival rates. Structural uncertainty is defined here as the variance across different ESMs used as input environmental data. I present results from three ESMs produced by: NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL-ESM); Max Planck

Institute for Meteorology (MPI-ESM); and Institute Pierre Simon Laplace Climate Modelling

Centre (IPSL-ESM). These models provide high-resolution data, surface and bottom data, and the necessary environmental variables required by the DBEM (Cheung et al., 2008a, 2011).

ESMs use various levels of atmospheric greenhouse gas concentrations, known as representative concentration pathways (RCP) to develop scenarios of environmental change. In my analysis, I use a low (RCP 2.6) and a high (RCP 8.5) carbon emissions scenario for DBEM simulations.

The numbers represent radiative forcing values in the year 2100 relative to pre-industrial periods.

The low emissions scenario will occur if carbon emissions are strongly mitigated and annual global greenhouse gas emissions peak within a decade but then substantially decline. This trajectory is the closest RCP scenario that aligns with targets set with the 2015 ‘Paris Agreement’ from the United Nations Framework Convention on Climate Change Conference of the Parties.

The high emissions scenario is our current pathway, where carbon emissions are not curbed and the burning of fossil fuels continues to be the primary source of energy for the industrial sector.

With this scenario, minimal efforts are made to reduce carbon emissions or develop green energy sources.

4.2.5 Analysis

Changes to species abundance were calculated as percent changes between initial and final conditions:

(! !! ) % �ℎ�� = !"# !"!# · 100 (4.8) !!"!# where � is the mean abundance of a 10 year period. Change in MCP was also calculated using this formula.

Latitudinal centroid LC for each species was calculated by multiplying the abundance of each occupied cell by the latitude Lat of each cell:

! !!! !"#!∗!! �� = ! (4.9) !!! !!

The rate of species distribution latitudinal shift was estimated by finding the slope of a linear regression of the latitudinal centroid for each year. The rate of latitudinal shift was converted to kilometers by multiplying the estimated slope by !" where � = 6378.2, the approximate radius !"# of the earth. This rate was then converted to decadal shifts.

Species range size for each year was calculated by multiplying each occupied 0.5˚ longitude by

0.5˚ latitude cell with the average area of the cell. Projected percent changes in range size were calculated using the same structure as equation (4.8).

Effects of OA were isolated from effects of other global change stressors by finding the difference in the outputs between simulations run with modelled effects of OA and without OA.

OA is modelled to amplify responses in aerobic scope and subsequent life histories, and these effects were isolated to illustrate the separate and additive effects of OA with other critical global change stressors (i.e. ocean warming, decreased oxygen content) (Cheung et al., 2016a).

4.3 RESULTS

Intermediary non-spatially explicit results show that modelled impacts of ocean surface warming and acidification show amplified effects on aerobic scope, reducing aerobic scope by as much as

75% under the high CO2 scenario by the end of the century (Figure 4.2a). Continuing with this scenario, maximum body size was projected to decrease by as much as 66% (Figure 4.2b). Ocean acidification had substantial effects, accounting for over 60% of the reduction in aerobic scope and 32% of the decrease in maximum body size. Applying these models to include spatial dynamics with the DBEM, global invertebrate MCP was projected to decrease by about 12% in the high CO2 scenario due to physiological (e.g. ocean warming and acidification), and habitat constraints (e.g. primary production, habitat suitability), of which ocean acidification accounts for over 3% of this decrease (Figure 4.2c). Impacts under the low CO2 scenario are considerably lower, where aerobic scope and maximum body size were projected to decrease by 15% and

23%, respectively, while the change in MCP was negligible (Figure 4.2).

a b c w 1960 2100 ∞ 0 5 0 2050 –11 2100 2000 ∆

–23 (%) size body maximum

2100 0 ∆ −20 0 10 [H acidity, Low CO 2050 20 –34 2 30 High CO2 −40 50 + eight (g) MCP (%)

](%) −5 W ∆ OA

aerobic scope (%) –66 −60 No OA

∆ 100

−10 −80 2100 150

0 1 2 3 4 1950 2000 2050 2100 ∆ SST (˚C) Age (years) Year

Figure 4.2. Scaling the multi-stressor responses from the organism level to fisheries catch under the low CO2 and high CO2 climate change scenarios (blue and red, respectively), averaged across all modelled species (N = 210). Differences between projections with and without the modelled effects of ocean acidification (OA) are shown with solid and dashed lines, respectively. (a) Effects of projected ocean warming and acidification on aerobic scope for growth. (b) Change in the von Bertalanffy growth curve and maximum body size in the 2091-2100 period. (c) Changes in global maximum catch potential (MCP) projected by the dynamic bioclimatic envelope model. Results shown are relative to the 1951-1960 period and are multi-model averages from the three earth system models used in this study.

Impacts of global change on the maximum catch potential (MCP) of marine invertebrate fisheries shows regional variation where tropical regions will generally see a loss in catch while northern regions will see an increase if we continue on the “business-as-usual” high CO2 trajectory relative to the strong mitigation low CO2 trajectory (Figure 4.3). Increases in catch at higher latitudinal regions are largely driven by ocean warming that result in species distribution shifts and increased species turnover (Cheung et al., 2016a). Regions around the coral triangle such as the Indonesian Sea are projected to lose the most (Cheung et al., 2016a), and the models project decreases of >30% for invertebrate catch potential.

Differences in ∆ MCP between RCP scenarios (high – low CO2)

<–25% –10% –2.5% 2.5% 10% 25% >50%

Figure 4.3. Projected multi-stressor impacts on maximum catch potential (MCP) of marine invertebrates across large marine ecosystems. Results shown are for the 2091-2100 period (relative to 1951-1960) in high

CO2 (RCP 8.5) relative to low CO2 scenario (RCP 2.6).

OA is projected to decrease annual global invertebrate fisheries MCP by 0.75% for each 0.1 unit decrease in pH, and 3.4% by the end of the century (Figure 4.2c and 4.4a), although this is highly variable across regions. West Arctic large marine ecosystems—including North Bering, Chukchi

Sea, Beaufort Sea, Queen Elizabeth Islands archipelago, Canadian high Arctic and North

Greenland—are likely to be most susceptible to OA where pH is projected to decrease by up to

0.5 units (Figure 4.4b), offsetting potential gains in catch potential primarily driven by warming

(Figure 4.3). While catch potential is projected to increase overall in Arctic regions, OA will largely reduce gains in potential catches for species sensitive to OA (Lam et al., 2014). My projections show that these impacts due to OA are exacerbated with greater changes in surface temperature and oxygen concentration (i.e. IPSL earth system model) (Figure 4.4b; Appendix

Figure C.1 and C.2). Currently, there are only a few commercial marine fisheries in the West

Arctic (Zeller et al., 2011), largely restricted by national and international agreements—such as

commercial fishing bans in the Beaufort Sea and central Arctic—due to our limited knowledge of the baseline state and the sustainability of fisheries exploitation across these regions (e.g.

Cobb et al., 2008).

a Global b West Arctic

1960 2060 2000 0 ● ● 0 ●●● ● ● ● ● ● ● 2020 1960 2060 2100 2100 2020 ●●●● ● ● ● 2000 2060 ● 2020 −10 −1 2020 ● ●

−20 ● −2 ● 2100 2060 −30 ● −3 −40 ● 2100

−4 ● Low CO2

A scenarios (%) −50

O ● High CO2 −0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0

c NE Pacifc d Central Indo−Pacifc GFDL

1960 1960 IPSL 0.0 ● ● 0 ● 2100● ● MPI 2060●●●● ● 2000 ● ● 2020 ● 2020 2100 ● 2000 2060 ● erence in ∆ MCP between f −2.5 ● ● 2060 2020 Dif ● −1 ● 2020

−5.0 ● ● 2060

−2 2100 ● ● −7.5 2100 ●

−10.0 −3

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

∆pH

Figure 4.4. Projected ocean acidification (OA) impacts on maximum catch potential (MCP) of marine invertebrates in addition to other climate change stressors. Thicker coloured lines are multi-model means and thinner lines are simulations with the different earth system models: GFDL – Geophysical Fluid Dynamics Laboratory; MPI – Max Planck Institute; IPSL – Institute Pierre Simon Laplace. Black lines and grey bands are selected regressions and 95% confidence limits. MCP data are smoothed by a 10-year running mean and relative to 1951-1960.

Large marine ecosystems of the Northeast Pacific Ocean show decreases in MCP up to 6% annually by year 2100 under the high CO2 scenario (Figure 4.4c). This region includes the highly productive fishing regions of the East Bering Sea, Gulf of Alaska, and California Current large marine ecosystems. Across this region, there are highly valuable capture fisheries including

Alaskan king crab and Dungeness crab fisheries, as well as open-system mariculture fisheries such as Pacific oyster and geoduck fisheries (Tai et al., 2017). Other studies using ecosystem models of the Northeast Pacific also found amplified negative impacts on species abundance with multiple global change stressors (Ainsworth et al., 2011). Similarly, projections of OA impacts on the California Current ecosystem showed negative direct impacts on epibenthic invertebrates and downstream indirect impacts on higher trophic level species assemblages

(Marshall et al., 2017).

Impacts of OA on fisheries catch potential in the Central Indo-Pacific region (i.e. Gulf of

Thailand, South China Sea, Sulu-Celebes Sea, Indonesian Sea) were much less significant, where

MCP decreases an additional 2% by year 2100 in the high emissions scenario (Figure 4.4d).

Overall, catch potential across tropical regions is projected to substantially decrease overall due to global change (Cheung et al., 2016a), and while the impacts of OA are negative, they may be overshadowed by temperature-driven changes (McNeil & Sasse, 2016). However, OA effects on critical habitat forming species (e.g. corals, mussels) or key intermediate trophic-level species

(e.g. seastars) could lead to substantial ecosystem changes (Hoegh-Guldberg et al., 2007; Sunday et al., 2017).

a 100 Banded carpet shell

Half-crenate ark ∆ abundance (%) )

−1 500 10 100 Sand Crayfsh 10

1 Northern quahog 0 −1 1

Latitudinal shift (km•decade −10

Atlantic bay scallop

0 −100 −10 −1 0 1 10 100 500 Crustaceans

b Molluscs

∆ abundance (%) ) 1 −1 Half-crenate ark 30

20 0 10

0 −1 −10

Sand Latitudinal shift (km•decade −20 Crayfsh Northern −30 −10 quahog Atlantic bay scallop

−50 Banded carpet shell

−50 −10 −1 0 1 ∆range size(%)

Figure 4.5. Biogeographical changes in range size, abundance, and distributional shift of latitudinal centroids for 210 invertebrate fisheries species in the high CO2 scenario. (a) Responses to global change stressors excluding ocean acidification, and (b) responses to ocean acidification separated from other stressors. Values shown are multi-model means for 2091-2100 period relative to 1951-1960. Correlations between variables are shown in Appendix Table C.1. Note log scales.

Biogeographical changes of range size showed a positive correlation (r = 0.78; Appendix Table

C.2) with changes in abundance (Figure 4.5a), while absolute changes in range sizes were positively correlated with increased rates of latitudinal centroid shift (r = 0.52; Appendix Table

C.2). For most species, OA had negative impacts on abundance and range size, as well as decreased rates of latitudinal shift (Figure 4.5b). Species that had large decreases in abundance and range size, quicker rates of latitudinal shift, and exacerbated effects due to OA are likely to be at greatest risk to global change (e.g. banded carpet shell, Atlantic bay scallop) (Figure 4.5).

Such species may also face substantially elevated risk of extinction as population viability is generally positively correlated to range size (Purvis et al., 2000). Some species such as the northern quahog showed positive responses (i.e. range expansion, abundance increase) to global stressors (Figure 4.5a) and negative responses to OA, likely due to an increase in suitable habitat but limited by the sensitivity to OA. Mollusc species showed greater losses in catch potential in high CO2 scenarios when compared with crustacean species (Appendix Figure C.3), explained by the greater effect size for the parameters used for molluscs than crustaceans (Appendix Table

C.1). However, changes in catch potential for molluscs showed more variability to the effects of

OA, suggesting that the interaction effects of OA and other stressors in my model are not consistent across mollusc species within the same group. Conversely, crustaceans appear to be more robust to OA than molluscs (Ries et al., 2009) possibly because they are generally more mobile and thus have the ability to quickly shift their geographic range.

Sensitivity analyses showed that my model results of OA impacts are most sensitive to parameter uncertainty. Parameter uncertainty (OA effect size) accounted for most of the uncertainty (50-

90%) in the first half of the model simulations, and decreased to ~60% in later years (Figure 4.6).

Therefore, it is imperative to obtain accurate empirical data for parameter effect size of OA responses in order to accurately project species responses to global change. This is especially important for more localized and species-specific analyses; my results summarize effects of OA

across large spatial extents and many species, therefore I used mean effect sizes across taxonomic groups. Furthermore, the proportion of total uncertainty was smallest for model uncertainty, suggesting the results are robust to the different structures of ESMs at the global scale. Scenario uncertainty initially accounted for >25% of total uncertainty but its absolute uncertainty was negligible during this early part of the simulation. Scenario uncertainty increased at the year (~2010) where environmental conditions diverged (Appendix Figure C.2) between low and high CO2 scenarios to account for >30% of the uncertainty by the end of the simulations

(Figure 4.6).

One component of uncertainty not specifically tested here is the various models of mechanistic physiological responses to environmental stressors. The models used here are much less complex than the alternatives (e.g. Then et al., 2015), which generally require a more thorough understanding of the mechanisms involved (Lefevre et al., 2017) and life history parameters that are not readily available for all species tested. Previous studies have highlighted that the chosen models are not universally applicable (Lefevre et al., 2017; Pörtner et al., 2018a), yet the

OCLTT model provides connections of physiological constraints to higher ecological phenomena (see response to Jutfelt et al., 2018). Thus, by combining empirical physiological and life history models, I was able to link environmentally driven changes in aerobic scope to trade- offs with life history parameters (Tai et al., 2018)—e.g. growth, survival—which then scale to population dynamics.

0 ios (%) r

−2 MCP between scena

∆ −4

erence in Source of uncertainty f

Dif −6 Parameter b 100 Model

Scenario

75 tainty (%) r

25 ercentage of total unce P

0 1950 2000 2050 2100 Year

Figure 4.6. Testing model variability for the projected impacts on maximum catch potential (MCP) due to ocean acidification (OA). (a) Projected changes using the upper and lower bounds of parameter, model, and scenario uncertainty; and (b) the proportion of total uncertainty allocated to each source of uncertainty. Default conditions held constant when testing each source of uncertainty were: 1) parameter = mean OA effect size; 2) model = GFDL earth system model; and 3) scenario = RCP 8.5. Results are smoothed by 10- year running means and relative to the 1951-1960 average.

4.4 DISCUSSION

Our current understanding indicates that marine invertebrates are at most risk to the direct effects of OA (Kroeker et al., 2013), although these effects will likely differ across regions (Figure 4.3).

OA effects on marine invertebrate fisheries will have major implications for food security, livelihoods, and the global economy. At over 50 billion USD annually, invertebrate fisheries account for over 1/3 of the value of internationally traded seafood (FAO, 2014a). Invertebrates 93

support many valuable fisheries, especially in developing countries where over 50% of internationally traded seafood is caught (FAO, 2014a). Developing tropical areas are already projected to be most susceptible to global change (Cheung et al., 2016a) and OA could exacerbate food and income insecurities currently faced in these regions. For example, in areas across the Fijian Islands invertebrates comprise over 70% of artisanal and 50% of subsistence harvests (Teh et al., 2009). Most of the non-subsistence catch in Fiji is exported with the largest invertebrate fisheries valued at $8.4 million FJD (~$3.8 million USD) in 2003 (Teh et al., 2009).

Additionally, small-island developing nations such as the Pacific Islands are highly dependent on invertebrates for both additional income sources and subsistence (Kronen et al., 2010). Ease of access and the minimal gear requirements for invertebrate fishing has facilitated women and even children to participate in subsistence harvesting and remains a source of easily attainable

‘backup’ protein when fishing at sea is unsuccessful (Harper et al., 2013; Codding et al., 2014).

For some larger fishing nations invertebrate fisheries make substantial economic contributions.

In Canada, lobster fisheries were valued at over $1.1 billion CAD in 2015 and its exports contributed over $2 billion CAD to the Canadian economy (DFO, 2015). Similarly, sea scallops in the USA had dockside revenues estimated at $559 million USD in 2012 (Cooley et al., 2015).

Fisheries catch in regions at higher latitudes are projected to increase with global change

(Cheung et al., 2016a), but this may not equate to benefits to fishers and the economy. Decreases in highly valuable species (e.g. shellfish) may outweigh any increases in catch of less valuable species, potentially resulting in lost revenues (Lam et al., 2016). Prices will also be affected by changes in fish supply, resulting in changes to overall revenues (Sumaila et al., 2019). If OA has largely negative effects on organism function, this could further reduce overall catch and amplify

any losses in fisheries catch. Simulation modelling of OA effects on marine fisheries resources has largely suggested negative impacts to fisheries resources (Ainsworth et al., 2011; Cheung et al., 2011; Lam et al., 2014).

Projection models such as these provide valuable insight for possible future scenarios to identify regions and species that may be most sensitive to global change, and where to concentrate adaptation and mitigation efforts. However, the extent of OA impacts remains uncertain. Impacts of OA on fisheries has been widely discussed and resulted in qualitative and quantitative modelling efforts (Cooley & Doney, 2009b; Ainsworth et al., 2011; Ekstrom et al., 2015). These studies are informative and provide a baseline understanding of the potential OA impacts on species and possible downstream effects on fisheries resources. This spatially explicit, multi- stressor model provides a step towards better understanding how the distribution and abundance of fisheries resources will change at a global scale. Such large-scale models highlight regions that are most susceptible to OA and global change, such as the Arctic (AMAP, 2013), and areas where OA may be overshadowed by other drivers, such as temperature in the tropics (Cheung et al., 2016a). Variability in responses to the different stressors across regions and species indicate which environmental variables are of most concern. Then, the capacity of fisheries to adapt and mitigate global change impacts can be evaluated together with biological models to better inform management decisions and conservation efforts. My study supports previous findings that strong mitigation of CO2 emissions lead to benefits in fisheries catch potential (Cheung et al., 2016a), and that these benefits are further extended due to the minimization of OA.

Accurate projections of OA and global change impacts on marine fisheries require interdisciplinary integration to determine how multiple stressors interact to affect species at

various levels of biological organization. Thus, development of multi-stressor models requires collaboration between physiologists, biologists, and modellers. The development of modelling impacts of multiple drivers on marine resources is relatively new, yet numerous advances have been made to facilitate efforts and develop a thorough understanding of multi-stressor impacts

(Haigh et al., 2015; Koenigstein et al., 2016). My study contributes to the development of modelling efforts for global change, as well as to better understanding potential interactions of multiple stressors on the spatial distribution of marine fisheries resources.

Chapter 5: Evaluating present and future potential of Arctic fisheries in

Canada

5.1 INTRODUCTION

Arctic marine commercial fisheries in Canada remains small as access has largely been restricted by short fishing seasons, sea ice cover, dangerous navigation routes, and the geographic separation from populated areas increasing operating costs. However, climate change is quickly altering the seascape of Arctic environments by decreasing seasonal sea ice extent and altering biological productivity (IPCC, 2013; Pörtner et al., 2014). This has led to increased exploration for new shipping routes, oil and gas extraction sites, and the potential for commercial capture fisheries. The Arctic presents a unique opportunity to apply diverse current knowledge and tools to establish a baseline of environmental conditions and ensure sustainable practices.

Management of Arctic environments and the exploitation of marine resources in these areas are challenged with balancing sustainable and long-term viability of natural resources, preventing overexploitation, degradation, and pollution of marine environments, and maintaining the economic, social, and cultural wellbeing of Arctic communities (Singh et al., 2017).

Currently, commercial fisheries in Canada’s Arctic are primarily located in Canada’s Eastern

Arctic, in areas including Baffin Bay, Davis Strait, and Hudson Bay and Strait (Figure 5.1).

Fisheries catch in the remainder of Canada’s Arctic is primarily small-scale (community-based operations) and consisting of Arctic char (Salvelinus alpinus alpinus) (Pauly & Zeller, 2015).

Arctic marine ecosystem resources are important to northern communities in Canada, especially

Inuit, in terms of culture, food, livelihoods, and income (Hoover et al., 2016; Watts et al., 2017).

With the onset of climate change impacts on Arctic marine ecosystems, current fisheries may be affected and non-exploited regions may be pressured with the development of new fisheries.

Climate change is rapidly altering Arctic marine systems and will likely continue at a rate never seen before in human existence. Ocean warming has increased Arctic mean summer sea surface temperatures by ~0.7˚C decade-1 from 1982 to 2017 (Timmermans et al., 2017). Arctic sea ice has drastically changed since satellite records began in 1979. Winter maximum sea ice was at a record low in 2017 and second lowest in 2018, now four straight years with record low winter maximums, while the 10 lowest recorded summer minimum sea ice extents have occurred in the last 11 years (Perovich et al., 2017; Vizcarra, 2018). Consequentially, this may lengthen fishing seasons, open up new fishing grounds, and potentially bring increased abundance of commercially valuable sub-polar species as they shift to higher latitudes (Cheung et al., 2016a).

On the other hand, elevated atmospheric CO2 will also increase ocean acidity. Global sea surface pH has already decreased by 0.1 units since pre-industrial levels, equal to a ~26% increase in acidity, with greater pH declines at higher latitudes (IPCC, 2013). Global projections show sea surface pH to decrease an additional 0.3 units by century end, a 140% increase in acidity since pre-industrial levels (IPCC, 2013). Ocean warming and acidification are exacerbated by the reduction in sea ice cover—i.e. increased absorption of solar energy due to decreased reflection by sea ice, and increased surface area exposed to uptake of atmospheric CO2 (Steiner et al.,

2015). While the effects of warming will likely increase species invasion and turnover (Cheung et al., 2016a), the effects of acidification are less predictable and may have diverse impacts on species and their biomass (AMAP, 2018; Tai et al., 2018). Understanding the effects of climate

change on current and future species distribution and abundance is essential to fisheries management with the elevated interest in new Arctic fisheries.

It is projected that climate change will increase abundance of commercially valuable species in

Canada’s Arctic (Lam et al., 2014; Cheung et al., 2016a). Current commercial fishing moratoria in some of these regions (i.e. Beaufort Sea) give us time to utilize the vast network of tools that have been developed to sustainably manage resources. In this paper, two main objectives for

Arctic marine fisheries in Canada are covered: 1) taking stock of existing fisheries in its current state; and 2) building scenarios using an integrated modelling approach to estimate current and future fisheries potential under different climate scenarios. Here, I used projection models to estimate changes in distribution and abundance of commercially valuable polar and sub-polar species. Potential fisheries catch was estimated along with the potential catch value of commercial species across Canada’s Arctic. These areas represent the few areas in the world that are potentially underexploited and present a unique opportunity to extensively evaluate the regions that can lead to sustainable outcomes. Projection scenarios are essential tools for mitigating future impacts of global change. These results can be utilized as key inputs for analyses and assessments of Arctic fisheries to support management and sustainability of marine resources into a changing future. This study evaluates the current and long-term (century-end) potential of Arctic fisheries in Canada in terms of its contribution to food security and the economy and outlines the implications of the development of large-scale commercial fisheries in the Arctic.

5.2 METHODS

5.2.1 Study areas

Canada’s Arctic exclusive economic zone (EEZ) overlaps four major large marine ecosystems

(LME) in the Arctic: 1) Canada Eastern Arctic and West Greenland; 2) Hudson Bay complex; 3)

Beaufort Sea; and 4) Canada high-Arctic and North Greenland (Figure 5.1). Canada’s EEZ also overlaps part of the central Arctic LME, but it is not considered in this analysis due to the lack of information and knowledge of species distribution through this area. LME regions were developed by the US National Oceanic and Atmospheric Administration (NOAA) to identify large (>200,000 km2) ocean regions for ecosystem-based management and conservation (Kelley

& Sherman, 2018). Extents for the 66 defined LMEs are characterized by factors that constrain ecosystem areas: bathymetry, hydrography, productivity, and trophic linkages. Almost all (64 of

66) LMEs have experienced some degree of warming and it is estimated that 50% of all fish stocks within LMEs are overexploited (IOC-UNESCO & UNEP, 2016). The four Arctic LMEs that border Canada are defined by very different physical conditions (e.g. bathymetry, currents, nutrients), annual cycles (e.g. sea ice extent), ecosystem structure (e.g. species), and marine resource dependence critical to nutrition, livelihoods, and culture (Berkes, 1990; Christiansen et al., 2014; Cisneros-Montemayor et al., 2016; Watts et al., 2017).

100

Figure 5.1. Large marine ecosystem boundaries of Canada’s Arctic and Canada’s Exclusive Economic Zone (EEZ). Large marine ecosystems: BS = Beaufort Sea; CEA-WG = Canada East Arctic and West Greenland; HAA = High Arctic Archipelago (Canada high Arctic and North Greenland); HB = Hudson Bay Complex.

5.2.1.1 Canada Eastern Arctic and West Greenland – LME #18

Davis Strait and Baffin Bay make up a large portion of the ~315,000 km2 of Canada Eastern

Arctic and West Greenland LME (CEA-WG). Primary productivity in this region is relatively high with inputs of nutrient-rich waters from the Pacific (Tremblay et al., 2002; PAME, 2013).

Sea ice covers this entire region throughout the months of January-April, with the exception of the southeastern areas off Greenland and the North Water Polynya, which acts as a significant silicate pump to deep water nutrient storage (Tremblay et al., 2002). Arctic cod (Boreogadus saida) is a significant forage species in this ecosystem—and the Arctic as a whole—supporting higher trophic levels including fish, birds, and marine mammals (Loseto et al., 2009; Divoky et al., 2015; IOC-UNESCO & UNEP, 2016). This region also serves as a migration route and feeding ground for many marine mammal and bird species.

101

Commercial fisheries are most active in the CEA-WG relative to the other Arctic LMEs in

Canada. The relative ease of access to this region for major fishing fleets in Canada, US and

Europe, as well as a longer fishing season has enabled the development of commercial fisheries.

Northern prawn represents over two-thirds of current fisheries catch in this region (IOC-

UNESCO & UNEP, 2016), while various flatfish species (e.g. Greenland halibut) are also important. The coastal population in this LME is the largest at ~70,000 (in 2010) from Greenland and Nunavat (IOC-UNESCO & UNEP, 2016).

5.2.1.2 Beaufort Sea – LME #55

Canada’s Western Arctic region is largely comprised of the Beaufort Sea LME (BS), which covers an area of ~622,000 km2. Sea ice cover and light penetration limit primary production in the BS, but nutrient-rich Pacific water and riverine inputs from the Mackenzie River provide a large influx of nutrients that support a large diversity of organisms (PAME, 2013). Arctic cod is the dominant forage fish species in the BS and is a significant trophic linkage for this ecosystem that supports major summer feeding grounds for bowhead and beluga whales (Parker-Stetter et al., 2011; PAME, 2013). Further, the Mackenzie River delta provides major breeding, feeding, and migration habitats for a number of seabird, shorebird, and waterfowl species (Dickson &

Gilchrist, 2002).

Marine fisheries catch in the BS is exclusively for subsistence purposes, as there are no extant commercial fisheries (Ayles et al., 2016). The US fishing moratorium in 2009 and the development of Canada’s Integrated Fisheries Management Plan—an agreement between the

Inuvialuit and Canada to manage marine mammal and fish resources cooperatively in the

Beaufort Sea (Cobb et al., 2008; Niemi et al., 2012)—between 2011-2014 prevents any 102

commercial fishing in the BS until there is more information available about the ecosystem and its capacity to sustain marine resource exploitation (“North Pacific Fisheries Management

Council, Arctic Fishery Management Plan for Fish Resources of the Arctic Management Area”;

Ayles et al., 2016). Subsistence marine fisheries are largely comprised of Arctic char (Salvelinus alpinus), while commercial fisheries, made up of small-scale fleets, primarily target various whitefishes species (Coregonidae) (Fisheries and Oceans Canada et al., 2014; Pauly & Zeller,

2015; IOC-UNESCO & UNEP, 2016). Marine mammals (e.g. seal, beluga, narwal) are also significant sources of food across this region (Hoover et al., 2016; Watts et al., 2017). The BS is of significant importance to the culture, livelihoods, food, and income to the ~18,000 (as of

2010) people inhabiting this coastal region (IOC-UNESCO & UNEP, 2016).

5.2.1.3 Hudson Bay complex – LME #63

The Hudson Bay LME (HB) is a semi-enclosed body of water that has an area of about 1.1 million km2, the largest of the four considered here and the only one completely under Canada’s jurisdiction. While the LME as a whole is low in productivity, many coastal, estuarine, and near island embankments have increased primary production due to nutrients from river runoff and the periodic upwelling of deep nutrient-rich waters (PAME, 2013). Through the winter, the HB is covered by sea ice, with the exception of polynyas along the northwest and east areas. These polynyas are biologically important features in the sea ice that heavily influence the physical and biological oceanography of the region. In summer, most of the HB complex is ice-free. As with other Arctic areas, the HB ecosystem is largely dependent on Arctic cod. Further, this region provides important habitat for a number of whale species (i.e. bowheads, narwals, belugas),

103

walrus, and polar bears, as well as providing migratory habitats for many other bird species

(Hoover et al., 2013).

Fisheries catch is relatively modest in this area and commercial harvest is approximately equal to that of subsistence catch (Pauly & Zeller, 2015). Catch was primarily subsistence based in the past, but has been split equally between commercial and subsistence since the early 2000s.

Fisheries catch composition has historically been dominated by Arctic char (primarily for subsistence), and in recent years the demand for northern prawn has increased commercial catches to equal that of Arctic char. The coastal population in the HB was ~44,000 in 2010 (IOC-

UNESCO & UNEP, 2016).

5.2.1.4 Canada high-Arctic and North Greenland – LME #66

Canada high-Arctic and North Greenland LME—hereon referred to as the High Arctic

Archipelago LME (HAA)—is the newest identified LME to reflect Canada’s Marine

Biogeographic Regions and changes to the Canadian Arctic Archipelago boundaries as a distinct region from Canada’s Eastern and Western Arctic. The HAA covers ~691,000 km2 of sea and is largely influenced by multi-year pack ice (PAME, 2013). Summer conditions expose the waters in the east and southwest portions of the LME. Primary production is low due to sea ice cover, and there is little ecological connectivity with adjacent areas of the Northwest Passage and

Baffin Bay (PAME, 2013). Circumpolar distributions of Arctic cod can also be found in this region. Other forage fish species found in neighbouring regions (e.g. capelin, Pacific herring) likely have distributions that overlap the HAA, although there is little sampling effort to confirm more accurate estimates of abundance. Some high Arctic bird species use this area for breeding

104

grounds. Whales and seals often avoid moving into this region due to heavy and variable sea ice, but may visit from neighbouring areas in favourable conditions.

Fisheries in this region are very small, and reconstructed from reports showing the majority of landings are for commercial purposes (Zeller et al., 2011; Pauly & Zeller, 2015). The short fishing season, distance from major ports, and dangerous conditions with heavy multi-year sea ice likely limit fisheries in this region. There are a few commercially valuable species in the

HAA, including arctic char, capelin, and pacific herring (PAME, 2013; Pauly & Zeller, 2015).

However, the potential of this area for fisheries is largely unknown. The LME has an estimated human population of <300 people as of 2010, the lowest population of the four LMEs considered here (IOC-UNESCO & UNEP, 2016).

5.2.2 Taking stock of Canada’s Arctic fisheries

Data from SAU (www.seaaroundus.org) were used to evaluate the current commercial and subsistence fisheries catch across the four Arctic LMEs (Zeller et al., 2011; Pauly & Zeller,

2015). Catch estimates within each LME include catches from all contributing fishing countries.

While country-specific catch is available for each LME, catch was aggregated to compare with projections of catch potential (described below). Landed values (LV) for total catches were calculated using global ex-vessel prices (Sumaila et al., 2007; Tai et al., 2017):

�� = � ∗ � (5.1) where H is the landings in tonnes and p is the ex-vessel price (per tonne). Catch amounts were broken down by various sectors, including small- and large-scale commercial, subsistence and recreational. Recreational amounts were negligible, while subsistence catch varied across the 105

region. Ex-vessel prices were applied to subsistence catches to provide a hypothetical value for those fisheries. Current landings and landed values were averaged for the most recent time period available from SAU, 2005-2014 (Pauly & Zeller, 2015).

5.2.3 Modelled species distribution and abundance

Current species distribution and abundance were simulated using a DBEM on a 0.5˚ longitudinal by 0.5˚ latitudinal grid (Cheung et al., 2008a) (see Supplementary material for further details on the DBEM). Species distributions from the SAU database (Jones & Cheung, 2015) were used as initial baseline conditions. Habitat suitability and environmental preference was characterized by overlaying variables such as temperature, salinity, dissolved oxygen concentration and depth, and were resolved vertically by the sea surface (for pelagic species) and bottom (for demersal species) layers. The baseline of initial habitat suitability was assumed to be the environmental conditions between 1951 and 1970. The model assumes that each occupied grid cell is at carrying capacity at this initial baseline time period, and carrying capacity will change as environmental conditions change (Cheung et al., 2008a). Outputs of environmental variables from ESMs were then used to project future scenarios (Bopp et al., 2013).

Species-specific preferences with habitat characteristics (e.g. sea ice cover, substrate) were also included. The effects of climate change (i.e. ocean warming, reduced dissolved oxygen concentration, ocean acidification) on body size, growth, and survival were simulated as physiological constraints using changes in aerobic scope (Tai et al., 2018). In the model, only invertebrate species were affected by ocean acidification because current literature has largely suggested that the effects of expected level of ocean acidification on growth and mortality of

106

finfish species are negligible or highly variable (Kroeker et al., 2013). Ocean acidification was modelled to affect aerobic scope and survival (Tai et al., 2018).

Population growth was modelled using a logistic growth function, determined by the intrinsic population growth rate and restricted by carrying capacity. Larval and adult stages disperse via advection-diffusion models (Cheung et al., 2008a), where adult emigration rates are greater if habitats within neighbouring cells are more favourable, while immigration is greater if habitat within the present cell is more preferable and abundance is below carrying capacity. Simulations were run for 72 polar and sub-polar species that are currently exploited: 60 finfish species and 12 invertebrate species (Appendix Table D.1). Metadata for each species were taken from FishBase and SeaLifeBase (Froese & Pauly, 2017; Palomares & Pauly, 2017). Of the 72 species, 64 were in Canada’s Eastern Arctic and West Greenland LME, 15 were in the Beaufort Sea, 57 were in

Hudson Bay, and 10 were in the High Arctic Archipelago.

5.2.4 Fisheries catch, prices and landed value potential

Fisheries catch potential across the four Arctic LMEs was calculated as the annual maximum sustainable yield of a species based on outputs of changes in abundance and carrying capacity from the DBEM, trophic level, geographic range, and primary production across its range

(Cheung et al., 2008b). Maximum sustainable yield (MSY) is the highest long-term equilibrium catch that will provide continuous returns, and in a simple Gordon-Schaefer model,

�� = !!∙! (5.2) !

where B∞ is the biomass at carrying capacity and r is the intrinsic rate of population increase.

107

Therefore, harvest rate (F) at MSY is

� = ! (5.3) !"# !

MSY in year t of a species is determined by the amount of energy (primary production) available to the population across its exploited range (P) and the area of its geographic range (A). Thus in the model, changes in primary production will lead to proportional changes in carrying capacity and ultimately the biomass that can be sustained in each spatial cell and therefore the MSY

(equation 5.1). Catch potential was not estimated for each country to avoid assumptions of catch allocations and transboundary fisheries management strategies between countries, as well as specific distributions of each species within each LME.

I used 2010 constant ex-vessel prices to estimate the current and future landed value potential of fisheries in each LME (Tai et al., 2017). Keeping real prices constant allows for direct comparison to current fisheries based on differences in potential catch amounts. Average annual catch and landed value potential for each LME for the period between 2005 and 2014 were calculated, and this represents the current sustainable potential of Arctic fisheries. Average annual numbers for the period from 2091 and 2100 was used to represent the future potential in the Arctic. The latter was projected to derive possible future scenarios of climate change and the value of fisheries. Data were aggregated by LME as they represent geographic regions with unique characteristics that are relatively isolated in terms of physical and ecological overlap.

108

5.2.5 Multi-model scenarios and simulations: characterizing uncertainty

Two primary sources of uncertainty were assessed in this analysis: model and scenario uncertainty. To characterize model uncertainty, environmental conditions were derived from outputs from three ESMs: NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL-ESM),

Max Planck Institute for Meteorology (MPI-ESM), and Institute Pierre Simon Laplace Climate

Modelling Centre (IPSL-ESM) (Bopp et al., 2013). Representative Concentration Scenarios

(RCPs) (van Vuuren et al., 2011) provide greenhouse gas concentration trajectories derived to reflect possible combinations of various socioeconomic assumptions. Two scenarios were used to characterize scenario uncertainty, RCP 2.6 and RCP 8.5, where the numbers represent radiative forcing values by the year 2100 based on greenhouse gas concentrations. RCP 2.6 is the low climate change scenario and assumes immediate mitigation of greenhouse gas (GHG) emissions where annual greenhouse gas emissions peak next decade (year 2025) but is reduced considerably, while RCP 8.5 is the high climate change scenario and our current trajectory if

GHG emissions continue to increase and no mitigation action is implemented. These ESMs were chosen due to the availability of sea surface and bottom data, the full range of environmental data required by the DBEM (i.e. sea temperature, dissolved oxygen, primary production, pH, current advection, salinity, sea ice extent), and data for both RCP scenarios (Cheung et al.,

2016a).

109

5.3 RESULTS

5.3.1 Current reported fisheries catch

Average annual catch between 2005 and 2014, the assumed current catch, was ~189,000 tonnes across all four Arctic LMEs (Table 5.1) while the average annual landed value for the same period was ~$560 million USD. Primary fisheries species in catch by weight were Northern prawn, Greenland halibut, and Atlantic cod, and were largely obtained from the CEA-WG LME.

Catch from CEA-WG comprised over 99% of total catch across the four LMEs in tonnage and landed value.

The proportion of catch for subsistence purposes varied by region and the BS had the highest percentage (80%) while the CEA-WG had the lowest (0.5%) (Table 5.1). However, subsistence catch in the CEA-WG was still greater than the total catch in the BS, with reports of annual catch at 922 and 264 tonnes, respectively. Subsistence catch is also significant in the HB at 717 tonnes but was negligible in the HAA (<0.1 tonnes).

Catch was reported for 88 different taxa across the Arctic. While many of these taxa names were reported at the species level, some were reported at higher taxonomic classifications (e.g. clams, sculpins). Therefore, it is likely that more than 88 species are exploited across these regions. The

CEA-WG recorded the greatest number of taxa in the last 10 years with 81, and the HAA recorded the fewest at 19 different taxa (Table 5.1).

110

Table 5.1. Average reported historical marine fisheries total catch and landed value for years 2005-2014. Data from www.seaaroundus.org (Pauly & Zeller, 2015). Landed value Large marine # of Catch (thousand Subsistence Primary fisheries ecosystem taxa† (tonnes) USD)†† catch (%)* species by tonnage Canada Northern prawn, Eastern Arctic 81 186,892 555,745 0.5 Greenland halibut, and West Atlantic cod Greenland

Beaufort Sea 21 330 1,060 80 Arctic char

Arctic char, Hudson Bay 53 1,299 3,827 55 Northern prawn

High Arctic 19 4 10 0.7 - Archipelago

Northern prawn, Total 88 188,527 560,643 1.1 Greenland halibut, Atlantic cod

†Number of unique taxa labels reported from the data. Note that some taxa names represent broad taxonomic groups (e.g. Family) or ‘species not identified’, and thus estimates of species numbers are conservative and likely comprise more than what is reported.

*Proportion of total catch for subsistence purposes.

††Landed values of subsistence catch are estimated using market ex-vessel prices from commercial catches (Tai et al., 2017).

5.3.2 Current and future catch potential

DBEM projections estimate current (2005-2014) maximum catch potential to be 760,000 tonnes and valued at $779 million USD annually across all Arctic regions (Table 5.2; Appendix Figure

D.1), 4 and 1.4 times greater than the current reported catch and landed value, respectively

(Table 5.1). This difference can be attributed to a greater proportion of lower value species (e.g. 111

capelin) in the projected catch potential from the DBEM. For estimates in future scenarios of climate change, models projected annual catch and landed value potentially to marginally increase under low climate change (RCP 2.6) to 833,000 tonnes and $859 million USD, respectively by the end of the century (2091-2100 average) (Table 5.2). Under high climate change (RCP 8.5), catch and landed value potential were projected to significantly increase to

1,278,000 tonnes and $1.3 billion USD, respectively, by century end.

Table 5.2. Model projections of annual catch and landed value potential for current and two future climate change scenarios (low and high) across all four arctic large marine ecosystems.† 2005-2014†† 2091-2100 RCP 2.6 RCP 8.5 Catch (thousand 760 833 1,278 tonnes) (632 – 875) (668 – 1,009) (1,152 – 1,472) Landed value 779 859 1,345 (million USD) (551 – 893) (572 – 1,069) (1,139 – 1,492)

†Values presented are multi-model means from the different earth system models used. Minimum and maximum values from model outputs are presented below in parentheses.

††Projected catches for the 2005-2014 period was averaged between RCP 2.6 and 8.5 scenarios.

Current projected catch potential in the CEA-WG was 250,000 tonnes annually, 34% larger than current reported catches (Figure 5.2a; Table 5.1). Landed value potential in this area was valued at $505 million USD, slightly less than current landed values of $556 million USD (Figure 5.2b;

Table 5.1). This difference is likely due to slightly more conservative estimates of catch potential of Northern prawn compared to reported catch, as well as the number of higher-priced species modelled for this region in the DBEM compared to the number reported in catch statistics (62 and 81 total species, respectively). Under low climate change, there was minimal change in catch

112

and landed value potential by year 2100, but under high climate change it was projected to increase to 572,000 tonnes and $892 million USD annually. Increased catch potential in this region were largely driven by increases in capelin, Atlantic cod, and Northern prawn, where catch potential for each species increased by 330%, 160% and 59%, respectively.

Catch potential for 2005-2014 in the BS was 9,050 tonnes and worth $4.6 million USD annually

(Figure 5.2c and 5.2d), 27 times greater than current reported catch and 4.3 times greater than current reported landed value (Table 5.1). Estimates of increased potential were largely due to capelin and Arctic cod—two species with low value (<$700 USD·tonne-1) that explains the disproportionate increases between catch and landed value. Future annual catch and landed value potential in the BS (in 2100) under high climate change was projected to be 50,000 tonnes and

$15.6 million USD respectively. These increases are due to a large influx of capelin into this region, increasing by 10 times in abundance and increasing catch potential by 42,000 tonnes,

83% of the total catch potential in this LME. Conversely, the low climate change scenario did not show significant changes into the future.

In the HB, current catch potential was projected to be 502,000 tonnes and valued at $268 million

USD annually (Figure 5.2e and 5.2f). This is estimated to be 386 and 70 times larger than the current reported catch and landed value, respectively (Table 5.1). A significant proportion of catch potential came from capelin, estimated to be 429,000 tonnes, while much of the remaining catch potential came from blue mussel and Arctic cod at 26,000 and 22,000 tonnes, respectively.

With high climate change, annual catch was 649,000 tonnes and landed value was $433 million

USD in 2100 (Figure 5.2e and 5.2f) due to increases in capelin, Northern prawn, and blue mussel. Again, the low climate change scenario resulted in little change.

113

In the HAA, current catch potential was 2,140 tonnes and landed value was at $1.6 million USD

(Figure 5.2g and 5.2h)—much greater than minimal numbers of current reported catch and landed value (Table 5.1). Arctic cod and Northern prawn make up over 99% of the projection catch potential (87.5% and 12.3%, respectively). High climate change scenario projections showed future annual catch of 6,700 tonnes and landed value of $3.8 million USD by year 2100

(Figure 5.2g and 5.2h) due to increases in capelin, Northern prawn, and Arctic cod. Low climate change scenario showed negligible change by year 2100.

114

a b 1000 600 800 Canada Eastern Arctic - West 400 600 Greenland

400 200

60 c d

15 40 Beaufort Sea 10

20 5

RCP 2.6 RCP 8.5 alue (million USD)

700 e v f Catch (thousand tonnes)

Landed 400 600 Hudson Bay

500 300

400 200

g h 10.0 4

7.5 3 High Arctic Archipelago 5.0 2 2.5

1 2000 2025 2050 2075 2100 2000 2025 2050 2075 2100 Year

Figure 5.2. Projected maximum sustainable catch and landed value potential in each of Canada’s Arctic Large Marine Ecosystems. Thin lines represent each model simulation using the different earth systems models, while bold lines are multi models means. Blue lines are projections of the low climate change scenario (RCP 2.6) and red lines are the high climate change scenario (RCP 8.5). Data are smoothed using a 10-year running mean.

Table 5.3 lists the top five species with the greatest projected annual catch and landed value by the end of the century under scenario RCP 8.5 across Canada’s Arctic (all four LMEs

115

aggregated). Capelin was projected to have the greatest current catch potential of 488,000 tonnes annually, about 72% of the total catch potential for the 72 species analyzed here (Table 5.2 and

5.3). Current catch potential for the top 5 species was 698,000 tonnes, over 90% of the total catch potential. Current landed value potential was highest for Northern prawn at $302 million

USD annually due to its relatively high value. The top five species in current landed value potential comprised 85% of the total landed value potential (Table 5.2 and 5.3).

Future projections of annual catch potential under scenario RCP 2.6 showed changes between

0% (blue mussel and Arctic cod) and 27% (Atlantic cod) for the top five species by 2100 (Table

5.3). Much greater changes were projected with high climate change: annual catch potential increased by 197% for Atlantic cod but decreased for Arctic cod by 26%. Arctic cod are a circumpolar species and thus have reached their northern range; decreases in abundance and catch potential are a result of an environmentally driven southern range contraction. Catch potential for the remaining top 5 species increased by ~50-70% with high climate change.

Similar changes were projected for landed value potential for both scenarios as prices remained constant. By year 2100, landed value potential from these five species was projected to be worth up to $831 million USD annually under RCP 8.5.

116

Table 5.3. Top 5 species for annual projected catch and landed value for years 2005-2014 and 2091-2100 for two representative concentration pathway (RCP) scenarios across all four LMEs.† 2005-2014†† 2091-2100 Common Name RCP 2.6 RCP 8.5 488 530 807 Capelin (434 – 570) (461 – 639) (707 – 961)

104 118 174 Northern prawn (68 – 122) (70 – 156) (150 – 193) 35 44 103 Atlantic cod (17 – 48) (17 – 61) (67 – 124) 36 36 54 Blue mussel (27 – 41) (29 – 40) (50 – 58) 36 36 26 Arctic cod (29 – 42) (30 – 45) (20 – 38) Catch (thousand tonnes) 698 764 1,165 Total (588 – 808) (624 – 925) (1,064 – 1,348) 302 346 510 Northern prawn (200 – 356) (206 – 455) (438 – 564) 131 141 216 Capelin (116 – 152) (123 – 170) (188 – 257) 72 92 214 Atlantic cod (36 – 99) (36 – 128) (140 – 258) 130 131 198 Blue mussel (98 – 147) (106 – 145) (182 – 212) Greenland 33 35 49 halibut (17 – 45) (15 – 47) (29 – 60)

Landed value (million USD)Landed (million value 668 745 1,186 Total (466 – 775) (486 – 937) (1,009 – 1,319)

†Values presented are multi-model means from the different earth system models used. Minimum and maximum values from the different earth system model outputs are presented below in parentheses.

††Projected catches for the 2005-2014 period were averaged between RCP 2.6 and 8.5 scenarios.

117

5.4 DISCUSSION

Current catches for each LME region presented here represent one set of estimates using data and reconstruction methods by SAU, and should be compared to other reports to determine accuracy. There are other sources of catch statistics for Arctic marine fisheries, but they contain catch numbers for Canada only or are region specific (Inuvialuit Harvest Study: Data and methods report 1988-1997, 2003; Priest & Usher, 2004; Watts et al., 2017). Furthermore, there are very few recent reports or articles on catch statistics in Canada’s Arctic. While comparisons between these reports are difficult due to the differences in temporal and spatial scale, it underscores the apparent gap in reliable and consistent data reports of Canada’s Arctic fisheries.

Projections of species distribution and abundance show significant potential for marine capture fisheries for current and both future scenarios of climate change across all four LMEs in

Canada’s Arctic, especially in the CEA-WG and the HB LMEs (Figure 5.2). The large difference between current reported catch and modelled estimates of current catch potential suggests that there is room for fisheries expansion in Canada’s Arctic, even in regions that already have substantial commercial catches (e.g. Baffin Bay in the CEA-WG). This is likely the reality of

Arctic ecosystems, where the fishing seasons are short and limited by dangerous conditions.

Increased catch potential in future scenarios of climate change are primarily driven by ocean warming as species shift poleward to cooler waters. General trends are consistent across the various model uncertainties (e.g. scenarios uncertainty), but simulations are uncertain in the precision of the estimates and exact projection values must be interpreted with caution. The combination of increased catch and accessibility has major implications for Arctic ecosystems,

Canada’s Arctic coastal populations, and potential industrial fisheries development.

118

Current subsistence catch of marine fisheries resources represents only 1% of the total catch across the four Arctic LMEs, but much higher in regions with few or non-existent commercial fisheries (e.g. Beaufort Sea at 80%) (Table 5.1). However, food derived from marine fish and invertebrate species comprise only a small proportion of marine food. For example, in Inuit communities the majority of ocean-derived food nutrition come from marine mammals (e.g. seal, narwal, beluga), estimated at between 72% and 80% across each of Canada’s Arctic LMEs

(Watts et al., 2017). These harvested marine mammals are largely dependent on lower trophic level species, especially on highly abundant species such as Arctic cod (Cui et al., 2012;

Whitehouse et al., 2014). Thus, changes in the abundance and distribution of these prey species could drastically alter the population dynamics of these marine mammals. In addition to impacts on subsistence catches due to ecological changes, climate change has already affected hunting patterns due to physical changes in sea ice (Ford et al., 2008). This includes both later formation and earlier break up of sea ice, and higher mobility and less stability of sea ice, ultimately increasing dangers of sea ice travel and hunting from the ice edge. If marine mammal catches decline due to direct and indirect effects of climate change, increased fisheries catch potential could serve to mitigate impacts and partially alleviate food insecurities. Thus, the allocation of fisheries resources (i.e. for commercial and subsistence purposes) in a changing and uncertain future is a critical discussion point for decision makers.

Climate change driven decreases in sea ice coupled with increases in catch potential has increased interest in the development of fisheries throughout the Arctic (Van Pelt et al., 2017).

Aside from Canada’s Eastern Arctic regions (i.e. Baffin Bay and Davis Strait) and Hudson Bay, commercial fisheries are either minimal or non-existent in the Central and Western Arctic.

119

Recent evaluations suggest that commercial fisheries in these areas might be feasible (Van Pelt et al., 2017), however, this may depend on region and species. Costs of travel to the remote areas such as the high Arctic may outweigh revenues and restrict any fisheries; but this has not prevented areas such as the high seas from being exploited (Sala et al., 2018). Certain species may also be more economically feasible. Capelin, the species with the greatest catch potential in a high climate change future (Table 5.3), has been primarily caught in CEA-WG LME and is projected to increase in this region and the adjacent waters of the HB.

For now, to ensure comprehensive knowledge of Arctic marine ecosystems and its ability to sustain exploitation, commercial fishing bans across these regions are currently in place. This includes fishing bans by the US and Canada in Alaskan waters north of the Bering Strait (i.e.

Beaufort and Chukchi Seas) and the adjacent waters of Canadian Beaufort Sea, respectively

(“North Pacific Fisheries Management Council, Arctic Fishery Management Plan for Fish

Resources of the Arctic Management Area”; Ayles et al., 2016). These bans prevent any new commercial fishing licenses from being issued, and aim to protect the Beaufort Sea until scientists and decision makers develop the knowledge, understanding, and a comprehensive management plan of whether current and future fish productivity can sustain commercial fisheries without negative impacts to food webs and the traditional lifestyle of the Inuvialuit.

Furthermore, recent research has identified significant areas in the Beaufort Sea and Mackenzie

River Delta area, and has led to the development of two marine protected areas in 2010 and

2016: Tarium Niryutait and Anguniaquiva Niqiqyuam marine protected area, respectively

(AMAP, 2018). Recently in October of 2018 following an agreement in December of 2017, an international agreement was signed by all five countries bordering the Arctic as well as other

120

fishing nations, including, China, Japan, South Korea and the EU to ban any commercial fisheries operations in the high Arctic for 16 years (Weber, 2018). This protects much of

Canada’s West Arctic region for the near future as more research is done on the sustainability and feasibility of commercial fisheries.

Results of this paper contribute towards better understanding Arctic marine species and their response to environmental change, outlines implications for Arctic marine resources, and discusses the underlying social and cultural considerations for advancing fisheries development.

These results can be used as input data for further analyses of the downstream ecological, economic, social, and cultural impacts of fisheries. While these results are informative, model projections such as the ones presented here are not free of uncertainties and limitations. For example, projections of future biogeochemical cycling, primary production, and interactions of sea ice dynamics are still highly variable among models (Vancoppenolle et al., 2013).

Furthermore, climate change impacts on species distribution and abundance may be affected by ecological interactions, including invasions and extinctions, which can drastically alter community structure and ecosystem dynamics (Doney et al., 2012). Community and ecosystem resistance to invasions may also restrict shifts in species distribution. The DBEM simulates biogeographical changes for single species, and therefore does not account for ecosystem interactions, yet ecosystem responses to climate change remain highly uncertain for the Arctic

(Michel et al., 2012). For these reasons, the Arctic may not be readily habitable for sub-arctic species and thus may not be able to support exploitation in these regions.

The effects of ocean acidification remain uncertain and may affect species more drastically than what is currently understood, especially in the Arctic where pH is projected to decrease

121

substantially (Denman et al., 2011; Steiner et al., 2015). Generally, current knowledge supports that invertebrate calcifying species show greater sensitivity to ocean acidification than finfish

(Kroeker et al., 2013), and changes in highly abundant key ecosystem species—such as the planktonic mollusc, Limacina helicina—can have major bottom up effects on higher trophic levels (Welch et al., 1992; Karnovsky et al., 2008). Nonetheless, past modelling results testing different sources of uncertainty found that projections were least sensitive to ocean acidification effect size (Tai et al., 2018). Between the two sources of uncertainty analyzed herein, model uncertainty accounted for almost all of the uncertainty at the start of the simulation (2004-2015) but this shifted to scenario uncertainty near the end of the simulation (2091-2100), suggesting trends were consistent across earth system models when scenarios diverged (Appendix Figure

D.2).

5.5 CONCLUSION

These models project the sustainable catch potential of fisheries in Canada’s Arctic LMEs to be much greater than current reported catches; yet further investigation should evaluate the economic feasibility. Future fisheries catch will potentially be much greater under the high climate change scenario (RCP 8.5) and moderate if carbon emissions are curbed and climate change is moderate (RCP 2.6). However, my results must be interpreted with caution. Arctic ecosystems may not be ecologically suitable as a refuge for sub-polar species or resilient to increased exploitation. Proper measures must be taken to ensure the sustainability of these marine resources and it is essential to consider the ecological, social, cultural, and economic impacts to some of the only largely unexploited and pristine marine regions left in the world.

122

Chapter 6: Conclusion – thesis contributions and implications for current and future research

This research addresses aspects across the broad interdisciplinary field of fisheries and how future global change will disrupt the flow of resources between natural marine environments and human society. It contributes global databases, modelling approaches, scenario development, and assessments of global change impacts. Interdisciplinary approaches are critical for research on impacts of climate change occurring at a rapid pace (IPCC, 2014). Building end-to-end frameworks to bridge natural and social sciences is essential to developing climate change adaptation and mitigation strategies (Mastrandrea et al., 2010).

The goal of this thesis was to develop the tools and knowledge to conduct analyses of global change impacts—with a strong focus on ocean acidification in a multi-stressor context—on biological systems and how it scales up to fisheries level impacts. My first main research objective was to update a database to quantify global fisheries prices and landed values that matches the reconstructed SAU catch database (Chapter 2). This database was constructed to contribute to ongoing research efforts that support many past, current, and future research projects. For my thesis, it was a natural progression to use this database for the economic analyses in Chapter 5. There are other studies that have already benefitted from this database, some of which I have been involved in (e.g. AMAP, 2018; Sumaila et al., 2019).

The second main objective addressed in my thesis was to identify the biological and socioeconomic impacts of OA on future fisheries. Chapter 3 to 5 used the DBEM to explore effects of OA on abundance, catch potential, and landed values. Chapter 3 provides an important

123

exercise in integrated modelling approaches for how physiological responses scale to population dynamics and distribution. This chapter provided guidance for the assumptions for the mechanistic modelling in subsequent Chapters 4 and 5. Chapters 3 to 5 are exercises in scenario development and building projections for global change impacts on fisheries. They each provide different aspects of uncertainty, while highlighting major assumptions and limitations. Chapter 3 specifically addresses the uncertainty of OA effects and modelling its effects on various life history traits. Chapter 4 focuses on the climate change trajectories and underscores the severe consequences of continuing on society’s current high climate change trajectory. In this modelling output, areas with greater OA will likely have greater and less predictable impacts. Similarly,

Chapter 5 focuses on the various climate change scenario trajectories in a specific region,

Canada’s Arctic, and includes projections of economic potential in a region that is still relatively pristine and of high interest due to its potential.

The third main objective of my thesis was to explore the different indicators that can be used at various spatial scales. In Chapter 3, I used specific species found in Canada’s water to display the variance in spatial changes to abundance in response to OA impacts. Indicators used here included changes in life history traits and how that scaled up to abundance. These species represent economically important but biologically diverse invertebrate species. In Chapter 4, I scaled up the model to explore the OA effects on global invertebrate species. Similarly, I used changes in life history traits and how they scaled up to changes in abundance and also catch potential. I aggregated changes in catch potential by spatial entities of large marine ecosystems to show the aggregated impacts at a global scale. Furthermore, in Chapter 4 I used latitudinal shift and changes in range size as indicators for species that are most sensitive to global change.

124

Chapter 5 is a regional analysis for Canada’s Arctic that looks at downstream economic responses to future global change. Potential fisheries expansion here must proceed with caution due to the relatively pristine nature and the importance of Arctic marine ecosystems to local Inuit communities. I used modelling projections to assess the current and future potential biomass and value of sustainable fisheries. While these indicators suggest the potential for resource exploitation is high, the resilience of Arctic ecosystems is still highly uncertain as these areas may face some of the greatest changes in in environmental conditions and human exploitation in the next few decades.

Price database (Ch. 2) Modelling OA impacts (Ch. 3) Scenario development (Ch. 3-5) • Global ex-vessel fsh prices (1950-2010) • Mechanisms of OA impacts on physiology • Disaggregated by fsheries end-products • Testing scenarios of OA mechanism • Scaling impacts to population dynamics (i.e. human consumption, fshmeal & fsh oil, other) • Future climate change impact scenarios • Defning assumptions, uncertainties, and • Landed value estimates • OA impacts with multiple climate-related stressors sensitivities of different mechanisms • Fisheries economic analyses

Tools developed Application Canada's Arctic fsheries (Ch. 5) Global OA impacts (Ch. 4) • Current and future potential catch and value of fsheries Future research • OA impacts on invertebrate fsheries • Increased catch as species shift north directions • Maximum catch potential • Interest in commercial operations In progress • Identify species and regions at greatest risk • Uncertainty of the resilience of the ecosystem Recommended

Lobster and shrimp fsheries Dynamic bionomic model Risk for Arctic communities OA impacts modelling • Modelling OA impacts • Risks for fsheries to OA and climate change • Integrate price, cost, feet size • Impacts to various life stages • Alternative mechanisms • Ecosystem and community resilience dynamics • Interactions with different fshing restrictions (i.e. metabolic theory) • Resource allocation • Management scenarios (i.e. size limit, harvest rate) • Ensemble of approaches • Integrated assessment model

Figure 6.1. Thesis contributions and research directions.

125

6.1 RESEARCH CONTRIBUTIONS

My thesis contributes a fish price database, methods on modelling OA impacts, scenarios of future global change impacts on fisheries, and assessments of global change impacts, important to current and future directions for interdisciplinary fisheries research. In the following sections,

I will describe the contributions and some of the major developments from my research, as well as potential future research directions.

6.1.1 Fisheries databases

My contributions of the ex-vessel fish price database (Chapter 2) provides a foundation for large- scale global and regional fisheries economic analyses, and potentially even applied research at more local scales. There have been three version of the database, with each update providing more detailed information to match the records from the global SAU catches (Sumaila et al.,

2007; Swartz et al., 2013; Tai et al., 2017).

This price database is part of a set of fisheries economics databases that correspond to the SAU catch database. There are ongoing efforts that have produced various databases including a fishing cost database (Lam et al., 2011) and a fishing gear database (Cashion et al., 2018), which report the cost of fishing (per tonne) and the type of gear used (e.g. purse seine, trawl), respectively, for each record in the SAU catch database. These databases can be combined for analyses of historical fisheries trends and then used for parameterizing projection scenarios (Lam et al., 2016). These databases play an important role for the scenario development and modelling in my thesis, i.e. Chapters 3 to 5. The global ex-vessel price database, along with the other global fisheries databases, are open source and provide valuable information to fisheries researchers,

126

fisheries managers, and policy makers around the world, especially for those where the institutional capacity for monitoring and data collection is limited (van Helden, 2012). Open source data and science is essential for research in climate change impacts where rapid development and dissemination of results is necessary to keep pace with climate change itself

(Tai & Robinson, 2018). This database has proven to be a valuable tool in recent publications

(e.g. Sala et al., 2018; Selig et al., 2018) and has provided me with opportunities for collaboration (e.g. Cisneros-Montemayor et al., 2018; Sumaila et al., 2019).

6.1.2 Ocean acidification modelling

In Chapter 3, I used a mechanistic approach to integrate a set of working hypotheses to model the impacts of OA and how it scales from physiology to fisheries. However, the effects of ocean acidification as a biophysical driver remain highly variable across species and there has yet to be a universal hypothesis to explain cause-and-effect relationships. Meta-analyses that have summarized the impacts of OA are extremely useful for parameterizing projection models to help develop a better understanding (e.g. Kroeker et al., 2010, 2013; Nagelkerken & Connell,

2015; Cattano et al., 2018). While the results from meta-analyses are more generalized, they provide an understanding of the potential impacts and sensitivity of the results as it scales up to higher levels of biological organization (e.g. organism, population, ecosystem). My research in

Chapter 3 explored possible mechanisms and—using the current working hypotheses and parameters—showed that OA impacts are generally small relative to the effects of warming.

I linked the effects of OA from physiological mechanisms to population dynamics using a set of hypotheses that have recently been challenged. The DBEM combines the OCLTT and the GOL hypotheses to model the effects of OA and other climate change stressors (Pörtner & Lannig, 127

2009; Pauly & Cheung, 2017). These effects are then combined with a growth model (von

Bertalanffy, 1957) to estimate how the effects of environmental change (i.e. temperature, pH, oxygen) scale up from aerobic scope to life history traits including growth rate, maximum body size, and natural mortality (Chapter 3). Natural mortality is then incorporated into the population dynamics model. While not all researchers or literature may support this particular mechanism, it is a viable working hypothesis and at the least it facilitates discussion for directions for advancing this scientific field.

The OCLTT hypothesis has been consistently challenged for its lack of general applicability to ectotherms and the growing number of studies that fail to support the hypothesis (Pörtner et al.,

2018a). For example, positive effects on aerobic scope in response to increasing acidity have been observed, opposite to what the OCLTT suggests (Kunz et al., 2018). Aerobic scope, the primary measure for the OCLTT, also does not provide a good proxy for the energy available for other life history processes such as growth and reproduction (Clark & Sandblom, 2013).

However, the OCLTT provides a working and usable hypothesis for “bridging ecology and physiology” and providing a mechanistic approach for projection models of climate related responses (Pörtner et al., 2017; Pörtner et al., 2018b). An alternative hypothesis, termed Multiple

Performance-Multiple Optima, provides a more complex extension of the OCLTT where different physiological processes have different thermal—and environmental—optima (Clark &

Sandblom, 2013), but there have been limited developments on resolving the mathematical formula required to utilize it in forecasting models (Falkenberg et al., 2018). Linking multiple hypotheses has been proposed to create a more integrated framework, advancing our understanding of global change impacts from physiology to ecology (Falkenberg et al., 2018).

128

While the ongoing debate is important to progress and advance our understanding of the links between ecology and physiology, current practices of using the OCLTT in projection modelling are sufficient and well received. Nonetheless, it is widely acknowledged that further discussion is required to facilitate future inter- and trans-disciplinary research (Falkenberg et al., 2018; Pörtner et al., 2018b).

6.1.3 Scenario development and identifying high impact areas

Scenario development is essential to forecast modelling of climate change impacts, and has been an important contribution for international organizations such as the IPCC. I used existing RCP scenarios (van Vuuren et al., 2011) to derive how the upper (RCP 8.5) and lower (RCP 2.6) climate forcings affect living marine resources. In Chapter 3, I explored various mechanisms and assumptions in modelling biological OA effects and how it translated to DBEM outputs of changes in population distribution and abundance across the different RCP scenarios.

Quantifying the uncertainty and relative magnitude of OA effects, in combination with other stressors, is essential to determine the biophysical drivers with the greatest impacts. Scenarios with and without the effects of OA allowed me to separate the effects of OA and determine the relationship between changes in pH and changes in physiology (aerobic scope), biology

(growth), and fisheries (maximum catch potential). Our current understanding suggests that temperature remains one of the biggest environmental drivers of change in marine ecosystems

(Chapter 4) (Cheung et al., 2016a; Tai et al., 2018).

In Chapter 4, I also used RCP scenarios to determine large-scale regional and global impacts of

OA with other stressors. In Chapter 5, scenarios of current and future fishing were used to present the potential of fishing in Canada’s Arctic. These scenarios provide estimates of the 129

value of fisheries in this region with varying levels of climate change, which can then be used to determine the economic feasibility of fishing operations in this region.

In Chapter 3 and 4, I identified species and regions with greatest responses to OA and climate change stressors. This includes changes in abundance, distribution shifts, and range size. The greatest changes in catch potential identify the fisheries and countries at greatest risk. The global analysis (Chapter 4) could potentially contribute to international assessments and policies (i.e.

IPCC reports), and recent reports suggest gaps in this field of OA impact projection modelling.

Effects of OA have been incorporated into various integrated modelling approaches, enhancing our understanding of OA’s end-to-end effects and advancing the landscape of tools available

(Olsen et al., 2018; Rheuban et al., 2018). These efforts to explore various models and assumptions are important for climate change adaptation while our understanding of bridging

OA effects on physiology to ecology remains uncertain.

6.2 CAVEATS AND LIMITATIONS

Limitations to the research in my thesis have been briefly outlined in each chapter. The database in Chapter 2 is a global construction of ex-vessel fish prices, and is best used for larger scale analyses that correspond to using the SAU catch database (Pauly & Zeller, 2015). This database makes assumptions about the relationship of prices between countries and applies global economic models to estimate these missing prices (Hill & Syed, 2010). These estimates do not account for within country or seasonal variation in ex-vessel prices, which can have major effects on the local—and sometimes global—economy (e.g. Atlantic lobster prices) (Pershing et al., 2018). An alternative to using the SAU catch is to use catch records from FAO. A different

130

ex-vessel fish price database was constructed from the FAO catch and landed value database

(Melnychuk et al., 2016), which would be better suited when using FAO’s database.

Modelling the effects of OA with other multiple stressors in Chapters 3 to 5 reflect one set of working hypotheses for the effects of environmental stressors on biological processes and how that scales up to higher levels of biological organization (i.e. population dynamics). A major assumption of the DBEM model used here is that aerobic scope is a suitable proxy for the energy available for growth and that aerobic scope responds to environmental stressors (Pauly &

Cheung, 2017; Pörtner et al., 2017). Therefore, it should be noted that the model and results are not derived from established theory, but instead derived from the most viable working hypothesis in current literature. Results from my research should be used in addition to other similar outputs with different assumptions and underlying mechanisms. Using an ensemble of models can provide a greater picture of the uncertainty in modelling (Bopp et al., 2013; Jones & Cheung,

2015; Rodgers et al., 2015; Cheung et al., 2016b).

While I explored various uncertainties with the DBEM, there were other factors that were not included or had limited information. Our knowledge on the scope for adaptation to global environmental change is limited (Chapter 3), but there are examples of ‘natural’ experiments

(e.g. volcanic CO2 seeps) that provide insight into adaptation to conditions similar to future projections (e.g. Harvey et al., 2016; Agostini et al., 2018). Experiments on evolutionary capacity can enhance the capacity of our modelling efforts to incorporate evolutionary responses

(Sunday et al., 2014). Furthermore, including mechanisms for ecological interactions would add more value to the spatial model, limiting species range shifts based on their capacity for invasion and trophic interactions. Modelling efforts could also be improved by adding dynamic

131

interactions of fishing and climate change to explore scenarios of various management strategies

(Rheuban et al., 2018).

6.3 NEXT STEPS

6.3.1 Continuing projects

My thesis has led to ongoing projects with co-authors and other researchers. I am currently working on a project to develop a model integrating results from experiments on OA effects

(Menu-Courey et al., 2019) on the physiology of American lobster (Homarus americanus) into the DBEM and using various climate and fishing scenarios to determine impacts on abundance, distribution, catch, and population demography. This project is part of a program to connect science to end-users, and in this case it involves communicating results to stakeholders in the fishery (i.e. fishers, managers, policymakers). One of the major developments I am working on is incorporating impacts of climate-related stressors and fishing on size distribution. Effects of fishing include changes in harvest rate and size limits. The Canadian lobster fishery has size limits based on estimates of reproductive maturity (Fisheries and Oceans Canada, 2012); size limits are region dependent where warmer regions (i.e. Northumberland Strait) have lower size limits due to quicker development and a smaller size-at-maturity (Campbell & Robinson, 2008).

With this addition to the DBEM, we can then apply the model to geographically distinct populations within a species, as well as generalize its application for other important species (i.e. northern prawn, Pandalus borealis).

My thesis research has also led to involvement with ongoing research in Canada’s Arctic regions. The Canadian government is concerned with the rapid change occurring across the

132

Arctic and they have highlighted major gaps in knowledge of climate-related impacts on ecosystems and marine resources. Collaborations to conduct an assessment of future scenarios of climate change and its impacts on marine resources, such as fish and marine mammals, are required for stakeholders and policymakers to mitigate and adapt to future changes. Current policies to halt any commercial harvesting have been put in place until a better understanding of exploitation effects on ecosystems and indigenous communities is developed (Weber, 2018). The work presented in Chapter 5 will be included in these research objectives and provide a foundation for further analyses of climate-related impacts.

6.3.2 Recommendations for future experimental research

It is widely recognized that collaboration and integration of experimental researchers and modellers is important for research on climate impacts on ecosystems and society, yet there are still relatively few examples (e.g. Queirós et al., 2015; Rheuban et al., 2018). There is still a major gap in translating OA effects on physiology to responses in life history traits and population dynamics (Pörtner et al., 2018b), thus I recommend future experiments focus on quantifying the relationship between physiological responses and population responses to global change. Specifically, multi-factorial experiments to incorporate and isolate the effects of multiple stressors (i.e. OA, warming, deoxygenation), and quantifying changes in life history traits, as well as population parameters in mesocosm settings. This experimental data can then be incorporated into modelling efforts and potentially advance our understanding of the mechanisms of multi-stressor responses in marine ectotherms. The scientific community needs more integration between these disciplines to co-design research approaches to address the complexities of bridging physiology and ecology (Boyd et al., 2018). Once a solid foundation of

133

literature has provided enough evidence to ascertain a ubiquitous or general mechanism of OA and multistressor responses, we can then challenge its applicability amongst other working hypotheses (i.e. OCLTT).

6.3.3 Recommendations for future modelling approaches

Future work involving the DBEM includes testing the sensitivity of model outputs to various models of the relationship between environmental drivers and population dynamics. The DBEM currently uses ecophysiological models (von Bertalanffy, 1957; Pörtner & Lannig, 2009; Pauly

& Cheung, 2017) and habitat suitability models (Cheung et al., 2008a) to simulate the effects of biophysical drivers on body size and population dynamics. This set of models could be replaced in the DBEM, or modified and combined with models of metabolic theory and population dynamics (Bernhardt et al., 2018). This alternative approach would use models that have been fit to explain the effects of temperature on body size, population growth rate, and carrying capacity.

Additional approaches can address this source of uncertainty and establish an ensemble of ecophysiological models in the DBEM. However, there are still major gaps in our understanding of multi-stressor effects (temperature, acidification, deoxygenation) and metabolic theory and how they scale to population dynamics.

The DBEM is effective to model climate-related effects on fisheries in its current state, yet it can be improved with dynamic socioeconomic components. A major advancement would be to integrate a more complex bionomic model to include the dynamics of fish price, cost of fishing, and fleet size (Walters & Martell, 2004). Furthermore, incorporating management measures such as harvest rates, quotas, and size limits, would improve our understanding of what efforts can mitigate climate change effects and ensure long-term sustainability of fisheries resources. 134

Converting the DBEM to be used in an Integrated Assessment Model would benefit stakeholders and policymakers by providing quantitative assessments of various management strategies with the dynamics of environmental and human drivers (Fisheries and Oceans Canada et al., 2014;

Cooley et al., 2015; Ayles et al., 2016; Rheuban et al., 2018).

6.4 FINAL REMARKS

Recent global environmental change is altering natural and social systems in unprecedented ways as atmospheric carbon dioxide is accumulating at unprecedented rates (IPCC, 2013). Global efforts are being made to address many of these impacts, such as the development of the 2015

UN Sustainable Development Goals to achieve global peace and prosperity for people and the planet (https://sustainabledevelopment.un.org/). Making efforts is imperative to limit global warming to 1.5˚ C to ensure a more sustainable and equitable society (IPCC, 2018). My research has direct implications for two of these objectives: Climate Action—to take urgent action to combat climate change and its impacts; and Life Below Water—to conserve and sustainably use the oceans, seas, and marine resources for sustainable development. In working towards these global challenges, collaboration will expedite research programs and facilitate rapid action for climate-related impacts, while communication of results to researchers, government, and public has never been more important to enhance adaptation strategies.

135

Bibliography

Agostini S, Harvey BP, Wada S, Kon K, Milazzo M, Inaba K, Hall-Spencer JM (2018) Ocean

acidification drives community shifts towards simplified non-calcified habitats in a

subtropical−temperate transition zone. Scientific Reports, 8, 11354.

Ainsworth CH, Samhouri JF, Busch DS, Cheung WWL, Dunne J, Okey TA (2011) Potential

impacts of climate change on Northeast Pacific marine foodwebs and fisheries. ICES

Journal of Marine Science, 68, 1217–1229.

Alder J, Pauly D (2006) On the multiple uses of forage fish: from ecosystem to markets.

Fisheries Centre Research Reports, 14, 109.

Alheit J, Niquen M (2004) Regime shifts in the Humboldt Current ecosystem. Progress in

Oceanography, 60, 201–222.

AMAP (2013) AMAP Assessment 2013: Arctic Ocean Acidification. Arctic Monitoring and

Assessment Programme (AMAP), Oslo, Norway, Oslo, Norway, viii + 99 p.

AMAP (2018) AMAP Assessment 2018: Arctic Ocean Acidification (ed Symon C). Arctic

Monitoring and Assessment Programme (AMAP), Tromsø, Norway, vi+187pp p.

Anthony KR, Kleypas JA, Gattuso J-P (2011) Coral reefs modify their seawater carbon

chemistry – implications for impacts of ocean acidification. Global Change Biology, 17,

3655–3666.

Aranda M (2009) Developments on fisheries management in Peru: The new individual vessel

quota system for the anchoveta fishery. Fisheries Research, 96, 308–312.

Asche F, Bjorndal T (1999) Demand elasticities for fish and seafood: A review. Unpublished.

Centre for Fisheries Economics, Norwegian School of Economics and Business

Administration. Bergen, Norway, 94 pp. 136

Asche F, Tveterås S (2004) On the Relationship Between Aquaculture and Reduction Fisheries.

Journal of Agricultural Economics, 55, 245–265.

Asche F, Oglend A, Tveteras S (2013) Regime Shifts in the Fish Meal ⁄ Soybean Meal Price

Ratio. Journal of Agricultural Economics, 64, 97–111.

Atkinson D (1994) Advances in ecological research, volume 25, 1st edn (eds Begon M, Fitter

AH). Academic Press, 1–58 pp.

Ayles B, Porta L, Clarke RM (2016) Development of an integrated fisheries co-management

framework for new and emerging commercial fisheries in the Canadian Beaufort Sea.

Marine Policy, 72, 246–254.

Bakun A (1990) Global Climate Change and Intensification of Coastal Ocean Upwelling.

Science, 247, 198–201.

Barange M, Merino G, Blanchard JL et al. (2014) Impacts of climate change on marine

ecosystem production in societies dependent on fisheries. Nature Climate Change, 4, 211–

216.

Barsky R, Kilian L (2002) Oil and the Macroeconomy Since the 1970s. Journal of Economic

Perspectives, 18, 115–134.

Barton A, Waldbusser GG, Feely RA et al. (2015) Impacts of coastal acidification on the Pacific

Northwest shellfish industry and adaptation strategies implemented in response.

Oceanography, 28, 146–159.

Bechmann RK, Taban IC, Westerlund S et al. (2011) Effects of Ocean Acidification on Early

Life Stages of Shrimp (Pandalus borealis) and Mussel (Mytilus edulis). Journal of

Toxicology and Environmental Health-Part a-Current Issues, 74, 424–438.

Bell FW, Nash DA, Carlson EW, Waugh F V, Kinoshita RK, Fullenbaum RF (1970) The future 137

of the world’s fishery resources: forecasts of demand, supply, and prices to the year 2000

with a discussion of implications for public policy. National Marine Fisheries Service

Economic Research Laboratory. Washington, D.C., USA., 338 pp.

Berkes F (1990) Native subsistence fisheries: a synthesis of harvest studies in Canada. Arctic, 43,

35–42.

Bernhardt JR, Sunday JM, O’Connor MI (2018) Metabolic Theory and the Temperature-Size

Rule Explain the Temperature Dependence of Population Carrying Capacity. The American

Naturalist, 192, 687–697. von Bertalanffy L (1951) Theoretische Biologie - Zweiter Band: Stoffwechsel, Wachstum. A.

Francke, Bern. 418 pp. von Bertalanffy L (1957) Quantitative laws in metabolism and growth. The Quarterly Review of

Biology, 32.

Bertrand A, Ballón M, Chaigneau A (2010) Acoustic Observation of Living Organisms Reveals

the Upper Limit of the Oxygen Minimum Zone. PLoS ONE, 5, e10330.

Beverton RJH, Holt SJ (1959) A review of the life span and mortality rate of fish in nature and

the relation to growth and other physiological characteristics. In: CIBA Foundation

Colloquia on Ageing: the Lifespan of Animals (eds Wolstenholme GEW, O’Connor M), pp.

1422–177. J. & A., Churchill, London.

Bindoff NL, Stott PA, AchutaRao KM et al. (2013) Detection and Attribution of Climate

Change: from Global to Regional. In: Climate Change 2013: The Physical Science Basis.

Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental

Panel on Climate Change. (ed [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen,

J. Boschung, A. Nauels, Y. Xia VB and PMM (eds. ).). Cambridge University Press, 138

Cambridge, United Kingdom and New York, NY, USA.

Bopp L, Monfray P, Aumont O et al. (2001) Potential impact of climate change on marine export

production. Global Biogeochemical Cycles, 15, 81–99.

Bopp L, Resplandy L, Orr JC et al. (2013) Multiple stressors of ocean ecosystems in the 21st

century: projections with CMIP5 models. Biogeosciences, 10, 6225–6245.

Börger T, Beaumont NJ, Pendleton L et al. (2014) Incorporating ecosystem services in marine

planning: The role of valuation. Marine Policy, 46, 161–170.

Boyd PW, Cornwall CE, Davison A et al. (2016) Biological responses to environmental

heterogeneity under future ocean conditions. Global change biology, 22, 2633–2650.

Boyd PW, Collins S, Dupont S et al. (2018) Experimental strategies to assess the biological

ramifications of multiple drivers of global ocean change—A review. Global Change

Biology, 24, 2239–2261.

Branch TA, Dejoseph BM, Ray LJ, Wagner CA (2013) Impacts of ocean acidification on marine

seafood. Trends in Ecology & Evolution, 28, 178–186.

Breitburg D, Levin LA, Oschlies A et al. (2018) Declining oxygen in the global ocean and

coastal waters. Science, 359.

Bruno JF, Carr LA, O’Connor MI (2015) Exploring the role of temperature in the ocean through

metabolic scaling. Ecology, 96, 3126–3140.

Caldeira K, Wickett ME (2003) Oceanography: anthropogenic carbon and ocean pH. Nature,

425, 365.

Campbell A, Robinson DG (2008) Reproductive Potential of Three American Lobster (Homarus

americanus) Stocks in the Canadian Maritimes. Canadian Journal of Fisheries and Aquatic

Sciences, 40, 1958–1967. 139

Cao L, Naylor R, Henriksson P, Leadbitter D, Metian M, Troell M, Zhang W (2015) China’s

aquaculture and the world’s wild fisheries. Science, 347, 133–135.

Capotondi A, Alexander MA, Bond NA, Curchitser EN, Scott JD (2012) Enhanced upper ocean

stratification with climate change in the CMIP3 models. Journal of Geophysical Research,

117, 1–23.

Cashion T (2016) The end use of marine fisheries landings. Fisheries Centre Research Reports,

24, 108.

Cashion T, Le Manach F, Zeller D, Pauly D (2017) Most fish destined for fishmeal production

are food-grade fish. Fish and Fisheries, 1–8.

Cashion T, Al-Abdulrazzak D, Belhabib D et al. (2018) Reconstructing global marine fishing

gear use: Catches and landed values by gear type and sector. Fisheries Research, 206, 57–

64.

Cattano C, Claudet J, Domenici P, Milazzo M (2018) Living in a high CO2 world: a global meta-

analysis shows multiple trait-mediated responses of fish to ocean acidification. Ecological

Monographs, 88, 320–335.

Checkley DM, Dickson AG, Takahashi M, Radich JA, Eisenkolb N, Asch R (2009) Elevated

CO2 enhances otolith growth in young fish. Science, 324, 1683.

Cheung WWL, Pauly D (2016) Impacts and effects of ocean warming on marine fishes. In:

Explaining ocean warming: Causes, scale, effects and consequences (eds Laffoley D,

Baxter JM), pp. 239–253. IUCN, Gland, Switzerland.

Cheung WWL, Lam VWY, Pauly D (2008a) Modelling Present and Climate-Shifted Distribution

of Marine Fishes and Invertebrates. Fisheries Centre Research Reports, 16, 76.

Cheung WW, Close C, Lam V, Watson R, Pauly D (2008b) Application of macroecological 140

theory to predict effects of climate change on global fisheries potential. Marine Ecology

Progress Series, 365, 187–197.

Cheung WWL, Lam VWY, Sarmiento JL, Kearney K, Watson R, Zeller D, Pauly D (2010)

Large-scale redistribution of maximum fisheries catch potential in the global ocean under

climate change. Global Change Biology, 16, 24–35.

Cheung WWL, Dunne J, Sarmiento JL, Pauly D (2011) Integrating ecophysiology and plankton

dynamics into projected maximum fisheries catch potential under climate change in the

Northeast Atlantic. ICES Journal of Marine Science, 68, 1008–1018.

Cheung WWL, Watson R, Pauly D (2013a) Signature of ocean warming in global fisheries

catch. Nature, 497, 365–368.

Cheung WWL, Sarmiento JL, Dunne J et al. (2013b) Shrinking of fishes exacerbates impacts of

global ocean changes on marine ecosystems. Nature Climate Change, 3, 254–258.

Cheung WWL, Reygondeau G, Frölicher TL (2016a) Large benefits to marine fisheries of

meeting the 1.5°C global warming target. Science, 6319, 1591–1594.

Cheung WWL, Frolicher TL, Asch RG et al. (2016b) Building confidence in projections of the

responses of living marine resources to climate change. ICES Journal of Marine Science, 1–

14.

Cheung WWL, Jones MC, Reygondeau G, Stock CA, Lam VWY, Froelicher T (2016c)

Structural uncertainty in projecting global fisheries catches under climate change.

Ecological Modelling, 325, 57–66.

Cheung WWL, Jones MC, Lam VWY, Miller D, Ota Y, Teh L, Sumaila UR (2017) Transform

high seas management to build climate-resilience in marine seafood supply. Fish and

Fisheries, 18, 254–263. 141

Chiu A, Li L, Guo S, Bai J, Fedor C, Naylor RL (2013) Feed and fishmeal use in the production

of carp and tilapia in China. Aquaculture, 414–415, 127–134.

Christensen V, De la Puente S, Sueiro JC, Steenbeek J, Majluf P (2014) Valuing seafood: The

Peruvian fisheries sector. Marine Policy, 44, 302–311.

Christiansen JS, Mecklenburg CW, Karamushko O V. (2014) Arctic marine fishes and their

fisheries in light of global change. Global Change Biology, 20, 352–359.

Ciais P, Sabine C, Bala G et al. (2013) Carbon and Other Biogeochemical Cycles. In: Climate

Change 2013: The physical science basis. Contribution of working group I to the fifth

assessment report of the intergovernmental panel on climate change (ed [Stocker, T.F., D.

Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia VB and PMM

(eds. ).). Cambridge University Press, Cambridge, United Kingdom and New York, NY,

USA, 465–570 pp.

Cisneros-Montemayor AM, Pauly D, Weatherdon L V., Ota Y (2016) A Global Estimate of

Seafood Consumption by Coastal Indigenous Peoples. Plos One, 11, e0166681.

Cisneros-Montemayor AM, Harper S, Tai TC (2018) The market and shadow value of informal

fish catch: A framework and application to Panama. Natural Resources Forum.

Clark TD, Sandblom E (2013) Aerobic scope measurements of fishes in an era of climate

change: Respirometry, relevance and recommendations. The Journal of Experimental

Biology, 216, 2771–2782.

Close C, Cheung W, Hodgson S, Lam V, Watson R, Pauly D (2006) Distribution ranges of

commercial fishes and invertebrates. Fishes in Databases and Ecosystems. Fisheries Centre

Research Reports, 14, 27–37.

Cobb D, Fast H, Papst MH, Rosenberg D, Rutherford R, Sareault JE (2008) Beaufort Sea Large 142

Ocean Management Area: Ecosystem Overview and Assessment Report, Vol. 2780. ii-ix +

188p. p.

Codding BF, Whitaker AR, Bird DW (2014) Global Patterns in the Exploitation of Shellfish

Global Patterns in the. The Journal of Island and Coastal Archaeology, 9, 145–149.

Cooley SR, Doney SC (2009a) Anticipating ocean acidification’s economic consequences for

commercial fisheries. Environmental Research Letters, 4, 8pp.

Cooley SR, Doney SC (2009b) Anticipating ocean acidification’s economic consequences for

commercial fisheries. Environmental Research Letters, 4, 024007.

Cooley SR, Kite-Powell HL, Doney SC (2009) Ocean acidification’s potential to alter Global

Marine ecosystem Services. Oceanography, 22, 172–181.

Cooley SR, Rheuban JE, Hart DR, Luu V, Glover DM, Hare JA, Doney SC (2015) An integrated

assessment model for helping the united states sea scallop (Placopecten magellanicus)

fishery plan ahead for ocean acidification and warming. PLoS ONE, 10, 1–27.

Corinne Le Quéré, Andrew RM, Friedlingstein P et al. (2018) Global Carbon Budget 2018.

Earth System Science Data, 10, 1–54.

Costello C, Ovando D, Clavelle T et al. (2016) Global fishery prospects under contrasting

management regimes. Proceedings of the National Academy of Sciences, 113, 5125–5129.

Crain CM, Kroeker K, Halpern BS (2008) Interactive and cumulative effects of multiple

stressors in marine systems. Ecology Letters, 11, 1304–1315.

Crain CM, Halpern BS, Beck MW, Kappel C V (2009) Understanding and Managing Human

Threats to the Coastal Marine Environment. Annals of the New York Academy of Sciences,

1162, 39–62.

Crim RN, Sunday JM, Harley CDG (2011) Elevated seawater CO2 concentrations impair larval 143

development and reduce larval survival in endangered northern abalone (Haliotis

kamtschatkana). Journal of Experimental Marine Biology and Ecology, 400, 272–277.

Cui X, Grebmeier JM, Cooper LW (2012) Feeding ecology of dominant groundfish in the

northern Bering Sea. Polar Biology, 35, 1407–1419.

Denman KL (2017) A model simulation of the adaptive evolution through mutation of the

coccolithophore Emiliania huxleyi based on a published laboratory study. Frontiers in

Marine Science, 3, 1–13.

Denman K, Christian JR, Steiner N, Po H (2011) Potential impacts of future ocean acidification

on marine ecosystems and fisheries: current knowledge and recommendations for future

research. ICES Journal of Marine Science, 68, 1019–1029.

Deutsch C, Ferrel A, Seibel B, Pörtner H-O, Huey RB (2015) Climate change tightens a

metabolic constraint on marine habitats. Science, 348, 1132–1136.

DeVries T, Le Quéré C, Andrews O et al. (2019) Decadal trends in the ocean carbon sink.

Proceedings of the National Academy of Sciences, 201900371.

DFO (2015) Fisheries and Oceans Canada National Fisheries Statistics.

Diaz RJ, Rosenberg R (2008) Spreading dead zones and consequences for marine ecosystems.

Science, 321, 926–929.

Dickson DL, Gilchrist HG (2002) Status of marine birds of the southeastern Beaufort Sea.

Arctic, 55, 46–58.

Divoky GJ, Lukacs PM, Druckenmiller ML (2015) Effects of recent decreases in arctic sea ice

on an ice-associated marine bird. Progress in Oceanography, 136, 151–161.

Doney SC, Ruckelshaus M, Emmett Duffy J et al. (2012) Climate Change Impacts on Marine

Ecosystems. Annual Review of Marine Science, 4, 11–37. 144

Dulvy NK, Sadovy Y, Reynolds JD (2003) Extinction vulnerability in marine populations. Fish

and Fisheries, 4, 25–64.

Dulvy NK, Rogers SI, Jennings S, Stelzenmller V, Dye SR, Skjoldal HR (2008) Climate change

and deepening of the North Sea fish assemblage: a biotic indicator of warming seas. Journal

of Applied Ecology, 45, 1029–1039.

Dunne JP, John JG, Shevliakova S et al. (2013) GFDL’s ESM2 global coupled climate-carbon

earth system models. Part II: Carbon system formulation and baseline simulation

characteristics. American Meteorological Society, 26, 2247–2267.

Dutkiewicz S, Morris JJ, Follows MJ, Scott J, Levitan O, Dyhrman ST, Berman-Frank I (2015)

Impact of ocean acidification on the structure of future phytoplankton communities. Nature

Climate Change, 1–19.

Dyck AJ, Sumaila UR (2010) Economic impact of ocean fish populations in the global fishery.

Journal of Bioeconomics, 12, 227–243.

Edwards M, Richardson AJ (2004) Impact of climate change on marine pelagic phenology and

trophic mismatch. Nature, 430, 881–884.

Ekau W, Auel H, Pörtner H-O, Gilbert D (2010) Impacts of hypoxia on the structure and

processes in pelagic communities (zooplankton, macro-invertebrates and fish).

Biogeosciences, 7, 1669–1699.

Ekstrom JA, Suatoni L, Cooley SR et al. (2015) Vulnerability and adaptation of US shellfisheries

to ocean acidification. Nature Climate Change, 5, 207–214.

Ellis RP, Urbina MA, Wilson RW (2016) Lessons from two high CO 2 worlds - future oceans

and intensive aquaculture. Global Change Biology, 2100, 1–8.

Fabry VJ, Seibel B a., Feely R a., Orr JC (2008) Impacts of ocean acidification on marine fauna 145

and ecosystem processes. ICES Journal of Marine Science, 65, 414–432.

Fagerli CW, Norderhaug KM, Christie HC (2013) Lack of sea urchin settlement may explain

kelp forest recovery in overgrazed areas in Norway. Marine Ecology Progress Series, 488,

119–132.

Falkenberg LJ, Dupont S, Bellerby RGJ (2018) Approaches to Reconsider Literature on

Physiological Effects of Environmental Change: Examples From Ocean Acidification

Research. Frontiers in Marine Science, 5, 1–12.

FAO (2014a) State of world fisheries and aquaculture: opportunities and challenges. Rome, 243

pp.

FAO (2014b) Fishery and aquaculture statistics.

FAO (2016) The State of World Fisheries and Aquaculture. Rome.

FAO (2018) The State of World Fisheries and Aquaculture - Meeting the sustainable

development goals. Rome.

Farrell AP, Hinch SG, Cooke SJ, Patterson DA, Crossin GT, Lapointe M, Mathes MT (2008)

Pacific salmon in hot water: applying aerobic scope models and biotelemetry to predict the

success of spawning migrations. Physiological and biochemical zoology, 81, 697–708.

Feely RA, Sabine CL, Lee K, Berelson W, Kleypas J, Fabry VJ, Millero FJ (2004) Impact of

anthropogenic CO (sub 2) on the CaCO (sub 3) system in the oceans. Science, 305, 362–

366.

Feely RA, Doney SC, Cooley SR (2009) Present conditions and future changes in a high-CO2

world. Oceanography, 22, 36–47.

Feely RA, Sabine CL, Hernandez-Ayon JM, Ianson D, Hales B (2014) Evidence for Upwelling

of Corrosive “Acidified” Water onto the Continental Shelf. Science, 320, 1490–1492. 146

Fisheries and Oceans Canada (2012) A brief history of the lobster fishery in the southern Gulf of

St. Lawrence. Moncton, New Brunswick.

Fisheries and Oceans Canada, Fisheries Joint Management Committee, Inuvialuit Game Council,

Inuvialuit Regional Corporation (2014) Beaufort Sea Integrated Fisheries Management

Framework for the Inuvialuit Settlement Region, Canada 2013-2017. 64p.

Ford J, Smit B, Wandel J, Allurut M, Shappa K, Ittusarjuat H, Qrunnut K (2008) Climate change

in the Arctic: current and future vulnerability of two communities in Canada. The

Geographical Journal, 174, 45–62.

Froese R, Pauly D (2017) FishBase. World Wide Web electronic publication, version (10/2017).

Gattuso J-P, Magnan a., Bille R et al. (2015) Contrasting futures for ocean and society from

different anthropogenic CO2 emissions scenarios. Science, 349, aac4722-1-aac4722-10.

Gattuso J, Magnan AK, Bopp L et al. (2018) Ocean Solutions to Address Climate Change and Its

Effects on Marine Ecosystems. Frontiers in Marine Science, 5, 337.

Gislason H, Daan N, Rice JC, Pope JG (2010) Size, growth, temperature and the natural

mortality of marine fish. Fish and Fisheries, 11, 149–158.

Gooding RA, Harley CDG, Tang E (2009) Elevated water temperature and carbon dioxide

concentration increase the growth of a keystone echinoderm. Proceedings of the National

Academy of Sciences, 106, 9316–9321.

Guinotte JM, Fabry VJ (2008) Ocean acidification and its potential effects on marine

ecosystems. Annals of the New York Academy of Sciences, 1134, 320–342.

Haigh R, Ianson D, Holt CA, Neate HE, Edwards AM (2015) Effects of ocean acidification on

temperate coastal marine ecosystems and fisheries in the northeast pacific. PLoS ONE, 10,

1–46. 147

Halpern BS, Walbridge S, Selkoe KA et al. (2008) A Global Map of Human Impact on Marine

Ecosystems. Science, 319, 948–952.

Hardy RW (2010) Utilization of plant proteins in fish diets: Effects of global demand and

supplies of fishmeal. Aquaculture Research, 41, 770–776.

Harley CDG, Hughes AR, Hultgren KM et al. (2006) The impacts of climate change in coastal

marine systems. Ecology Letters, 9, 228–241.

Harper S, Zeller D, Hauzer M, Pauly D, Sumaila UR (2013) Women and fisheries: Contribution

to food security and local economies. Marine Policy, 39, 56–63.

Hartmann D., Klein Tank AMG, Rusticucci M et al. (2013) Observations: Atmosphere and

Surface. In: Climate Change 2013: The Physical Science Basis. Contribution of Working

Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.

(ed [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels,

Y. Xia VB and PMM (eds. ).). Cambridge University Press, Cambridge, United Kingdom

and New York, NY, USA.

Harvey BP, Gwynn-Jones D, Moore PJ (2013) Meta-analysis reveals complex marine biological

responses to the interactive effects of ocean acidification and warming. Ecology and

Evolution, 3, 1016–1030.

Harvey BP, McKeown NJ, Rastrick SPS et al. (2016) Individual and population-level responses

to ocean acidification. Scientific Reports, 6, 1–7.

Hawkins E, Sutton R (2009) The potential to narrow uncertainty in regional climate predictions.

Bulletin of the American Meteorological Society, 90, 1095–1107.

Hein CL, Öhlund G, Englund G (2012) Future distribution of arctic char Salvelinus alpinus in

Sweden under climate change: Effects of temperature, lake size and species interactions. 148

Ambio, 41, 303–312. van Helden P (2012) The cost of research in developing countries. EMBO Reports, 13, 395.

Heston A, Summers R, Aten B (2012) Penn World Table Version 7.1 Center for International

Comparisons of Production, Income and Prices at the University of Pennsylvania.

Heuer RM, Grosell M (2014) Physiological impacts of elevated carbon dioxide and ocean

acidification on fish. AJP: Regulatory, Integrative and Comparative Physiology, 307,

R1061–R1084.

Hilborn R, Walters CJ (1992) Quantitative Fisheries Stock Assessment. Springer US, Boston,

MA.

Hill RJ, Syed I (2010) The International Comparisons Program and the Country-Product-Dummy

Method: An Application to the Asia-Pacific Region.

Hoegh-Guldberg O, Mumby J, Hooten J et al. (2007) Coral reefs under rapid climate change and

ocean acidification. Science, 318, 1737.

Hoegh-Guldberg O, Cai R, Poloczanska ES et al. (2014) The Ocean. In: 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.

(eds Barros VR, Field CB, Dokken DJ, Mastrandrea MD, Mach KJ, Bilir TE, Chatterjee M,

Ebi KL, Estrada YO, Genova RC, Girma B, Kissel ES, Levy AN, MacCracken S,

Mastrandrea PR, White LL), pp. 1655–1731. Cambridge University Press, Cambridge,

United Kingdom and New York, NY, USA.

Hoover C, Pitcher T, Christensen V (2013) Effects of hunting, fishing and climate change on the

Hudson Bay marine ecosystem: II. Ecosystem model future projections. Ecological

Modelling, 264, 143–156. 149

Hoover C, Ostertag S, Hornby C, Parker C, Hansen-Craik K, Loseto L, Pearce T (2016) The

Continued Importance of Hunting for Future Inuit Food Security. Solutions, 7, 40–51.

Howarth RW (2008) Coastal nitrogen pollution : A review of sources and trends globally and

regionally. Harmful Algae, 8, 14–20.

Hughes GM (1983) Allometry of gill dimensions in some British and American decapod

crustacea. Journal of Zoology, 200, 83–97.

Inuvialuit Harvest Study: Data and methods report 1988-1997 (2003) Inuvik, Nunavut, 209 pp.

IOC-UNESCO, UNEP (2016) Large Marine Ecosystems: Status and Trends, Vol. 4. United

Nations Environment Programme, Nairobi.

IPBES (2019) Global assessment report on biodiversity and ecosystem services of the

Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. E. S.

Brondizio, J. Settele, S. Díaz, and H. T. Ngo (editors). Bonn, Germany, XXXX p.

IPCC (2013) Climate Change 2013: The Physical Science Basis. Contribution of Working Group

I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds

Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex

V, Midgley PM). Cambridge University Press, Cambridge, United Kingdom and New York,

NY, USA, 1535 pp p.

IPCC (2014) 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 (eds Field CB, Barros VR, Dokken DJ, Mach

KJ, Mastrandrea MD, Bilir TE, Chatterjee M, Ebi KL, Estrada YO, Genova RC, Girma B,

Kissel ES, Levy AN, MacCracken S, Mastrandrea PR, White LL). Cambridge University

Press, Cambridge, United Kingdom and New York, NY, USA. 150

IPCC (2018) Summary for Policymakers (eds Masson-Delmotte V, Zhai P, Pörtner H-O, Roberts

D, Skea J, Shukla PR, Pirani A, Moufouma-Okia W, Péan C, Pidcock R, Connors S,

Matthews JBR, Chen Y, Zhou X, Gomis MI, Lonnoy E, Maycock T, Tignor M, Waterfield

T). World Meteorological Organization, Geneva, Switzerland, 32 pp. p.

Islam MS, Tanaka M (2004) Impacts of pollution on coastal and marine ecosystems including

coastal and marine fisheries and approach for management: a review and synthesis. Marine

Pollution Bulletin, 48, 624–649.

Johnson L, Rees CJC (1988) Oxygen consumption and gill surface area in relation to habitat and

lifestyle of four crab species. Comparative Biochemistry and Physiology -- Part A:

Physiology, 89, 243–246.

Jones MC, Cheung WWL (2015) Multi-model ensemble projections of climate change effects on

global marine biodiversity. ICES Journal of Marine Science, 72, 741–752.

Jones HD, Richards OG, Southern TA (1992) Gill dimensions, water pumping rate and body size

in the mussel Mytilus edulis L. Journal of Experimental Marine Biology and Ecology, 155,

213–237.

Jones MC, Dye SR, Pinnegar JK, Warren R, Cheung WWL (2012) Modelling commercial fish

distributions: Prediction and assessment using different approaches. Ecological Modelling,

225, 133–145.

Kappel C V. (2005) Losing pieces of the puzzle: threats to marine estuarine, and diadromous

species. Frontiers in Ecology and the Environment, 3, 275–282.

Karnovsky NJ, Hobson KA, Iverson S, Hunt GL (2008) Seasonal changes in diets of seabirds in

the North Water Polynya: A multiple-indicator approach. Marine Ecology Progress Series,

357, 291–299. 151

Kelley E, Sherman K (2018) Trends of the Large Marine Ecosystem assessment and

management approach as reflected in the literature. Ocean and Coastal Management, 155,

104–112.

Kleypas JA, Feely RA, Fabry VJ, Langdon C, Sabine CL, Robbins LL (2006) Impacts of Ocean

Acidification on Coral Reefs and Other Marine Calcifiers : A Guide for Future Research.

St. Petersburg, Florida, 88 pp.

Koenigstein S, Mark FC, Gößling-Reisemann S, Reuter H, Poertner HO (2016) Modelling

climate change impacts on marine fish populations: process-based integration of ocean

warming, acidification and other environmental drivers. Fish and Fisheries, 17, 972–1004.

Kovach RP, Gharrett AJ, Tallmon DA (2012) Genetic change for earlier migration timing in a

pink salmon population. Proceedings of the Royal Society B, 279, 3870–3878.

Kremer JN (1983) Ecological implications of parameter uncertainty in stochastic simulation.

Ecological Modelling, 18, 187–207.

Kroeker KJ, Kordas RL, Crim RN, Singh GG (2010) Meta-analysis reveals negative yet variable

effects of ocean acidification on marine organisms. Ecology Letters, 13, 1419–1434.

Kroeker KJ, Kordas RL, Crim R et al. (2013) Impacts of ocean acidification on marine

organisms: Quantifying sensitivities and interaction with warming. Global Change Biology,

19, 1884–1896.

Kroeker KJ, Gaylord B, Hill TM, Hosfelt JD, Miller SH, Sanford E (2014) The role of

temperature in determining species’ vulnerability to ocean acidification: a case study using

Mytilus galloprovincialis. PLoS ONE, 9, e100353.

Kroeker KJ, Kordas RL, Harley CDG (2017) Embracing interactions in ocean acidification

research: Confronting multiple stressor scenarios and context dependence. Biology Letters, 152

13, 20160802.

Kronen M, Vunisea A, Magron F, McArdle B (2010) Socio-economic drivers and indicators for

artisanal coastal fisheries in Pacific island countries and territories and their use for fisheries

management strategies. Marine Policy, 34, 1135–1143.

Kunz KL, Claireaux G, Pörtner H-O, Knust R, Mark FC (2018) Aerobic capacities and

swimming performance of polar cod ( Boreogadus saida ) under ocean acidification and

warming conditions. The Journal of Experimental Biology, 221, jeb184473.

Lam VWY, Sumaila UR, Dyck A, Pauly D, Watson R (2011) Construction and first applications

of a global cost of fishing database. ICES Journal of Marine Science, 68, 1996–2004.

Lam VWY, Cheung WWL, Sumaila UR (2014) Marine capture fisheries in the Arctic: winners

or losers under climate change and ocean acidification? Fish and Fisheries.

Lam VWY, Cheung WWL, Reygondeau G, Sumaila UR (2016) Projected change in global

fisheries revenues under climate change. Scientific Reports, 6, 32607.

Lefevre S (2016) Are global warming and ocean acidification conspiring against marine

ectotherms? A meta-analysis of the respiratory effects of elevated temperature, high CO2

and their interaction. Conservation Physiology, 4, 1–31.

Lefevre S, Mckenzie DJ, Nilsson GE (2017) Models projecting the fate of fish populations under

climate change need to be based on valid physiological mechanisms. Global Change

Biology, 1–11.

Lluch-Belda D, Crawford RJM, Kawasaki T, MacCall a. D, Parrish RH, Schwartzlose R a.,

Smith PE (1989) World-wide fluctuations of sardine and anchovy stocks: the regime

problem. South African Journal of Marine Science, 8, 195–205.

Lohbeck KT, Riebesell U, Reusch TBH (2012) Adaptive evolution of a key phytoplankton 153

species to ocean acidification. Nature Geoscience, 5, 346–351.

Loseto LL, Stern GA, Connelly TL et al. (2009) Summer diet of beluga whales inferred by fatty

acid analysis of the eastern Beaufort Sea food web. Journal of Experimental Marine

Biology and Ecology, 374, 12–18.

Lotze HK, Lenihan HS, Bourque BJ et al. (2006) Depletion, degradation, and recovery potential

of estuaries and coastal seas. Science, 312, 1806–1809.

Lovenduski NS, McKinley GA, Fay AR, Lindsay K, Long MC (2016) Partitioning uncertainty in

ocean carbon uptake projections: Internal variability, emission scenario, and model

structure. Global Biogeochemical Cycles, 30, 1276–1287.

Mackas DL, Goldblatt R, Lewis AG (1998) Interdecadal variation in developmental timing of

Neocalanus plumchrus populations at Ocean station P in the subarctic North Pacific.

Canadian Journal of Fisheries and Aquatic Sciences, 55, 1878–1893.

Marshall KN, Kaplan IC, Hodgson EE et al. (2017) Risks of ocean acidification in the California

Current food web and fisheries: ecosystem model projections. Global Change Biology, 23,

1525–1539.

Masson-Delmotte V, Schulz M, Abe-Ouchi A et al. (2013) Information from Paleoclimate

Archives. In: Climate Change 2013: The Physical Science Basis. Contribution of Working

Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.

(ed [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels,

Y. Xia VB and PMM (eds. ).). Cambridge University Press, Cambridge, United Kingdom

and New York, NY, USA.

Mastrandrea MD, Heller NE, Root TL, Schneider SH (2010) Bridging the gap: Linking climate-

impacts research with adaptation planning and management. Climatic Change, 100, 87–101. 154

Mathis JT, Cooley SR, Lucey N et al. (2015) Ocean acidification risk assessment for Alaska’s

fishery sector. Progress in Oceanography, 136, 71–91.

McCauley DJ, Pinsky ML, Palumbi SR, Estes JA, Joyce FH, Warner RR (2015) Marine

defaunation: Animal loss in the global ocean. Science, 347, 1255641–1255641.

McNeil BI, Sasse TP (2016) Future ocean hypercapnia driven by anthropogenic amplification of

the natural CO2 cycle. Nature, 529, 383–386.

McNeil BI, Matear RJ, Barnes DJ (2004) Coral reef calcification and climate change: The effect

of ocean warming. Geophysical Research Letters, 31, 2–5.

Melnychuk MC, Clavelle T, Owashi B, Strauss K (2016) Reconstruction of global ex-vessel

prices of fished species. ICES Journal of Marine Science.

Menu-Courey K, Noisette F, Piedalue S et al. (2019) Energy metabolism and survival of the

juvenile recruits of the American lobster (Homarus americanus) exposed to a gradient of

elevated seawater pCO2. Marine Environmental Research, 143, 111–123.

Michel C, Bluhm B, Gallucci V et al. (2012) Biodiversity of Arctic marine ecosystems and

responses to climate change. Biodiversity, 13, 200–214.

Monaco CJ, Helmuth B (2011) Tipping Points, Thresholds and the Keystone Role of Physiology

in Marine Climate Change Research, Vol. 60. 123–160 pp.

Moore C (2015) Welfare estimates of avoided ocean acidification in the U.S. mollusk market.

Journal of Agricultural and Resource Economics, 40, 50–62.

Munday PL, Dixson DL, Donelson JM, Jones GP, Pratchett MS, Devitsina G V, Døving KB

(2009a) Ocean acidification impairs olfactory discrimination and homing ability of a marine

fish. Proceedings of the National Academy of Sciences of the United States of America, 106,

1848–1852. 155

Munday PL, Crawley NE, Nilsson GE (2009b) Interacting effects of elevated temperature and

ocean acidification on the aerobic performance of coral reef fishes. Marine Ecology

Progress Series, 388, 235–242.

Munday PL, Dixson DL, McCormick MI, Meekan M, Ferrari MCO, Chivers DP (2010)

Replenishment of fish populations is threatened by ocean acidification. Proceedings of the

National Academy of Sciences, 107, 12930–12934.

Myers RA, Worm B (2003) Rapid worldwide depletion of predatory fish communities. Nature,

423, 280–283.

Nagelkerken I, Connell SD (2015) Global alteration of ocean ecosystem functioning due to

increasing human CO2 emissions. Proceedings of the National Academy of Sciences, 112,

13272–7.

Naylor RL, Goldburg RJ, Primavera JH et al. (2000) Effect of aquaculture on world fish

supplies. Nature, 405, 1017–24.

Naylor RL, Hardy RW, Bureau DP et al. (2009) Feeding aquaculture in an era of finite resources.

Proceedings of the National Academy of Sciences, 106, 15103–15110.

Nerem RS, Beckley BD, Fasullo JT, Hamlington BD, Masters D, Mitchum GT (2018) Climate-

change-driven accelerated sea-level rise detected in the altimeter era. Proceedings of the

National Academy of Sciences of the United States of America, 115, 2022–2025.

Niemi A, Johnson J, Majewski A, Melling H, Reist J, Williams W (2012) State of the Ocean

Report for the Beaufort Sea Large Ocean Management Area. Canadian Manuscript Report

of Fisheries and Aquatic Sciences, 2977, vi + 51.

Nienhuis S, Palmer AR, Harley CDG (2010) Elevated CO2 affects shell dissolution rate but not

calcification rate in a marine snail. Proceedings of the Royal Society B, 277, 2553–2558. 156

Noonburg EG, Byers JE (2016) Bad neighbors: how spatially disjunct habitat degradation can

cause system-wide population collapse. Ecology, 97, 2858–2866.

North Pacific Fisheries Management Council, Arctic Fishery Management Plan for Fish

Resources of the Arctic Management Area.

Nunoo FKE, Asiedu B, Amador K, Belhabib D, Lam V, Sumaila R, Pauly D (2014) Marine

Fisheries Catches in Ghana: Historic Reconstruction for 1950 to 2010 and Current

Economic Impacts. Reviews in Fisheries Science & Aquaculture, 22, 274–283.

O’Connor MI, Bruno JF, Gaines SD, Halpern BS, Lester SE, Kinlan BP, Weiss JM (2007)

Temperature control of larval dispersal and the implications for marine ecology, evolution,

and conservation. Proceedings of the National Academy of Sciences of the United States of

America, 104, 1266–71.

Olsen E, Kaplan IC, Ainsworth C et al. (2018) Ocean Futures Under Ocean Acidification,

Marine Protection, and Changing Fishing Pressures Explored Using a Worldwide Suite of

Ecosystem Models. Frontiers in Marine Science, 5, 1–23.

Palomares MLD, Pauly D (2017) SeaLifeBase. World Wide Web Electronic Publication, version

(06/2017).

PAME (2013) Large Marine Ecosystems (LMEs) of the Arctic area - Revision of the Arctic LME

map 15th of May 2013. 1–20 pp.

Parker-Stetter SL, Horne JK, Weingartner TJ (2011) Distribution of polar cod and age-0 fish in

the U.S. Beaufort Sea. Polar Biology, 34, 1543–1557.

Parmesan C (2006) Ecological and Evolutionary Responses to Recent Climate Change. Annual

Review of Ecology, Evolution, and Systematics, 37, 637–669.

Paul C, Matthiessen B, Sommer U (2015) Warming, but not enhanced CO 2 concentration, 157

quantitatively and qualitatively affects phytoplankton biomass. Marine Ecology Progress

Series, 528, 39–51.

Pauly D (1980a) On the interrelationships between natural mortality, growth parameters, and

mean environmental temperature in 175 fish stocks. ICES Journal of Marine Science, 39,

175–192.

Pauly D (1980b) On the interrelationships between natural mortality, growth parameters, and

mean environmental temperatuer in 175 fish stocks. Journal du Conseil CIEM, 39, 175–

192.

Pauly D (1981) The relationship between gill surface area and growth performance in fish: a

generalization of von Bertalanffy’s theory of growth. Berichte der Deutschen

Wissenschaftlichen Kommission für Meeresforschung, 28, 251–282.

Pauly D, Cheung WWL (2017) Sound physiological knowledge and principles in modeling

shrinking of fishes under climate change. Global Change Biology, 1–12.

Pauly D, Zeller D (2015) Sea Around Us Concepts, Design and Data (seaaroundus.org).

Pauly D, Zeller D (2016) Catch reconstructions reveal that global marine fisheries catches are

higher than reported and declining. Nature Communications, 7, 10244.

Pauly D, Christensen V, Walters C (2000) Ecopath, Ecosim, and Ecospace as tools for evaluating

ecosystem impact of fisheries. ICES Journal of Marine Science, 57, 697–706.

Pauly D, Christensen V, Guénette S et al. (2002) Towards Sustainability in World Fisheries.

Nature, 418, 689–695.

Van Pelt TI, Huntington HP, Romanenko O V., Mueter FJ (2017) The missing middle: Central

Arctic Ocean gaps in fishery research and science coordination. Marine Policy, 85, 79–86.

Perovich D, Meier W, Tschudi M et al. (2017) Sea ice [in Arctic Report Card 2017], 158

http://www.arctic.noaa.gov/Report-Card.

Perry AL, Low PT, Ellis JR, Reynolds JD (2005) Climate Change and Distribution Shifts in

Marine Fishes. Science, 308, 1912–1915.

Pershing AJ, Mills KE, Dayton AM, Franklin BS, Kennedy BT (2018) Evidence for adaptation

from the 2016 marine heatwave in the Northwest Atlantic Ocean. Oceanography, 31, 152–

161.

Pikitch EK, Rountos KJ, Essington TE et al. (2014) The global contribution of forage fish to

marine fisheries and ecosystems. Fish and Fisheries, 15, 43–64.

Pistevos JCA, Nagelkerken I, Rossi T, Connell SD (2016) Ocean acidification alters temperature

and salinity preferences in larval fish. Oecologia.

Poloczanska ES, Burrows MT, Brown CJ et al. (2016) Responses of Marine Organisms to

Climate Change across Oceans. Frontiers in Marine Science, 3, 1–21.

Pörtner H (2008) Ecosystem effects of ocean acidification in times of ocean warming: a

physiologist’s view. Marine Ecology Progress Series, 373, 203–217.

Pörtner H (2010) Oxygen- and capacity-limitation of thermal tolerance : a matrix for integrating

climate-related stressor effects in marine ecosystems. The Journal of Experimental Biology,

213, 881–893.

Pörtner HO, Farrell AP (2008) Physiology and climate change. Science, 322, 690–692.

Pörtner H-O, Knust R (2007) Climate change affects marine fishes through oxygen limiation of

thermal tolerance. Science, 315, 95–97.

Pörtner HO, Lannig G (2009) Oxygen and Capacity Limited Thermal Tolerance. In: Fish

Physiology, 1st edn, Vol. 27 (eds Richards JG, Farrell AP, Brauner CJ), pp. 143–191.

Elsevier Inc., Burlington. 159

Pörtner HO, Berdal B, Blust R et al. (2001) Climate induced temperature effects on growth

performance, fecundity and recruitment in marine fish: Developing a hypothesis for cause

and effect relationships in Atlantic cod (Gadus morhua) and common eelpout (Zoarces

viviparus). Continental Shelf Research, 21, 1975–1997.

Pörtner HO, Bennett AF, Bozinovic F et al. (2006) Trade‐Offs in Thermal Adaptation: The Need

for a Molecular to Ecological Integration. Physiological and Biochemical Zoology:

Ecological and Evolutionary Approaches, 79, 295–313.

Pörtner H-O, Karl DM, Boyd PW et al. (2014) IPCC - Chapter 6 Ocean Systems. 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, 411–484.

Pörtner H-O, Bock C, Mark FC (2017) Oxygen- and capacity-limited thermal tolerance: bridging

ecology and physiology. The Journal of Experimental Biology, 220, 2685–2696.

Pörtner H-O, Bock C, Mark FC (2018a) Oxygen- and capacity-limited thermal tolerance:

blurring ecology and physiology. The Journal of experimental biology, 221, jeb169615.

Pörtner H-O, Bock C, Mark FC (2018b) Connecting to ecology: a challenge for comparative

physiologists? Response to ‘Oxygen- and capacity-limited thermal tolerance: blurring

ecology and physiology.’ The Journal of Experimental Biology, 221, jeb169615.

Priest H, Usher PJ (2004) The Nunavut Wildlife Harvest Study. Iqaluit, Nunavut, 816 pp.

Przeslawski R, Byrne M, Mellin C (2015) A review and meta-analysis of the effects of multiple

abiotic stressors on marine embryos and larvae. Global Change Biology, 21, 2122–2140.

Purvis A, Gittleman JL, Cowlishaw G, Mace GM (2000) Predicting extinction risk in declining

species. Proceedings of the Royal Society B: Biological Sciences, 267, 1947–1952. 160

Queirós AM, Fernandes JA, Faulwetter S et al. (2015) Scaling up experimental ocean

acidification and warming research: from individuals to the ecosystem. Global Change

Biology, 21, 130–143.

Le Quéré C, Andrew RM, Canadell JG et al. (2016) Global Carbon Budget 2016. Earth System

Science Data, 8, 605–649.

Le Quesne WJF, Pinnegar JK (2012) The potential impacts of ocean acidification : scaling from

physiology to fisheries*. Fish and Fisheries, 13, 333–344.

R Core Team (2018) R: A Language and environment for statistical computing.

Ramcharran H (2002) Oil production responses to price changes: An empirical application of the

competitive model to OPEC and non-OPEC countries. Energy Economics, 24, 97–106.

Rao DSP (2004) The country-product-dummy method: a stochastic approach to the computation

of purchasing power parities in the ICP. In: Price and Productivity Measurement, Volume

4: International Comparisons and Trade (eds Diewert WE, Balk BM, Fixler D, Fox KJ,

Nakamura AO), p. 23. Trafford Press, Brisbane, Australia.

Reed JK (2002) Deep-water Oculina coral reefs of Florida: biology, impacts, and management.

Hydrobiologia, 471, 43–55.

Rhein M, Rintoul SR, Aoki S et al. (2013) Observations: Ocean. In: Climate Change 2013: The

Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of

the Intergovernmental Panel on Climate Change. (ed [Stocker, T.F., D. Qin, G.-K. Plattner,

M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia VB and PMM (eds. ).). Cambridge

University Press, Cambridge, United Kingdom and New York, NY, USA.

Rheuban JE, Doney SC, Cooley SR, Hart DR (2018) Projected impacts of future climate change

, ocean acidification , and management on the US Atlantic sea scallop (Placopecten 161

magellanicus) fishery. PLoS ONE, 1–21.

Ricke KL, Orr JC, Schneider K, Caldeira K (2013) Ricks to coral reefs from ocean carbonate

chemistry changes in recent earth system model projections. Environmental Research

Letters, 8, 1–6.

Ries JB, Cohen AL, Mccorkle DC (2009) Marine calcifiers exhibit mixed responses to CO2-

induced ocean acidification. Geology, 37, 1131–1134.

Rodgers KB, Lin J, Frölicher TL (2015) Emergence of multiple ocean ecosystem drivers in a

large ensemble suite with an Earth system model. Biogeosciences, 12, 3301–3320.

Sabine CL, Feely RA, Gruber N et al. (2004) The Oceanic Sink for Anthropogenic CO2.

Science, 305, 367–371.

Sala E, Mayorga J, Costello C et al. (2018) The economics of fishing the high seas. Science

Advances, 4, 1–14.

Salze G, McLean E, Battle PR, Schwarz MH, Craig SR (2010) Use of soy protein concentrate

and novel ingredients in the total elimination of fish meal and fish oil in diets for juvenile

cobia, Rachycentron canadum. Aquaculture, 298, 294–299.

Schlüter MH, Merico A, Reginatto M, Boersma M, Wiltshire KH, Greves W (2010)

Phenological shifts of three interacting zooplankton groups in relation to climate change.

Global Change Biology, 16, 3144–3153.

Schmidtko S, Stramma L, Visbeck M (2017) Decline in global oceanic oxygen content during

the past five decades. Nature, 542, 335–339.

Selig ER, Hole DG, Allison EH et al. (2018) Mapping global human dependence on marine

ecosystems. Conservation Letters, 1–10.

Shepherd CJ, Jackson AJ (2013) Global fishmeal and fish-oil supply: Inputs, outputs and 162

markets. Journal of Fish Biology, 83, 1046–1066.

Sheridan JA, Bickford D (2011) Shrinking body size as an ecological response to climate

change. Nature Climate Change, 1, 401–406.

Sibert JR, Hampton J, Fournier DA, Bills PJ (1999) An advection-diffusion-reaction model for

the estimation of fish movement par ... Canadian Journal of Fisheries and Aquatic

Sciences, 56, 925–938.

Silver M (2009) The Hedonic Country Product Dummy Method and Quality Adjustments for

Purchasing Power Parity Calculations. IMF Working Paper WP/09/271, International

Monetary Fund. Washington, D.C., USA.

Singh GG, Cisneros-Montemayor AM, Swartz W et al. (2017) A rapid assessment of co-benefits

and trade-offs among Sustainable Development Goals. Marine Policy.

Small C, Nicholls RJ (2003) A Global Analysis of Human Settlement in Coastal Zones. Journal

of Coastal Research, 19, 584–599.

Solomon S, Plattner G-K, Knutti R, Friedlingstein P (2009) Irreversible climate change due to

carbon dioxide emissions. Proceedings of the National Academy of Sciences of the United

States of America, 106, 1704–9.

Somero GN (2012) The Physiology of Global Change : Linking Patterns to Mechanisms. Annual

Review of Marine Science, 4, 39–61.

Srinivasan UT, Cheung WWL, Watson R, Sumaila UR (2010) Food security implications of

global marine catch losses due to overfishing. Journal of Bioeconomics, 12, 183–200.

Srinivasan UT, Watson R, Rashid Sumaila U (2012) Global fisheries losses at the exclusive

economic zone level, 1950 to present. Marine Policy, 36, 544–549.

Steiner N, Azetsu-scott K, Hamilton J et al. (2015) Observed trends and climate projections 163

affecting marine ecosystems in the Canadian Arctic. Environmental Reviews, 23, 191–239.

Steneck RS, Graham MH, Bourque BJ, Corbett D, Erlandson JM, Estes JA, Tegner MJ (2002)

Kelp forest ecosystems: biodiversity, stability, resilience and future. Environmental

Conservation, 29.

Stramma L, Johnson GC, Sprintall J, Mohrholz V (2008) Expanding oxygen-minimum zones in

the tropical oceans. Science (New York, N.Y.), 320, 655–658.

Stramma L, Schmidtko S, Levin LA, Johnson GC (2010) Ocean oxygen minima expansions and

their biological impacts. Deep Sea Research Part I: Oceanographic Research Papers, 57,

587–595.

Stramma L, Prince ED, Schmidtko S et al. (2011) Expansion of oxygen minimum zones may

reduce available habitat for tropical pelagic fishes. Nature Climate Change, 2, 33–37.

Sumaila UR, Marsden AD, Watson R, Pauly D (2007) A Global Ex-vessel Fish Price Database:

Construction and Applications. Journal of Bioeconomics, 9, 39–51.

Sumaila UR, Cheung W, Dyck A et al. (2012) Benefits of Rebuilding Global Marine Fisheries

Outweigh Costs. PLoS ONE, 7, e40542.

Sumaila UR, Tai TC, Lam VWY et al. (2019) Benefits of the Paris Agreement to ocean life,

economies, and people. Science Advances, 5, eaau3855.

Summers R (1973) International Price Comparisons Based Upon Incomplete Data. Review of

Income and Wealth, 19, 1–16.

Sunday JM, Crim RN, Harley CDG, Hart MW (2011) Quantifying Rates of Evolutionary

Adaptation in Response to Ocean Acidification. PLoS ONE, 6, e22881.

Sunday JM, Calosi P, Dupont S, Munday PL, Stillman JH, Reusch TBH (2014) Evolution in an

acidifying ocean. Trends in Ecology & Evolution, 29, 117–125. 164

Sunday JM, Fabricius KE, Kroeker KJ et al. (2017) Ocean acidification can mediate biodiversity

shifts by changing biogenic habitat. Nature Climate Change, 1, 1–6.

Swartz W, Sala E, Tracey S, Watson R, Pauly D (2010) The Spatial Expansion and Ecological

Footprint of Fisheries (1950 to Present). PLoS ONE, 5, e15143.

Swartz W, Sumaila R, Watson R (2013) Global Ex-vessel Fish Price Database Revisited: A New

Approach for Estimating ‘Missing’ Prices. Environmental and Resource Economics, 56,

467–480.

Tacon AGJ, Metian M (2008) Global overview on the use of fish meal and fish oil in industrially

compounded aquafeeds: Trends and future prospects. Aquaculture, 285, 146–158.

Tacon AGJ, Metian M (2015) Feed Matters : Satisfying the Feed Demand of Aquaculture.

Reviews in Fisheries Science & Aquaculture, 23, 1–10.

Tai TC, Robinson JPW (2018) Enhancing Climate Change Research With Open Science.

Frontiers in Environmental Science, 6, 1–5.

Tai TC, Cashion T, Lam VWY, Swartz W, Sumaila UR (2017) Ex-vessel Fish Price Database:

Disaggregating Prices for Low-Priced Species from Reduction Fisheries. Frontiers in

Marine Science, 4, 1–10.

Tai TC, Harley CDG, Cheung WWL (2018) Comparing model parameterizations of the

biophysical impacts of ocean acidification to identify limitations and uncertainties.

Ecological Modelling, 385, 1–11.

Teh LCL, Sumaila UR (2013) Contribution of marine fisheries to worldwide employment. Fish

and Fisheries, 14, 77–88.

Teh LCL, Teh LSL, Starkhouse B, Rashid Sumaila U (2009) An overview of socio-economic

and ecological perspectives of Fiji’s inshore reef fisheries. Marine Policy, 33, 807–817. 165

Thackeray SJ, Sparks TH, Frederiksen M et al. (2010) Trophic level asynchrony in rates of

phenological change for marine, freshwater and terrestrial environments. Global Change

Biology, 16, 3304–3313.

Then AY, Hoenig JM, Hall NG, Hewitt DA (2015) Evaluating the predictive performance of

empirical estimators of natural mortality rate using information on over 200 fish species.

ICES Journal of Marine Science, 72, 82–92.

Timmermans M-L, Ladd C, Wood K (2017) Sea Surface Temperature [in Arctic Report Card

2017], http://www.arctic.noaa.gov/Report-Card.

Tremblay J-É, Gratton Y, Carmack EC, Payne CD, Price NM (2002) Impact of the large-scale

Arctic circulation and the North Water Polynya on nutrient inventories in Baffin Bay.

Journal of Geophysical Research, 107, 3112.

Trenkel VM, Pinnegar JK, Dawson WA, Buit MH, Tidd AN (2005) Spatial and temporal

structure of predator – prey relationships in the Celtic Sea fish community. Marine Ecology

Progress Series, 299, 257–268.

Vancoppenolle M, Bopp L, Madec G, Dunne J, Ilyina T, Halloran PR, Steiner N (2013) Future

arctic ocean primary productivity from CMIP5 simulations: Uncertain outcome, but

consistent mechanisms. Global Biogeochemical Cycles, 27, 605–619.

Vaughan DG, Comiso JC, Allison I et al. (2013) Observations: Cryosphere. In: Climate Change

2013: The Physical Science Basis. Contribution of Working Group I to the Fifth

Assessment Report of the Intergovernmental Panel on Climate Change. (ed [Stocker, T.F.,

D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia VB and

PMM (eds. ).). Cambridge University Press, Cambridge, United Kingdom and New York,

NY, USA. 166

Vizcarra N (2018) Arctic sea ice maximum at second lowest in the satellite record. NSIDC:

National Snow and Ice Data Center Press Release. van Vuuren DP, Edmonds J, Kainuma M et al. (2011) The representative concentration

pathways: An overview. Climatic Change, 109, 5–31.

Waldbusser GG, Gray MW, Hales B et al. (2016) Slow shell building, a possible trait for

resistance to the effects of acute ocean acidification. Limnology and Oceanography, 1969–

1983.

Walters C, Martell S (2004) Fisheries Ecology and Management, 1st editio edn. Princeton

University Press.

Watson R, Kitchingman A, Gelchu A, Pauly D (2004) Mapping global fisheries: sharpening our

focus. Fish and Fisheries, 5, 168–177.

Watson S-A, Southgate PC, Tyler PA, Peck LS (2009) Early Larval Development of the Sydney

Rock Oyster Saccostrea Glomerata Under Near-Future Predictions of Co2-Driven Ocean

Acidification. Journal of Shellfish Research, 28, 431–437.

Watts P, Koutouki K, Booth S, Blum S (2017) Inuit food security in canada: arctic marine

ethnoecology. Food Security, 9, 421–440.

Weber B (2018) Canada to join international moratorium on High Arctic commercial fishing.

CBC.

Welch HE, Bergmann MA, Siferd TD et al. (1992) Energy Flow through the Marine Ecosystem

of the Lancaster Sound Region, Arctic Canada. Arctic, 45, 343–357.

Whitehouse GA, Aydin K, Essington TE, Hunt GL (2014) A trophic mass balance model of the

eastern Chukchi Sea with comparisons to other high-latitude systems. Polar Biology, 37,

911–939. 167

Wickham H, Francois R (2016) dplyr: A grammar of data manipulation.

Willeit M, Ganopolski A, Calov R, Brovkin V (2019) Mid-Pleistocene transition in glacial cycles

explained by declining CO 2 and regolith removal. Science Advances, 5, eaav7337.

Williams SL, Smith JE (2007) A Global Review of the Distribution, Taxonomy, and Impacts of

Introduced Seaweeds. Annual Review of Ecology and Systematics, 38, 327–359.

Zeller D, Booth S, Pakhomov E, Swartz W, Pauly D (2011) Arctic fisheries catches in Russia ,

USA , and Canada: baselines for neglected ecosystems. Polar Biology, 34, 955–973.

168

Appendices

Appendix A – Supplementary material for “Ex-vessel fish price database: disaggregating prices for low-priced species from reduction fisheries.”

The raw price data collected for estimation were primarily sourced from developed nations in

North America and Europe (Appendix Table A.1), and a greater proportion of data from more recent time periods (Appendix Table A.2). The rule-based stepwise schematic used in price estimation to match raw price data to each taxon-year-country reported catch is show in

Appendix Table A.3. It also shows the number of prices that fall within each category. The matching year is always retained in the rule-based schematic.

Linear regression model analysis between the previous version and version of the price database show a significant difference in average price trends over time (Appendix Table A.3). However, there was no significant difference in model fit between the model that included interaction effects (Model 1) and the model that included only additive effects (Model 2).

For full figure, see online supplementary material Appendix Figure A.1.

For full figure, see online supplementary material Appendix Figure A.2.

and fish oil, and other uses), and previous methods with single prices estimates for all product types. Linear trends show the value of estimating prices separately in specific countries.

Appendix Table A.1. Reported ex-vessel price data records used in the estimation for direct human consumption, DHC, and non-DHC purposes (i.e., fishmeal, fish oil, other), by country and region.† No. of records: No. of records: Region Country DHC : non-DHC Region Country DHC : non-DHC Europe UK 4621 : 2 Asia Japan 1771 Portugal 3473 Indonesia 921 : 2 Spain 2954 South Korea 315 Denmark 2536 : 9 Philippines 305 : 128 France 2369 : 2 Thailand 207 : 1305 Norway 2061 : 177 Brunei Dar. 91 Greece 1605 Malaysia 76 : 103 Italy 1592 India 49 : 10 Belgium 1567 Vietnam 0 : 275 Ireland 1534 China 0 : 30 Germany 1140 Bangladesh 0 : 2 Iceland 1090 : 4 Netherlands 1017 Oceania Australia 969 Sweden 895 : 4 Amer. Samoa 815 Finland 473 : 12 N. Marianas 430 Malta 422 Guam 388 Poland 323 N. Cyprus 321 S. America Brazil 1200 : 1 Slovenia 295 Chile 152 : 258 Estonia 257 Peru 0 : 6 Lithuania 236 Argentina 0 : 2 Latvia 194 Bulgaria 191 Africa Namibia 170 Romania 115 Mauritania 57 S. Cyprus 17 South Africa 22 : 3 Morocco 7 : 4 N. America USA 17,530 : 267 Canada 2353 : 2 Eurasia Russian Fed 225 Mexico 170 : 4 Greenland 58

†Cells with only one value represent DHC. 170

Appendix Table A.2. Reported ex-vessel price data records used in the estimation for direct human consumption, DHC, and non-DHC purposes (i.e., fishmeal, fish oil, other), and the proportion of reported prices to unique species-country-year global catch records by time period. No. of price Proportion of price records: to catch records: Period DHC : non-DHC DHC : non-DHC (%) 1950-54 1505 : 24 4.2 : 0.12 1955-59 1662 : 40 4.5 : 0.20 1960-64 1681 : 84 4.2 : 0.38 1965-69 1821 : 93 4.2 : 0.38 1970-74 1978 : 78 4.1 : 0.28 1975-79 2490 : 90 4.7 : 0.30 1980-84 3648 : 53 6.5 : 0.17 1985-89 4002 : 28 6.8 : 0.08 1990-94 4640 : 141 7.7 : 0.42 1995-99 6978 : 41 11.1 : 0.12 2000-04 7862 : 794 11.8 : 2.1 2005-09 16,845 : 878 24.5 : 2.3 2010 4466 : 268 32.4 : 3.5 Total 59,578 : 2612 9.25 : 0.72

Appendix Table A.3. Chronological order for estimating prices based on certain criteria and the number of ex-vessel prices in our price database that were matched to each category.

Matching category Step DHC FMFO/Other Reported data (no estimation) A 29744 740 Species B 95250 4477 Genus & functional group C 60103 1245 Genus D 27847 473 Family & functional group E 140748 9528 Family F 66602 13457 FAO ISSCAAP & functional group G 164454 12987 FAO ISSCAAP H 65331 22596 Order & functional group I 4361 2599 Order J 7 47149 Class & Functional Group K 3976 13884 Class J 187 137985 Functional group M 8863 1658 Taxon median N 218 94947 Year median O 0 538 Total 667691 364263 171

Residual Model Formula K AIC ΔAIC AIC weight DF

1 Price ~ Year * Method 3 110 1493.46 0.44 0.45

2 Price ~ Year + Method 2 111 1493.03 - 0.55

3 Price ~ Year 1 112 1508.25 15.2 2.8*10-4

172

Appendix B – Supplementary material for “Comparing model parameterizations of the biophysical impacts of ocean acidification to identify limitations and uncertainties.”

B.1 Dynamic bioclimatic envelope model

Details of the original model can be found in Cheung et al. (2011, 2016b). We provide a simplified conceptual schematic of the model structure in Appendix Figure B.1 and summarize relevant aspects that pertain to this study.

Initial species distribution

Initial species distributions were obtained from the SAU database, which uses a sequence of steps based on specific criteria to generate the geospatial habitat of each species on a 30’ longitude, by 30’ latitude grid (for details see http://www.seaaroundus.org/) (Close et al., 2006;

Pauly and Zeller, 2015). Species distributions were first restricted based on observed presences in statistical areas defined by Food and Agricultural Organization (FAO). Next, their distributions were restricted by latitudinal range within FAO area(s) and then by an expert reviewed range-limiting polygon. Distributions were further filtered based on species-specific parameters including depth range, habitat preference (e.g. coral, estuaries, seagrass), and equatorial submergence. Habitat data were primarily collected from SeaLifeBase

(www.sealifebase.org) (Palomares and Pauly, 2017).

Habitat suitability and carrying capacity

Environmental and habitat suitability was determined by overlaying environmental variables

(from ESMs) such as temperature, salinity, depth, sea-ice, and dissolved oxygen concentration

173

over initial species distribution maps. Initial environmental conditions are taken from ESMs between the years 1951-1970. Habitat preferences based on species-specific data collected primarily from SeaLifeBase were also incorporated to characterize a bioclimatic envelope. A major assumption is that each geographical 30’x30’ longitude-latitude cell from the initial species distribution is at carrying capacity at time zero, and as conditions within each cell changes carrying capacity also changes. Carrying capacity is positively correlated with habitat suitability (Cheung et al., 2008, 2016).

Individual growth and population dynamics

Individual growth is modelled using the von Bertalanffy growth function with species-specific parameters (obtained from SeaLifeBase) and constrained by ecophysiological conditions including temperature, oxygen, and pH (Cheung et al., 2011). Population growth is modelled using the logistic growth function (Hilborn and Walters, 1992), where the intrinsic rate of population increase was calculated from the natural mortality rate and carrying capacity was determined by the initial distribution and habitat suitability (Cheung et al., 2008). Population mortality rates were calculated from an empirical equation using: the maximum weight of an individual, von Bertalanffy growth parameter K, and mean temperature of a species’ range

(Pauly, 1980).

Dispersal and movement

Dispersal for larvae was modelled using a combination of an advection-diffusion and a pelagic duration model, and thus larval recruitment is ultimately determined by oceanic currents obtained from respective ESMs, diffusivity, and pelagic larval duration (Sibert et al., 1999;

174

O’Connor et al., 2007; Cheung et al., 2008). Net adult migration is determined by: 1) the distance between two geographical gridded cells and their relative size (e.g. large-bodied pelagic species, small reef-dwelling demersal species), and 2) a fuzzy logic model based on differences in habitat suitability between neighbouring cells. Individuals will tend to emigrate if neighbouring cells have a more favourable habitat than their present cell, and more individuals will tend to immigrate if the present cell is preferable to surrounding neighbour cells.

B.2 Tables and figures

Appendix Table B.1. Sensitivity test of maximum body size to environmental change using varying gill-body size scaling coefficients.

† Change in maximum body size, W∞ (%) Gill-body size scaling coefficient, Temperature Acidity increase of Temperature + d No change increase of 5% 10% acidity 0.5 0 -4.3 -3.1 -7.2 0.6 0 -5.3 -3.8 -8.9 0.7 0 -7.0 -5.0 -11.7 0.8 0 -10.3 -7.5 -17.0 0.9 0 -19.6 -14.4 -31.1 †The example here assumes a baseline maximum body size of 10,000 g.

175

Dynamic Bioclimatic Envelope Model

Oceanic Species Organism conditions parameters physiology & OCEAN Life history ACIDIFICATION Primary productivity Productivity, mortality

Recruitment

Habitat Population suitability dynamics Adults Juveniles Species distribution

Appendix Figure B.1. Conceptual diagram of the dynamic bioclimatic envelope model.

For full figure, see online supplementary material Appendix Figure B.2.

Appendix Figure B.2. Initial species distribution maps and the abundance (tonnes) for the ten invertebrate species (Table 3.2) used in modeling simulations.

176

0.0 0.000 4 C) C) ° −0.1 ° 0.50

−0.025 ace

f 3 ace pH f −0.2 2 −0.050 0.25 sea bottom sea sur

−0.3 sea bottom pH sea sur

1 ∆ ∆ temperature ( ∆ temperature ( ∆ −0.075 −0.4 0 0.00

1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100

0.2 0 ) ) ) 1 1 1 − − − 0.0 0 ace O2 ear f litre litre −10 y y production ⋅ ⋅ r ⋅ −0.2 Low CO2 ima mol mol r µ µ sea bottom O2 sea sur ( ( −5 −20 (Pg C −0.4 High CO2 ∆ ∆ net p

∆ −0.6 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100

0.0 0.00 1.2 3 C) −0.05 C) ° −0.1 °

ace 0.8 f ace pH

f −0.10 2 −0.2 −0.15 0.4 sea bottom −0.3 sea sur 1 sea bottom pH sea sur ∆ ∆ temperature ( ∆ temperature ( ∆ −0.20 0.0 −0.4 0 −0.25 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100

0 0.0 ) ) ) 1 1 1

0 − − − −5 −0.1 ace O2 ear f litre litre y y production ⋅ ⋅ −10 r ⋅ −0.2 Low CO2 ima mol mol −10

−15 r µ µ sea bottom O2 sea sur ( (

(Pg C −0.3 High CO2 ∆ ∆ −20 net p −0.4 −20 ∆ 1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100

Year

177

For full figure, see online supplementary material Appendix Figure B.5.

Appendix Figure B.5. Projected changes in abundance (relative to 2000) for all ten species due to OA under different assumptions for the relationship between OA and changes in both life history parameters (growth and survival): linear, exponential, and shifting baseline. Results shown are for low and high CO2 scenarios (RCP 2.6 and RCP 8.5) using GFDL earth system model and abundances are presented as 10 year running means.

178

Appendix C – Supplementary material for “Ocean acidification amplifies multi-stressor impacts on global marine invertebrate fisheries.”

0.0 −0.35

−0.1

−0.40 −0.2

−0.3 −0.45 −0.4

4.5 4

Low CO2 3.5 3 High CO2

2.5 2 Earth system model 1.5 1 GFDL IPSL 0 0.5 MPI

0 0

−5 −10

−10 −20

−15 −30 1950 2000 2050 2100

Appendix Figure C.1. Projections of environmental surface conditions from three earth system models used to drive changes in species distribution and abundance through the impacts of ocean acidification (a and b), ocean warming (c and d), and reduced O2 (e and f). Panels on the left (a, c, e) show projected environmental changes by year 2100 (relative to 1950) in a high CO2 scenario (RCP 8.5) and show multi-model means from model simulations. Panels on the left (b, d, e) show projected environmental changes over time (relative to

1950) in a high (RCP 8.5) and low (RCP 2.6) CO2 scenario where bold lines represent multi-model means and faint lines are results from individual earth system model projections simulations: GFDL – Geophysical Fluid Dynamics Laboratory; MPI – Max Planck Institute; IPSL – Institute Pierre Simon Laplace.

179

West Arctic NE Pacifc Central Indo-Pacifc

0.0 0.0 0.0

−0.1 −0.1 −0.1

−0.2 −0.2 −0.2 −0.3 −0.3 −0.3 −0.4 −0.4 −0.5 −0.4

Low CO2 4 3 High CO2 4 3

2 2 Earth system 2 model

1 1 GFDL

0 0 IPSL 0 MPI

0.01 0.00 0.000

0.00

−0.01 −0.005 −0.01

−0.02 −0.010 −0.02

1950 2000 2050 2100 1950 2000 2050 2100 1950 2000 2050 2100 Year

Appendix Figure C.2. Projected environmental changes to surface conditions in a high (RCP 8.5) and low

(RCP 2.6) CO2 scenarios. Bold lines represent multi-model means and faint lines represent the individual earth system model projections from: GFDL – Geophysical Fluid Dynamics Laboratory; MPI – Max Planck Institute; IPSL – Institute Pierre Simon Laplace.

180

Scallops

Clams

Mussels

Oysters

Crabs Low CO2

Shrimps and prawns High CO2

Cephalopods

Abalones

Lobsters

All

−20 −10 0 10 Difference in ∆ MCP between OA scenarios (%)

181

East Bering Sea Gulf of Alaska California Current

0 0.0 0.0

−10 −2.5 −0.5

−20 −5.0 −1.0 −30

−7.5 −1.5 −40

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

Gulf of California Gulf of Mexico SE U.S. Continental Shelf

1 0 0 ios (%) r

−1 0 −1 Low CO2 −2

High CO2

MCP between scena −1 ∆ −3 −2

−2 erence in f −4 Dif

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

NE U.S. Continental Shelf Scotian Shelf Newfoundland−Labrador Shelf

0 0 0

−1 −2 −2

−2 −4 −4

−3

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

182

Insular Pacifc−Hawaiian Pacifc Central−American Coastal Caribbean Sea 0.5 0.0 0.0 0.0

−0.5 −0.5 −0.5

−1.0 −1.0 −1.5 −1.0

−1.5 −2.0 −1.5 −2.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.3 −0.2 −0.1 0.0 −0.3 −0.2 −0.1 0.0

Humboldt Current Patagonian Shelf South Brazil Shelf

0 0.0 0 ios (%) r

−2.5 −2 −1

Low CO2 −5.0 −4 −2 High CO2 MCP between scena ∆ −7.5 −6 −3 erence in f −10.0 −8 Dif −4 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

East Brazil Shelf North Brazil Shelf Canadian Eastern Arctic−West Greenland

0.0 0 0

−1 −1 −0.5 −2

−3 −2 −1.0 −4

−3 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

183

Greenland Sea Barents Sea Norwegian Sea 50 0.0

40 10

−2.5 30 5 20 −5.0

10 0 −7.5 0 −5 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

North Sea Baltic Sea Celtic−BiscayShelf

0 4000 ios (%) r 0

3000

−1 −10 Low CO 2000 2

High CO2 MCP between scena 1000 −2 ∆

−20 0 −3 erence in f

Dif −1000 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.6 −0.4 −0.2 0.0 −0.4 −0.3 −0.2 −0.1 0.0

Iberian Coastal Mediterranean Sea CanaryCurrent

0 0.0 0.0

−1 −0.5 −0.5 −2

−1.0 −3 −1.0

−1.5 −4 −1.5

−2.0 −5

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

184

Guinea Current Benguela Current Agulhas Current

0.5 0.3 1.0

0.5 0.0 0.0

0.0 −0.3 −0.5

−0.5 −0.6

−1.0 −1.0

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

Somali Coastal Current Arabian Sea Red Sea

0 0 ios (%) 0.00 r

−0.25 −1 −1 Low CO2

−0.50 High CO2 MCP between scena −2 ∆

−2 −0.75 erence in

f −3 Dif

−0.4 −0.3 −0.2 −0.1 0.0 −0.3 −0.2 −0.1 0.0 −0.3 −0.2 −0.1 0.0

BayofBengal Gulf of Thailand South China Sea

8 0 0

−1 −1 4

−2 −2 2

0 −3 −3

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

185

Sulu−Celebes Sea Indonesian Sea North Australian Shelf

0.0 0.0 0.00

−0.5 −0.25 −0.5

−1.0 −0.50

−1.5 −1.0 −0.75

−2.0 −1.00 −1.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

NE Australian Shelf East−Central Australian Shelf SE Australian Shelf

0 0.0 0 ios (%) r

−0.5 −1 −2 Low CO2 −1.0

High CO2

MCP between scena −2 ∆ −1.5 −4

erence in −3 f −2.0 Dif

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

SW Australian Shelf West−Central Australian Shelf NW Australian Shelf

0 0.5 0.0

0.0 −1 −0.5

−0.5 −1.0 −2

−1.0 −1.5

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

186

NewZealandShelf East China Sea Yellow Sea

0 0 0

−1 −1 −2

−2 −4 −2

−3 −6

−3 −4 −8 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

Kuroshio Current Sea of Japan Oyashio Current 10.0

ios (%) 0.0 r 7.5 0 −0.5

5.0 Low CO2 −1.0 −200 High CO2 MCP between scena 2.5 ∆ −1.5

−400 0.0 erence in f −2.0 Dif

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

Sea of Okhotsk West Bering Sea NorthernBering−Chukchi Seas

0 0 0

−10 −2

−10 −4 −20

−6 −30 −20

−8 −40

−0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

187

BeaufortSea East Siberian Sea Kara Sea 3e−04 0 0

2e−04 −10 −400

1e−04

−20 −800

0e+00

−0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.2 0.0

Iceland Shelf and Sea Faroe Plateau Antarctica

0 0

ios (%) 0 r

−1 −2 −10 Low CO −2 2

High CO2 MCP between scena −3 −4 ∆

−20 −4

erence in −6 f

Dif −5 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0

Black Sea Hudson BayComplex Aleutian Islands

0.0 0 1.0

−2.5 0.5 −2

−5.0 0.0 −4

−7.5 −0.5

−6

−0.4 −0.3 −0.2 −0.1 0.0 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.3 −0.2 −0.1 0.0 ∆ pH

188

Canadian High Arctic−North Greenland ios (%) r

Low CO2

−1 MCP between scena High CO2 ∆ erence in

f −2 Dif

−0.4 −0.2 0.0 ∆ pH

Appendix Figure C.4. Projected changes in maximum catch potential (MCP) and ocean surface pH in each individual large marine ecosystem. Impacts to MCP were calculated by finding the difference between model runs with and without modelled impacts ocean acidification. Multi-model mean projections are shown for the high (RCP 8.5) and low (RCP 2.6) CO2 scenario and smoothed by a 10-year running mean.

Appendix Table C.1. Effect sizes (with 95% confidence limits in parentheses) of OA impacts on life history traits (modified from Kroeker et al., 2013).

Growth† Survival† Mean Mean (Lower, Upper) (Lower, Upper) Crustaceans -16% -13% (-33, 6) (-25, 0)

Molluscs -17% -35% (-26, -10) (-56, -10)

†Effect sizes represent the percent change with a doubling of hydrogen ion concentration.

189

Appendix Table C.2. Pearson’s correlation coefficients (r)† between changes in abundance, range size, and st distributional shift of latitudinal centroid in a high CO2 scenario by the end of the 21 century (2091-2100) relative to the 1951-1960 period.

Global change impacts Added impacts of ocean excluding ocean acidification r acidification r Abundance x Range size 0.78* Abundance x Range size 0.60*

(CL = [0.72, 0.83]; (CL = [0.50, 0.68];

t = 18.05; df = 205) t = 10.62; df = 205)

Absolute abundance x Latitudinal shift 0.60* Abundance x Latitudinal shift 0.37*

(CL = [0.50, 0.68]; (CL = [0.25, 0.48];

t = 10.61; df = 205) t = 5.75; df = 205)

Absolute range size x Latitudinal shift 0.52* Range size x Latitudinal shift 0.58*

(CL = [0.42, 0.62]; (CL = [0.48, 0.66];

t = 8.83; df = 205) t = 10.23; df = 205)

*Correlation coefficient significant to the level of p<0.001.

†Two-tailed t-test for Pearson’s correlation coefficient: CL = correlation coefficient confidence limits; t = t-value; df = degrees of freedom. One species was removed as an outlier.

190

Appendix D – Supplementary material for “Evaluating present and future potential of

Arctic fisheries in Canada.”

D.1 Dynamic bioclimatic envelope model

Changes in temperature, oxygen, and pH were modelled in a multi-stressor framework to affect organism physiology and subsequent effects on life-history rates. The DBEM combines the oxygen- and capacity- limited thermal tolerance model (Pörtner & Lannig, 2009) and the gill- oxygen limitation model (Pauly & Cheung, 2017) to determine how stressors affect somatic growth and mortality rates (Tai et al., 2018). The model uses a derived equation of the von

Bertalanffy growth function (von Bertalanffy, 1951) to determine changes in growth rate (change in biomass, B, as a function of time, t):

!" = ��! − ��! (D.1) !" where H and k represent the coefficients for anabolism and catabolism, respectively. Growth rate is dependent on the available oxygen (anabolism) and oxygen demand for maintenance metabolism (catabolism). Anabolism scales with body weight (W) to the exponent d < 1, catabolism scales linearly with (W), i.e. b = 1. In this model, d = 0.7, although values typically range from 0.5 and 0.95. Sensitivity analyses showed that changes in temperature and acidity with low values of d (< 0.7) slightly decreases sensitivity, while larger values of d (>0.7) markedly increases sensitivity (Pauly & Cheung, 2017; Tai et al., 2018). The use of 0.7 is thus a conservative value as smaller values of d only marginally decrease sensitivity to multiple stressors and larger values of d only increases sensitivity.

191

!!! Next, solving for dB/dt = 0 when maximum body size (W∞) is reached results in � = ��! .

Integrating equation (D.1) into a generalized von Bertalanffy growth function:

! !!! !! !!!! �! = �! 1 − � (D.2) where K is the von Bertalanffy growth parameter where � = � 1 − � . The von Bertalanffy growth parameter K represents the rate at which maximum body size is reached.

Equations (3) and (4) show how temperature affects metabolism: H and k coefficients as a

!! ! function of the Arrhenius equation, � , where � = �! �, with �! and � equal to the activation energy and Boltzmann constant, respectively. Furthermore, oxygen availability and acidification affect aerobic scope changing oxygen supply (H; anabolism) and oxygen demand

(k; catabolism), respectively. Multi-stressor impacts were modelled following the most viable and parsimonious current working hypothesis (oxygen- and capacity- limited thermal tolerance hypothesis), to link physiological responses to life history traits. These effects can be modelled using the equations:

!!! ! � = � �! � (D.3) and

� = ℎ �! �!!! ! (D.4)

where constants �! and �!are equal to Ea/R where Ea (for anabolism and catabolism, respectively) and R are the activation energy and Boltzmann constant, respectively, while T is the absolute temperature (in Kelvin) (Cheung et al., 2011). The coefficients g and h from equations (3) and

192

(4), respectively, were derived for each species from the average W∞, K, and environmental temperature �! reported in the literature (Cheung et al., 2011):

!!! !! ! � = !! ! (D.5) !! ! ! and

ℎ = ! !!! (D.6) !! !!!! !

The DBEM links changes in aerobic scope to changes in the asymptotic weight (W∞) and von

Bertalanffy growth parameter K:

! !!! � = ! (D.7) ! ! and

� = � 1 − � (D.8)

Other parameters—e.g. asymptotic length and the length at maturity—that scale with weight can also be predicted (Beverton & Holt, 1959).

Natural population mortality rates (M) were estimated from the Pauly’s (Pauly, 1980a) empirical equation:

� = −0.4851 − 0.0824�� ! + 0.6757�� + 0.4687�� ′ (D.9)

193

where �′ is the average water temperature of a species range in degrees Celsius—other parameters are defined above. The DBEM uses this empirical model due to data availability and its widespread use for fish stock and assessments (Tai et al., 2018).

Population dynamics

Populations were modelled using an intrinsic population growth model (Hilborn & Walters,

1992; Cheung et al., 2008a):

!! �! = � ∙ �! ∙ 1 − (D.10) !"!

where Gi is the population growth in cell i, r is the intrinsic rate of population increase, A is the abundance, and KC is the carrying capacity. The change in abundance in each cell was modelled as:

!!! = ! � + � + � (D.11) !" !!! ! !" !" where L and I are the settled and net migrated adults from surrounding cells j into focal cell i.

Carrying capacity is modelled to change in subsequent time steps as a function of habitat suitability, P, with the equations (Cheung et al., 2008a):

!!!! ��!!! = ��! ∙ (D.12) !! and

� = � � ∙ � �� ∙ � � ∙ � �� (D.13)

194

where T, Dep, H, and Ice are corresponding temperature, depth, habitat type, and sea ice coverage for each cell, respectively.

D.2 Tables and figures

Appendix Table D.1. List of species modelled in the dynamic bioclimate envelope model.

Species name Common name Alepocephalus bairdii Baird's slickhead Amblyraja hyperborea Arctic skate Anarhichas denticulatus Northern wolffish Anarhichas lupus Atlantic wolffish Anarhichas minor Spotted wolffish Aphanopus carbo Black scabbardfish Arctica islandica Ocean quahog Argentina silus Greater argentine Atheresthes stomias Arrowtooth flounder Boreogadus saida Arctic cod Centroscyllium fabricii Black dogfish Chaceon quinquedens Red deepsea crab Chimaera monstrosa Rabbit fish Chionoecetes opilio Snow crab Chlamys islandica Iceland scallop Ciliata mustela Fivebeard rockling Clupea pallasii pallasii Pacific herring Conger conger European conger Coryphaenoides rupestris Roundnose grenadier Cyclopterus lumpus Lumpfish Eleginus nawaga Navaga Enchelyopus cimbrius Fourbeard rockling Etmopterus spinax Velvet belly Eutrigla gurnardus Grey gurnard Gadus morhua Atlantic cod Gadus ogac Greenland cod Galeus melastomus Blackmouth catshark Gobius niger Black goby Helicolenus dactylopterus Blackbelly rosefish Hippoglossoides elassodon Flathead sole Hippoglossoides platessoides American plaice Hippoglossus hippoglossus Atlantic halibut

195

Illex illecebrosus Northern shortfin squid Leucoraja circularis Sandy ray Leucoraja fullonica Shagreen ray Leucoraja naevus Cuckoo ray Limanda aspera Yellowfin sole Liza saliens Leaping mullet Macrourus berglax Roughhead grenadier Mactromeris polynyma Arctic surfclam Mallotus villosus Capelin Meganyctiphanes norvegica Norwegian krill Melanogrammus aeglefinus Haddock Microstomus kitt Lemon sole Microstomus pacificus Dover sole Molva dypterygia Blue ling Mustelus asterias Starry smooth-hound Mya arenaria Sand gaper Mytilus edulis Blue mussel Nezumia aequalis Common Atlantic grenadier Oncorhynchus gorbuscha Pink salmon Ophiodon elongatus Lingcod Osmerus mordax mordax Rainbow smelt Pandalus borealis Northern prawn Pandalus montagui Aesop shrimp Petromyzon marinus Sea lamprey Platichthys stellatus Starry flounder Pleuronectes platessa European plaice Pleuronectes quadrituberculatus Alaska plaice Pollachius virens Saithe Raja brachyura Blonde ray Rajella lintea Sailray Reinhardtius hippoglossoides Greenland halibut Salmo salar Atlantic salmon Salvelinus alpinus alpinus Arctic char Scophthalmus rhombus Brill Scyliorhinus stellaris Nursehound Sebastes flavidus Yellowtail rockfish Sebastes mentella Beaked redfish Sebastes norvegicus Golden redfish Solen vagina European razor clam Trigla lyra Piper gurnard

196

1500 1500

1250 1250

RCP 2.6 1000 1000 RCP 8.5 alue (million USD) v 750 750 Catch (thousand tonnes)

Landed 500 2000 2025 2050 2075 2100 2000 2025 2050 2075 2100 Year

Appendix Figure D.1. Projected maximum sustainable catch and landed value potential in all four of Canada’s Arctic Large Marine Ecosystems tested: Canada Eastern Arctic – West Greenland; Beaufort Sea; Hudson Bay complex; and Canada high-Arctic – North Greenland. Thin lines represent each model simulation using the different earth systems models, while bold lines are multi models means. Data are smoothed using a 10-year running mean.

197

100

1200 75

1000 50 All 4 LMEs

800 25

600 0

600 100

500 75

400 Canada Eastern 50 Arctic - West Greenland 300 25 200

100 50

tainty (%) Source of

r 75 40 uncertainty

30 50 Beaufort Sea Model

20 Scenario 25

Catch (thousand tonnes) 10 ercentage of total unce

P 0

100

600 75

50 Hudson Bay 500 25

400 0

100

6 75

High Arctic 4 50 Archipelago

25 2

0 2000 2025 2050 2075 2100 2000 2025 2050 2075 2100 Year Year

198

uncertainty. Model uncertainty addresses the different earth system models, and scenario uncertainty addresses the different representative concentration pathway scenarios (RCP 2.6 and RCP 8.5).

199