UNDERSTANDING SUSTAINABLE DEVELOPMENT PROGRESS IN A METACOUPLED WORLD By Yuqian Zhang A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Fisheries and Wildlife – Doctor of Philosophy Environmental Science and Policy—Dual Major 2024 ABSTRACT With industrialization and human development over the past centuries, one of the primary challenges to humans is the global biodiversity loss at a massively accelerated rate. The United Nations (UN) has adopted the 17 Sustainable Development Goals (SDGs), aiming to provide human welfare and conserve the planet, now and into the future. Two of the SDGs directly address biodiversity conservation and sustainable development – SDG 14 (life below water) and SDG 15 (life on land). Although the UN has issued annual reports on SDGs, the reports do not consistently reveal the progress over time, because of inconsistent methods such as estimation based on different indicators across years. Besides the lack of a consistent assessment of integrated efforts (e.g., SDGs 14 and 15) in biodiversity conservation and sustainable development, the other challenge for conservation science is to identify key drivers for the socioecological changes and achieve environmental and socioeconomic sustainability within and across boundaries. The main objective of this dissertation is to fill the knowledge gaps by providing a consistent assessment of SDGs 14 and 15 over time (Chapter 2), exploring the key drivers for socioecological changes (Chapter 3), and conducting scenario analysis through the metacoupling framework and modeling approaches (Chapter 4). This dissertation would better inform countries to review their sustainable development progress associated with Life below Water and Life on Land and empower decision- makers with support for future conservation planning and sustainable development. The open- source database would contribute to future research in biodiversity conservation, sustainability science, and other disciplines. The methodology used in this study can also be generalized and contribute to the broader scientific community and beyond. To my dearest parents, grandparents, and myself iii ACKNOWLEDGEMENTS The journey of completing a PhD and writing a dissertation was full of dreams, adventures, challenges, happiness, and achievements, which cannot be completed alone. I am so lucky to have received so much support, company, and enlightenment over the past six years from so many great minds and souls that guided me through the long winters in Michigan and dark nights of my life. First and foremost, I want to thank my parents, Ying Feng and Liangmin Zhang, for their unconditional love and wholehearted support that made me feel safe to explore the world. As a first-generation international student, I am grateful for all the encouragement and respect that my parents offered me for making decisions, following my heart, and being who I am confidently. Second, I would like to thank my advisor, Jack Liu, for being the best model of a diligent, humble, inspiring, and serious scientist. I always appreciate Jack for training my scientific instinct and curiosity, teaching me how to work independently and collaboratively, providing me with the great platform and opportunities in lab, center, and conference to learn and grow, and passing down the scientific passion, joy, and inspiration to me like he received from greatest scholars of previous generations. All these served where I am now and will be carried forward into the future. I would also like to thank my committee, Drs. Tom Dietz, Dan Kramer, and Laura Schmitt Olabisi, for introducing me to the modeling world, teaching me academic writing and science communication, mentoring me in career development, and demonstrating community research. Many thanks to my friends, supporting groups, mentors, and colleagues in the lab, Center for Systems Integration and Sustainability, my home Department of Fisheries and Wildlife, Environmental Science and Policy Program, Sustainable Michigan Endowed Project, Michigan State University, Indiana University Bloomington, and beyond. iv Finally, I want to thank my funders. My PhD was supported by National Science Foundation, MSU AgBioResearch, Environmental Science and Policy Program Doctoral Recruiting Fellowship, Sustainable Michigan Endowed Project Scholarship, William W. and Evelyn M. Taylor Endowed Fellowship, Kellogg Biological Station Long Term Ecological Research Network Graduate Fellowship, MSU Dissertation Completion Fellowship, and NASA- MSU Professional Enhancement Award. I am also grateful for additional training support from travel grants through Environmental Science and Policy Program, Department of Fisheries and Wildlife, College of Agriculture and Natural Resources, and Graduate School. v PREFACE The chapters in this dissertation were conceptualized as individual research projects under the primary theme and common goals. While the chapters principally represent my own work, I use the pronoun we throughout the dissertation as an acknowledgement for the contributions of my collaborators. I am deeply honored to work with them and genuinely grateful for their contribution, enlightenment, and guidance along this journey. Without them, none of this would have been possible. vi TABLE OF CONTENTS CHAPTER 1: INTRODUCTION ................................................................................................... 1 1.1 Background ........................................................................................................................... 1 1.2 Theoretical framework .......................................................................................................... 2 1.3 Objectives and research questions......................................................................................... 4 CHAPTER 2: GLOBAL DECADAL ASSESSMENT OF LIFE BELOW WATER AND ON LAND ............................................................................................................................................. 6 2.1 Abstract ................................................................................................................................. 6 2.2 Summary ............................................................................................................................... 6 CHAPTER 3: ANALYZING GLOBAL THREATS AND OPPORTUNITIES FOR LIFE BELOW WATER AND ON LAND............................................................................................... 8 3.1 Abstract ................................................................................................................................. 8 3.2 Introduction ........................................................................................................................... 8 3.3 Methodology ....................................................................................................................... 12 3.4 Results ................................................................................................................................. 19 3.5 Discussion ........................................................................................................................... 29 3.6 Conclusions ......................................................................................................................... 34 CHAPTER 4: SUSTAINING LIFE ON LAND THROUGH A METACOUPLING APPROACH: SIMULATING SPAIN’S SDG 15 PROGRESS ................................................... 36 4.1 Abstract ............................................................................................................................... 36 4.2 Introduction ......................................................................................................................... 37 4.3 Methodology ....................................................................................................................... 38 4.4 Results ................................................................................................................................. 47 4.5 Discussion ........................................................................................................................... 61 CHAPTER 5: SYNTHESIS .......................................................................................................... 65 REFERENCES ............................................................................................................................. 67 APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 2 ......................................... 72 APPENDIX B SUPPORTING INFORMATION FOR CHAPTER 3 ......................................... 73 APPENDIX C SUPPORTING INFORMATION FOR CHAPTER 4 ....................................... 107 vii CHAPTER 1: INTRODUCTION 1.1 Background With industrialization and human development over the past centuries, one of the primary challenges to humans is the global biodiversity loss at a massively accelerated rate (Mace et al. 2005, Rockström et al. 2009). The United Nations (UN) has called for sustainable development and adopted the 17 Sustainable Development Goals (SDGs), aiming to provide human welfare and conserve the planet, now and into the future. Two of the SDGs take the initiative for an integrative assessment of biodiversity conservation efforts and economic development – SDG 14 (Life below Water) and SDG 15 (Life on Land). This initiative appears hopeful to fill the current gap of estimating conservation efforts and economic development separately. Although annual reports were produced by the United Nations to inform how sustainable development progress is being made on a global scale, those annual assessments were considered problematic, because the assessed values and indicators selections were inconsistent from one year to the other (Xu et al. 2020). Therefore, this dissertation aims to evaluate global SDGs 14 and 15 progress over time, identify countries that have high and low SDG scores, and explore the drivers for countries’ SDGs 14 and 15 score variation. It is challenging for conservation science to achieve environmental and socioeconomic sustainability within and across boundaries due to the complex system dynamics (interactions among system components, emergent behavior, etc.). To advance the knowledge of complex socio-environmental interactions within and across systems, this dissertation applies the metacoupling framework (Liu 2017) and uses System Dynamics to simulate the complex system interactions and processes. 1 The outcomes of this dissertation (1) fill the current knowledge gap in the SDGs 14 and 15 assessments at a global scale, (2) identify countries that did better or worse in SDGs 14 and 15, (3) provide potential explanations that drive the SDGs score variation, and (4) discover the impact of endogenous and exogenous environmental and social variables on SDG 15. This dissertation hopes to better inform countries on how to review their sustainable development progress associated with Life below Water and Life on Land and empowers decision-makers with support for future conservation planning and sustainable development. The methodology used in this study can also be generalized and contribute to the broader scientific community and beyond. 1.2 Theoretical framework The metacoupling framework (Liu 2017) is a powerful tool for understanding the complex system interactions within and across different scales and borders. Three types of human-nature interactions (couplings) are delineated under the complete metacoupling framework (Figure 1.1): (1) within a coupled system (intracoupling), (2) between distant coupled systems (telecoupling), and (3) between adjacent coupled systems (pericoupling). 2 Adapted from Liu 2017 Figure 1.1. Three categories of the conceptual metacoupling framework – intracoupling, telecoupling, and pericoupling (Liu, 2017). Systems can be defined as sending, receiving, and/or spillover systems depending on the directional movement of flows. Within each system, causes, agents, and effects are included for analysis. Between systems, there are direct or indirect flows (e.g., material, money, information). 3 Adapted from Liu et al. 2013 Figure 1.2. Sending, receiving, spillover systems and major system components under the metacoupling framework (Liu et al., 2013). Within each system, causes and effects are interrelated through agents. Between systems, flows of directional movement (e.g., materials, energy, and information) influence system interactions. 1.3 Objectives and research questions Chapter 2: Global Decadal Assessment of Life below Water and on Land Research questions: (1) How had sustainable development in terms of life below water and on land progressed, as measured in SDGs 14 and 15? (2) How did the SDG scores change before and after the adoption of SDGs in 2015? (3) Which countries had high or low SDG scores? (4) Which countries experienced drastic changes (increase or decrease) in SDG scores? This chapter evaluates countries’ SDGs 14 and 15 scores (at goal and target levels) between 2010 and 2020, based on the indicator selection and guidance from the United Nations. I also compare countries’ SDG progress before and after 2015 (when SDGs were adopted by United Nations member states). 4 Chapter 3: Analyzing Global Threats and Opportunities for Life below Water and on Land Research questions: (1) What drives the sustainable development progress variation among countries, in terms of the SDGs 14 and 15 measurements? (2) How different are the drivers for different groups of countries (e.g., income level, biodiversity hotspot)? (3) Are there any synergies and trade-offs between SDGs and their Targets? This chapter uses multivariate regressions with regularization techniques to explore the drivers for countries’ SDG variation. Several environmental and social variables are used for analysis. The data are either from publicly available databases or from the previous chapter. Chapter 4: Sustaining Life on Land through a Metacoupling Approach: Simulating Spain’s SDG 15 Progress Research questions: (1) How does the change of forest area impact a country’s SDG 15 progress? (2) What parameter has the largest impact on a country’s SDG 15 progress? This chapter frames the interactions among forest, land transformation, population, and SDG 15 with the metacoupling framework, then applies the system dynamics model (SDM) to simulate the stocks (e.g., forest area, population) change over the interactions. The data are either from publicly available databases or from the previous chapters. 5 CHAPTER 2: GLOBAL DECADAL ASSESSMENT OF LIFE BELOW WATER AND ON 2.1 Abstract LAND The United Nations (UN) has adopted the 17 Sustainable Development Goals (SDGs), aiming to provide human welfare and conserve the planet, now and into the future. Two of the SDGs directly address biodiversity conservation and sustainable development – SDG 14 (life below water) and SDG 15 (life on land). Although the UN has issued annual reports on SDGs, the reports did not consistently reveal the progress over time, because of inconsistent methods such as estimation based on different indicators across years. Our research examined the dynamics of the same 10 indicators for SDGs 14 and 15 between 2010 and 2020. Results indicate that the overall SDG 14 scores had a small growth between 2010 and 2020, whereas the substantial increase in SDG 15 scores spotlighted the conservation efforts and sustainable use of terrestrial ecosystem services, especially in countries with biodiversity hotspots. Globally, there was more progress in terms of SDG 15 scores during 2015–2020 than during 2010–2015 (before the UN adopted SDGs in 2015). Surprisingly, SDG 14 score had smaller progress during 2015–2020 than during 2010– 2015. Special attention should be given to low-income countries lagging in sustainable development performance when implementing the post-2020 global biodiversity framework. 2.2 Summary In this chapter, I evaluated countries’ SDGs 14 and 15 performances between 2010 and 2020, based on the indicator selection and guidance from the United Nations. This delineates how countries did in SDGs 14 and 15 over the past decade, and that through comparisons, which countries did well or poorly. This evaluation step fills the current knowledge gap at a global scale of estimating conservation efforts and economic development separately, and it also provides 6 significant data for the following chapters. With collaborative efforts, I designed the research, collected raw data, performed data analysis, interpreted the results, and wrote the chapter. This chapter has been published in an open-access journal with details below. Material from: Zhang, Y., Li, Y., & Liu, J. (2023). Global decadal assessment of life below water and on land. Iscience, 26(4). For the full text of this work, please go to: https://doi.org/10.1016/j.isci.2023.106420 7 CHAPTER 3: ANALYZING GLOBAL THREATS AND OPPORTUNITIES FOR LIFE BELOW WATER AND ON LAND 3.1 Abstract Anthropogenic activities have increasingly altered the environment and challenged global socioecological sustainability. Two of the 17 Sustainable Development Goals (SDGs) – SDGs 14 (life below water) and 15 (life on land) - aim to conserve biodiversity and sustainably use natural resources for sustainable development. Countries have achieved significant positive progress in SDGs 14 and 15 in the past decade at different rates. But what drives or impedes countries’ SDG progress remains unknown. Here, we identified key factors that directly and indirectly affect countries’ SDG 14 (52 countries) and 15 (143 countries) progress between 2010 and 2020. Our results demonstrate mixed expected and unexpected impacts of multiple drivers on SDG progress for countries across different income and biodiversity hotspot groups. Fish Production has the most profound negative impact on SDG 14 progress, and the impact on SDG 15 progress for countries of different income levels and biodiversity hotspot status varied substantially among drivers such as Agricultural Land Percentage, Forestry Import, Forestry Production, Forest Rents in GDP Percentage, and Political Stability. Synergies and trade-offs between SDGs and their Targets call for special attention for policy making to maximize the common benefits of multiple socioecological sustainability goals while minimizing the conflicting interests. Incorporating the significant direct and indirect drivers for SDG progress in future planning is imperative as the deadline for the 2030 agenda approaches. 3.2 Introduction In 2015, the United Nations Member States adopted 17 Sustainable Development Goals (SDGs) and aimed to address big sustainability challenges globally. Two of the 17 SDGs directly 8 aim to prevent biodiversity loss and buttress sustainable natural resources management: SDG 14 (life below water) – Conserve and sustainably use the oceans, seas and marine resources for sustainable development and SDG 15 (life on land) – Protect, restore and promote sustainable use of terrestrial ecosystems, sustainably manage forests, combat desertification, and halt and reverse land degradation and halt biodiversity loss. The progress of SDGs 14 and 15 revealed countries’ integrated efforts in biodiversity conservation and socioeconomic sustainable development. Countries have achieved significant positive progress in SDGs 14 and 15 between 2010 and 2020 at different paces (Zhang et al., 2023); however, what drives countries’ SDG progress and what factors affect the rate of progress remain yet unclear. Studies have shown the drivers for environmental stress and sustainability from the disciplines of sociology, economics, geography, etc. and interdisciplinary perspectives over past decades (Stern et al., 1992, Dietz and Rosa, 1994, Dietz et al., 2015, Dietz, 2017, Jorgenson et al., 2019). The spirit of SDGs is to achieve both environmental and socioeconomic sustainability; therefore, it is important to investigate socioecological stressors for environment and human well- being. But the literature that explained the drivers for the SDG progress is rather limited, nor does an exhaustive theory exist. Earlier studies have explored the linkage between environmental impact and population, affluence, and technology (Stern et al., 1992, Dietz and Rosa, 1994, Ehrlich and Holdren, 1971), which structured the debate about the effects of population, affluence and technology on the environment and provided a simple and robust framework for broader study references. Recent research has examined additional factors such as social dimensions of economic system, power, social stratification, inequality, and governance impact on global climate change (Jorgenson et al., 2019). The study of direct and indirect drivers (Díaz et al., 2019) further investigates their environmental impacts on terrestrial, freshwater, and marine ecosystems. 9 Important direct factors include land/sea use change, direct exploitation, climate change, pollution, invasive alien species; indirect factors are grouped into four categories: demographic and sociocultural, economic and technological, institutions and governance, conflicts and epidemics (Díaz et al., 2019). This provides guidelines for research on environmental impact and sheds light on studying the drivers for SDG progress. The metacoupling framework (Liu, 2017) that helps understand environmental and socioeconomic interactions within and across adjacent and distant systems is also useful to identify important natural and social, internal and external variables and map the interactions among them within and across systems (Wu et al., 2021, Chung and Liu, 2022). The metacoupling framework is more general and broader than the world systems framework that has sometimes been used to explain differences across countries in stress placed on the environment (Burns et al., 1994, Burns et al., 2003, Jorgenson and Givens, 2013). A major barrier to social scientific inquiry into the human–environment relationship is the difficulty in selecting appropriate analytic techniques and models that allow for a precise specification of the functional form of the relationship between driving forces and environmental impacts (York et al., 2003). Although linear regression is a simple, interpretable, and useful tool to estimate the direct and indirect drivers’ impact on SDG progress, it is limited by the knowledge of specification and data availability, and the misspecification of regression can lead to biased, inconsistent, inefficient, and misleading predictions (Dewey et al., 2000). To reduce the number of irrelevant variables while balancing the explaining power of the regression model, the Least Absolute Shrinkage and Selection Operator (LASSO) (Tibshirani, 1996) regression is a machine learning process to regulate the number of variables by adding a penalty term to the traditional regression model and shrinking some coefficients towards zero. Studies that used the LASSO regression for variable selection have yielded interpretable models by selecting appropriate 10 variables and reducing the risk of overfitting (Muthukrishnan and Rohini, 2016, Shortreed and Ertefaie, 2017, Wang et al., 2018). LASSO regression is useful as an exploratory method and a parsimonious model, but it may produce spurious conclusions if interpreted causally without care. The Environmental Kuznets Curve (Kuznets, 2019, Grossman and Krueger 1991) has shown that income differences among countries could lead to different patterns of energy use, economic growth, and the environmental outcomes (Stern, 2004, Leal and Marques, 2022). Earlier research has observed that countries of different income levels and biodiversity hotspot status performed significantly differently in terms of SDG 14 and 15 progress (Zhang et al., 2023). To prevent capturing only the average impact and to draw policy implications that are salient for specific countries, in this article we studied different drivers’ impact on SDG progress by allowing interactions of countries’ income level and biodiversity hotspot status with other independent variables. In particular, we addressed the following questions: (1) What drives the sustainable development progress variation among countries, in terms of the SDGs 14 and 15 measurements? (2) How different are the drivers for different groups of countries (e.g., income level, biodiversity hotspot)? (3) Are there any synergies and trade-offs between SDGs and their Targets? We first selected drivers for SDG progress analysis based on the inclusion of relevant direct and indirect drivers (Díaz et al., 2019) with the best available data for the study period between 2010 and 2020: 21 independent variables for SDG 14 among 52 countries/regions and 25 variables for SDG 15 among 143 countries/regions. Then interaction terms were generated based on countries’ income level and biodiversity hotspot status, which expanded to 40 independent variables for SDG 14 (19 high-income countries, 33 low-income countries) and 71 for SDG 15 (53 biodiversity-hotspot countries, and 90 non-hotspot countries. See Methods section for details about income level and biodiversity hotspot status classification for SDGs 14 and 15). We utilized the 11 LASSO technique to reduce the number of irrelevant variables and establish reliable statistical inferences. Besides SDG progress at the Goal level, we regressed the drivers against each SDG Target, and analyzed 3 Targets under SDG 14 and 6 Targets under SDG 15. Finally, we compared the multiple regression results and scrutinized the synergies and trade-offs between SDGs and Targets. 3.3 Methodology 3.3.1 Selection of drivers for SDG score change The goal of this study is to find drivers for SDGs 14 and 15 score change and analyze their impact as completely as possible. Studies have shown that land/sea use change, direct exploitation, climate change, pollution, and invasive alien species were considered the direct drivers for terrestrial, freshwater, and marine ecosystem change (Díaz et al., 2019, Didham et al., 2005, Nelson, 2005, Nelson et al., 2006). Other indirect drivers that may cause those social and ecosystem changes were categorized as demographic and sociocultural (e.g., population size and growth, age distribution), economic and technological (e.g., economic growth, consumption), institutions and governance (e.g., rule of laws, governance performance), and conflicts and epidemics (Díaz et al., 2019, Didham et al., 2005, Nelson, 2005). Based on the metacoupling framework, we developed a conceptual framework of drivers and effects between natural and human systems to understand the relationship between direct and indirect drivers for SDG progress within and across countries (Figure 3.1). To be inclusive whilst relatable to SDG score change with data limitation, we first included 21 variables for SDG 14 (Control of corruption index, Crops and animals export, Crops and animals import, Fish export, Fish import, Fish production, GDP, Government effectiveness index, Political stability index, Population density, Population growth rate, Total population, Regulatory quality index, Rule of law index, Temperature change, Tourist 12 number, Voice and accountability index, Population ages between 0 and 14, Population ages between 15 and 64, Population ages over 65, GDP per capita), and 25 variables for SDG 15 (Agricultural land percentage of total land area, Agricultural land square kilometer, Control of corruption index, Crops and animals export, Crops and animals import, Forest area in square kilometer, Forest Rents percentage of GDP, Forest export, Forest import, Forestry production, GDP, Government effectiveness index, Political stability index, Air Pollution of PM 2.5, Population density, Population growth rate, Total population, Regulatory quality index, Rule of law index, Temperature change, Voice and accountability index, Population ages between 0 and 14, Population ages between 15 and 64, Population ages over 65, GDP per capita). Figure 3.1. Direct and indirect causes of natural and human elements for SDG progress. The internal causes are natural processes and human activities within a country. The effect of the focal country (e.g., country X) could be impacted by its neighboring (country Y) and distant (country Z) countries through trade, which is considered as an external cause. 13 This study period was between 2010 and 2020, with a coverage of 52 countries/regions for SDG 14 (in 2011, 2013, 2015, 2017, and 2019) and 143 for SGD 15 (annually from 2010 to 2019) analysis (Figure 3.2). We used the SDGs 14 and 15 scores from a published database (Zhang et al., 2023), and we collected the independent variable data from publicly available sources including the World Bank Group (Worldwide Governance Indicators, World Development Indicators), World Health Organization, and United Nations Food and Agriculture Organization (FAO Statistics and Climate). 14 A B Figure 3.2. Countries’ spatial distribution by (A) income level for SDG 14 analysis, (B) income level and biodiversity hotspot status for SDG 15 analysis. 3.3.2 Regression form specification from STIRPAT and empirical observation We used the ordinary least square regression model to analyze the impact of drivers for SDG score change. To minimize the residual square of error term and make the estimated impact (coefficient) of drivers (independent variable) on SDG score (dependent variable) change comparable, we normalized each independent variable with scale function in R (Becker et al., 1988). 15 𝑋𝑠𝑐𝑎𝑙𝑒𝑑 = (𝑋𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 − 𝑋̅) ⁄ 𝑆 Where 𝑋𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 is the original X value, 𝑋̅ is the sample mean, and 𝑆 is the sample standard deviation. We did not use this method to standardize SDG score (dependent variable) because they had been normalized in sourced data (ranging from 0 to 100). To determine the appropriate specification form of variables included in the model, we plotted each independent variable against SDG 14 and 15 scores (dependent variable) separately (Figures A3.2 and A3.4]. This step provided empirical evidence besides theories of anthropogenic impacts on the environment such as the Stochastic Impacts by Regression Population, Affluence and Technology (STIRPAT) (Dietz and Rosa, 1994, York et al., 2003) of determining the appropriate form (e.g., original form, log form) of each independent variable in the regression. We kept the following variables in the original form: for SDG 14, they are Control of corruption index, Government effectiveness index, Political stability index, Population growth rate, Regulatory quality index, Rule of law index, Temperature change, Voice and accountability index, GDP per capita; and for SDG 15, they are Agricultural land percentage of total land area, Control of corruption index, Forest Rents percentage of GDP, Forest export, Forest import, Forestry production, Government effectiveness index, Political stability index, Population density, Population growth rate, Regulatory quality index, Rule of law index, Temperature change, Voice and accountability index, GDP per capita. We converted the following variables into the log form: for SDG 14, they are Crops and animals export, Crops and animals import, Fish export, Fish import, Fish production, GDP, Population density, Total population, Tourist number, Population ages between 0 and 14, Population ages between 15 and 64, Population ages over 65; for SDG 15, they are Agricultural land square kilometer, Crops and animals export, Crops and animals import, 16 Forest area in square kilometer, GDP, Air Pollution of PM 2.5, Total population, Population ages between 0 and 14, Population ages between 15 and 64, Population ages over 65. Because some independent variables explained SDG score change differently across income levels and/or biodiversity hotspots, we created additional interaction terms of high-income * independent variable in the regression for SDG 14; for SDG 15, we added biodiversity-hotspot * independent variable interaction terms besides the income level [Supplementary Methods]. The classification of countries into high/low-income and biodiversity/non-biodiversity hotspots (Figure 3.2) was adapted from the sourced SDG 14 and 15 data (Zhang et al., 2023, Chung and Liu, 2022). For SDG 14, countries with more than $12,696 gross national income per capita (World Bank Country and Leading Groups, 2021) were categorized as high-income countries (n=19); otherwise, they were low-income countries (n=33). For SDG 15, countries identified as high biodiversity hotspots in the literature (Zhang et al., 2023, Chung and Liu, 2022) were categorized as biodiversity-hotspot countries (n=53); otherwise, they were non-hotspot countries (n=90) in this study. The addition of interaction terms effectively differentiated the impact of several variables on SDG scores when countries were in different income and biodiversity groups. 3.3.3 LASSO regression model building With all those interaction terms included in the Ordinary Least Square regression, 40 independent variables were analyzed for SDG 14, and 71 for SDG 15. To reduce the number of irrelevant variables while balancing the explaining power of the regression model, we applied the machine learning regression shrinkage and selection approach via the LASSO regularization technique (Tibshirani,1996) to eliminate those statistically insignificant variables (Figures A3.5 and A3.6). The LASSO regression is intended to balance model simplicity and accuracy, by adding a penalty term to the traditional linear regression model and shrinking some coefficients towards 17 zero. The LASSO regression provides an interpretable model and reduces the risk of overfitting. Studies that used the LASSO regression for variable selection while comparing with other selection approaches have shown the effectiveness in selecting appropriate variables (Muthukrishnan and Rohini, 2016, Shortreed and Ertefaie, 2017, Wang et al., 2018). However, the limitation of the LASSO regression as a variable selector is that when there exist dependence structures among variables (Freijeiro‐González et al. 2022), the model did not fully resolve multicollinearity issues in the regression. Therefore, we manually removed all variables with generalized variation inflation factor (GVIF) larger than 5 (Fox and Monette, 1992, O’brien, 2007): we only removed one variable at a time when the variable had the highest GVIF, while we observed the adjusted R-square value change and the GVIFs for other independent variables. After several iterations of eliminating variables, we concluded the final regression model. The coefficient values across different independent variables (shown in Table 3.1 and Table 3.2) were comparable because of data normalization in previous steps. 3.3.4 Regression at Target level Each SDG has several Targets, and those Targets are closely linked to the SDG and can be used as subgoals to quantify and measure the SDG progress. After we had the regressions for SDGs 14 and 15 at a Goal level, we used the same independent variables to regress against each SDG Target score. This allowed us to detect potential similar and different estimates of variables at a Target level from a Goal level and examine synergies and trade-offs between SDGs and their Targets. For SDG 14, the Targets are 14.1 (By 2025, prevent and significantly reduce marine pollution of all kinds, in particular from land-based activities, including marine debris and nutrient pollution), 14.5 (By 2020, conserve at least 10 per cent of coastal and marine areas, consistent with national and international law and based on the best available scientific information), and 14.7 (By 18 2030, increase the economic benefits to small island developing States and least developed countries from the sustainable use of marine resources, including through sustainable management of fisheries, aquaculture and tourism). For SDG 15, the Targets are 15.1 (By 2020, ensure the conservation, restoration and sustainable use of terrestrial and inland freshwater ecosystems and their services, in particular forests, wetlands, mountains and drylands, in line with obligations under international agreements), 15.4 (By 2030, ensure the conservation of mountain ecosystems, including their biodiversity, in order to enhance their capacity to provide benefits that are essential for sustainable development), 15.5 (Take urgent and significant action to reduce the degradation of natural habitats, halt the loss of biodiversity and, by 2020, protect and prevent the extinction of threatened species), 15.6 (Promote fair and equitable sharing of the benefits arising from the utilization of genetic resources and promote appropriate access to such resources, as internationally agreed), 15.8 (By 2020, introduce measures to prevent the introduction and significantly reduce the impact of invasive alien species on land and water ecosystems and control or eradicate the priority species), and 15.9 (By 2020, integrate ecosystem and biodiversity values into national and local planning, development processes, poverty reduction strategies and accounts). Different Targets were explained differently by those independent variables across income levels and biodiversity hotspot statuses. 3.4 Results 3.4.1 Drivers for SDGs 14 and 15 at a Goal level After controlling the multicollinearity in the regression, 11 variables are used in the regression for SDG 14 including 1 interaction term. Nine out of the 11 variables are statistically significant with the significance level of p < 0.05. Two variables are not significant, one of which is the interaction term (Political Stability * High Income) meaning there is no difference in the 19 effect of Political Stability between high-income and low-income countries for its impact on SDG 14 progress. The significant variables are relevant to climate change, direct exploitation, institutions and governance, and demographic and sociocultural, economic, and technological pathways (Figure 3.3.A). Among the 9 significant variables, Fish Export (in the log form) has the most important positive role contributing to SDG 14 progress, with the estimate of 0.58, followed by Tourist (in the log form, 0.4) and GDP per Capita (0.23). Fish Production (in the log form) has the most important negative role dragging the SDG 14 progress, with an estimate of -0.51. Many other variables also have negative impacts on SDG 14 progress, such as Fish Import (in the log form, - 0.26), Political Stability (-0.25), Population Density (in the log form, -0.18), and Population Growth Rate (-0.12). Both direct and indirect drivers have impacts on SGD 14 progress, but the average effect that direct drivers have is negative, while the effect of indirect drivers is positive. By summing the coefficient estimates of direct and indirect drivers respectively, the direct sum is -0.36 and the indirect sum is 0.4 (Figure 3.4.A). 20 A Figure 3.3. Statistically significant drivers’ impact on SDGs. (A) Impact on SDG 14 for both high- and low-income countries. (B) Impact on SDG 15 for high-income, biodiversity-hotspot [HB] countries. (C) Impact on SDG 15 for high-income, no-biodiversity-hotspot [HN] countries. (D) Impact on SDG 15 for low-income, biodiversity-hotspot [LB] countries. (E) Impact on SDG 15 for low-income, no-biodiversity-hotspot [LN] countries. (F) Drivers differentiate impact on SDG 15 across country groups. 21 Figure 3.3 (cont’d) B D C E 22 Figure 3.3 (cont’d) F 23 A Direct and Indirect Drivers' Impact on SDG 14 Direct driver Indirect driver High income 0.4 Low income 0.4 -0.36 -0.36 Direct and Indirect Drivers' Impact on SDG 15 Direct driver Indirect driver 0.42 0.35 0.44 0.31 0.26 0.21 0.22 0.23 e t a m i t s e m u S 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 B e t a m i t s e m u S 0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 High-income, Biodiversity Hotspot High-income, Non- hotspot Low-income, Biodiversity-hotspot Low-income, Non- hotspot Figure 3.4. Sum estimates of direct and indirect drivers’ impact on SDGs. (A) Impact on SDG 14 for both high- and low-income countries. (B) Impact on SDG 15 for countries of different income levels and biodiversity hotspot status. For SDG 15 regression, 24 variables are analyzed in the model, including 13 interaction terms. Seventeen out of the 24 variables are statistically significant, including 8 primary variables 24 and 9 interaction terms, ranging from direct and indirect driver categories of climate change, land use change, pollution, institutions and governance, and economic and technological pathways. The differences are significant among high-income vs. low-income, and biodiversity- hotspot vs. non-hotspot countries across different variables (Figures 3.3.B, 3.3.C, 3.3.D, 3.3.E). Specifically, Political Stability has the most important positive impact (with an estimate of 0.3) on SDG 15 progress in high-income, non-hotspot countries (HN), while for low-income, non-hotspot countries (LN), the most important variable for positive impact is Forest Rents Percentage in GDP (0.26). Forest Area has the same estimate (0.23) across all countries, which is deemed as the most important positive factor in both high-income, biodiversity-hotspot (HB) and low-income, biodiversity-hotspot (LB) countries, as well as considered as the second most important positive driver for HN and LN countries. On the contrary, Forestry Production has the most important negative impacts on SDG 15 progress in HB (-0.23), HN (-0.14), and LB (-0.08) countries, followed by Population Growth Rate, which also has profound negative impacts for HN (-0.17) and HB (-0.15) countries. Some variables have opposite impacts on different country groups. For example, Agricultural Land Percentage has a positive impact (0.12) for SDG 15 progress in HB and HN countries and a negative impact (-0.07) in LB and LN countries; Forestry Import has a positive impact (0.01) in HB and LB countries and a negative impact (-0.04) in HN and LN countries (Figure 3.3.F). The average effects of both direct and indirect drivers on SGD 15 progress are positive yet they are different across country groups. For HB countries, the direct drivers have more prevailing impacts where the direct impact is summed at 0.26 (to the indirect impact sum estimate of 0.21). However, for other country groups (HN, LB, LN), the indirect drivers play more important roles. 25 For instance, the sum estimated for HN countries is 0.42 (to the direct impact sum estimate of 0.35) and indirect driver estimates are higher than direct driver estimates for all low-income countries (Figure 3.4.B). Table 3.1. Regression variable estimates for SDG 14. Variable name Log of Fish Export Log of Fish Import Log of Fish Production GDP per Capita Political Stability Log of Population Density Population Growth Rate Temperature Change Log of Tourist Note: p (<0.1)*, (<0.05)**, (<0.01)*** SDG 14 0.58*** -0.26*** -0.51*** 0.23*** -0.25*** -0.18*** -0.12** 0.15** 0.40*** Coefficient estimate of Target 14.5 0.6*** -0.14 -0.65*** 0.23*** -0.25*** -0.18*** 0.1 0.05 0.48*** Target 14.1 -0.18 0.1 0.18 -0.06 0.12 0.14** 0.01 0.16* 0.1 Target 14.7 0.27*** -0.43*** 0.25** 0.22** 0.38*** 0.02 0.23** 0.15** -0.35*** 26 Coefficient estimate of Table 3.2. Regression variable estimates for SDG 15. Variable name SDG 15 -0.07*** Target 15.1 -0.15*** Target 15.4 -0.03 Target 15.5 -0.05* 0 0 0.05 0.05 0.01 -0.01 -0.01 -0.04* 0.4*** 0.2*** 0.05** -0.08** -0.08** -0.07** -0.06** 0.43*** 0.12*** 0.11*** 0.21*** 0.26*** 0.19*** -0.09*** -0.17*** -0.12*** -0.19*** -0.09*** -0.37*** -0.27*** Agricultural Land Percentage Agricultural Land Percentage * High Income (0/1) Percentage of Forest Rents in GDP Percentage of Forest Rents in GDP * Biodiversity Hotspot (0/1) Percentage of Forest Rents in GDP * High Income (0/1) Forestry Import Forestry Import * Biodiversity Hotspot (0/1) Forestry Production Forestry Production * Biodiversity Hotspot (0/1) Forestry Production * High Income (0/1) Log of Forest Land Area Log of PM2.5 Population Growth Rate Population Growth Rate * Biodiversity Hotspot (0/1) Population Growth Rate * High Income (0/1) Political Stability Political Stability * Biodiversity Hotspot (0/1) Political Stability * High Income (0/1) Regulatory Quality Temperature Change Note: p (<0.1)*, (<0.05)**, (<0.01)***; all variables of interaction terms are italicized. 0.14*** 0.04 -0.19*** 0.48*** 0.14*** -0.31*** -0.13*** 0.26*** 0.01 0.23*** 0.1*** -0.04 0.13*** 0.04** -0.11*** 0.01 -0.13*** -0.11*** -0.35*** -0.11*** -0.11*** -0.11*** -0.14*** -0.15*** -0.15*** 0.17*** 0.11*** 0.23*** 0.19*** 0.13*** 0.14*** 0.11*** 0.22*** 0.15*** 0.16*** -0.08** -0.07** -0.1*** 0.07** -0.05 -0.05 0.02 0.06 0 0 27 Target 15.6 -0.07** Target 15.8 0.15*** 0.17*** -0.07* 0.03* -0.1*** 0.1*** -0.04 -0.01 0.12*** 0.03 0.05** 0.05** Target 15.9 -0.04 0.09*** 0.01 -0.04 -0.03 0.12*** 0.02 -0.03 -0.13*** -0.11*** -0.14*** -0.03 0.09*** 0.08** 0.16*** -0.09*** -0.11*** 0.21*** -0.12*** 0.2*** 0.07 0.05* -0.06* -0.02 0.03 -0.04 0.04* 0.02 0.01 3.4.2 Synergies and Trade-offs Between Sustainable Development Goals and Targets At the Target level, regression estimates showed different impacts. Some variables have the same positive impacts on SDG Targets as they do on SDG progress, while others have the opposite negative impacts. Here, we list some variables that have statistically significant estimates in regressions. For instance, Fish export has both positive impacts on Targets 14.5 and 14.7, with estimates of 0.6 and 0.27 respectively; GDP per capita also has both positive impacts on those Targets (0.23 and 0.22). Fish import has a negative impact (-0.43) on Target 14.7 aligning with the negative impact at the Goal level (-0.26). However, Fish production has a negative impact (- 0.65) on Target 14.5, consistent with the negative impact at the Goal level (-0.51) but has a positive impact (0.25) on Target 14.7, opposite to the Goal level estimate. Besides, Population density has the same negative impact (-0.18) on Target 14.5 and SDG 14, but the impact is positive (0.14) on Target 14.1. Tourist has a positive impact (0.48) on Target 14.5 (consistent with a positive impact on SDG 14) and a negative impact (-0.35) on Target 14.7 (Table 3.1). The same impacts (both positive or negative) among Targets and SDGs are considered synergies, meaning the variable contributes to achieving or preventing the Target and SDG progress at the same time. The same positive impacts are considered positive synergies (win-win), and the same negative impacts are considered negative synergies (lose-lose). For example, GDP per capita has both positive impacts on Targets 14.5 and 14.7, which is a positive synergic effect meaning that GDP per capita helps achieve both Targets 14.5 and 14.7. On the contrary, Fish production has both negative impacts on SDG 14 and Target 14.5, which is a negative synergic effect meaning that Fish production suppresses both SDG 14 and Target 14.5 progress. The opposite impacts (one positive, while other negative, vice versa) among Targets and SDGs are seen as trade-offs (win-lose), meaning the variable buttresses to fulfill one Target/SDG while 28 compromising the other (Zhao et al., 2021, Xing et al., 2024). Both synergies and trade-offs exist in SDGs 14 and 15 among their Targets. From the results above, many trade-offs have been detected among SDG 14 and their Targets. Most variables in the SDG 15 Target regressions have consistent impacts (both positive or both negative) on SDG 15 and their Targets, so synergies are more prevailing (Table 3.2). Nevertheless, a few trade-offs are noticeable. For example, Forestry Import for non-hotspot countries has negative impacts on SDG 15 (-0.04), Targets 15.1 (-0.08), 15.4 (-0.12) and 15.5 (- 0.08), but it has a positive impact (0.12) on Target 15.9. Forestry Import for biodiversity hotspot countries has positive impacts on SDG 15 (0.01) and Target 15.5 (0.04). Forestry Production for LB countries has negative impacts on SDG 15 (-0.08), Targets 15.1 (-0.3), 15.4 (-0.11), and 15.5 (-0.25), but a positive impact on Target 15.8 (0.1). Forestry Production for LN countries has negative impacts on SDG 15 (-0.14), Targets 15.1 (-0.3), 15.4 (-0.14), 15.6 (-0.01), 15.8 (-0.06), and 15.9 (-0.17), but a positive impact on Target 15.5 (0.06). In addition, Forest Land Area has a negative impact (-0.13) on Targets 15.5, while it has all positive impacts on SDG 15 (0.23), Targets 15.1 (0.48), 15.4 (0.14), 15.6 (0.09), and 15.9 (0.2). The synergies and trade-offs among SDGs and their Targets could reveal insights into further actions and policy implications to achieve sustainable development holistically. 3.5 Discussion Our LASSO regression approach and results identified the key variables and their impact on SDGs 14 and 15 progress among countries of different income levels and biodiversity hotspot status. Fish Production has the most profound negative impact on SDG 14, so does Forestry Production on SDG 15. Forest Area and Forest Rents in GDP Percentage have the most positive impact on SDG 15, while Fish Export, surprisingly, has the most positive correlation with SDG 29 14. The drivers for SDG progress and their significant levels largely vary among countries of different income and biodiversity hotspot status. Both synergies and trade-offs exist among SDGs and their Targets, highlighting potential challenges for future sustainable planning and opportunities to maximize the common benefits of multiple socioecological sustainability goals while minimizing the conflicting interests. The mixed expected and unexpected variable estimates on SDG progress are not fully understood. Several variables have either positive or negative effects on SDG progress, aligning with theories and expectations of drivers for environmental change. For example, Fish harvest and human population pressures have negative impacts on SDG 14 progress, which is illustrated by the negative estimates of Fish production, Human population density, and Population growth rate. Economic factors such as GDP per capita have a positive impact on SDG progress. However, fish trade has an interestingly mixed impact when Fish export has a positive estimate and Fish import has a negative outcome, refuting the assumption that SDG 14 scores should be higher when countries import more fish and conserve their domestic fish stocks, and lower SDG 14 scores when countries export more fish and consume their own natural resources. Namely, countries such as Croatia (high-income) and Morocco (low-income) that made great SDG 14 progress report mixed impacts of Fish export (negative for both countries), Fish import (positive for Croatia, negative for Morocco), and Fish production (positive for both countries); other countries such as Finland (high- income) and Tonga (low-income) that had retrogress in SDG 14 also show mixed impacts of Fish export (positive for Finland, negative for Tonga), Fish import (negative for Finland, positive for Tonga), and Fish production (positive for both countries). Possibly, the increase in domestic aquaculture that is highly correlated with Fish production and Fish export reduced the negative exploitation impact and sufficed sustainable fish capture. This might also result from the potential 30 reverse causation when higher global sustainable fisheries standards are imposed, countries with better SDG 14 progress practice more sustainable fishing and hence are likely to have more fish exports. Meanwhile, political stability also has an unexpected negative impact on SDG 14 progress when separating countries by their income levels. This is contradictory to the literature and beyond established knowledge (Feng, 1997, Aisen and Veiga, 2013, Ali, 2019), likely resulting from the limitation of LASSO technique that causal inference was not fully established during the regularization and modeling process. Due to data limitations, our study does not capture all variables of direct and indirect drivers for SDG progress. The data analyzed in this study include (1) as many variables as possible, (2) as many countries as available, and (3) as long-time span as possible. The panel data and regression analysis can reveal a significant part of the relationship in how different drivers impact SDG 14 and 15 progress as proxy of global socioecological change. However, it requires caution in examining the causal inference. The LASSO technique is useful as an exploratory approach to provide an initial understanding of significant variables that correlate with target-dependent variables while balancing the simplicity of regression models. However, limited causal inference, which is important for theory testing or policy making, is produced due to a lack of explicit pathways and mechanisms identified. For example, it is unlikely that the increased temperature or pollutants would improve SDG progress. These may not be the perfect indicators to choose based on data insufficiency. Other variables, such as fish trade, may involve dual directional causalities, meaning that those variables may have impacts on SDG progress at the same time being affected by SDG progress. Hence, further pathway studies are needed to discover the mechanisms for theory testing or policy recommendations. 31 For SDG 15 progress, the expected and unexpected impacts are also mixed across variables and vary among different country groups (income level, biodiversity hotspot). Specifically, both Forest land area and Agricultural land percentage have positive impacts on high-income countries, but Agricultural land percentage has a negative impact on low-income countries regardless of biodiversity hotspot difference. Forest land area is truly important for all countries to improve SDG 15 progress, with positive impacts across all country types. The expansion of Agricultural land percentage remains controversial, because agricultural land percentage has a positive impact on SDG 15 progress for high-income countries and a negative impact for low-income countries. Further investigation should focus on potential different mechanisms of how agricultural lands impact countries’ SDG 15 progress while considering other hidden factors. For example, agroecosystems that provide habitats for wildlife and enhance biodiversity while achieving food supply goals in highly developed countries with limited land would require exhaustive study and careful design. Besides, Forestry import has positive impacts on both high- and low-income countries when they are biodiversity hotspots. This can be explained by the fact that countries show better sustainability progress by conserving domestic resources through import while transferring environmental costs to other countries (Chung and Liu, 2022, Xu et al., 2020). It is critical and efficient to conserve forest ecosystems in countries with rich biodiversity. But Forestry import has a negative impact on non-hotspot countries. This could be the fact that they rely more on domestic forest consumption and reduce forestry import when biodiversity is not a primary goal to protect local forests and thus countries are not motivated to plant trees. For those countries, it is unlikely that Forestry import will directly impact their SDG 15 progress, but instead, domestic forest consumption could be more significant, calling for closer scrutiny. Additional causal path diagrams with quantitative analysis would be beneficial to enhance the understanding. Forest rent 32 percentage of GDP plays a more important and positive role in non-hotspot countries than that in biodiversity hotspot countries. Considering the concept of forest rent (roundwood harvest times the product of regional prices and a regional rental rate) and determinants to this variable, it could be explained that non-biodiversity countries produce more quantities of forestry products with a lower cost. All these assumptions and explanations need further examination and empirical studies, particularly those that are inconsistent with existing theories or established knowledge. The indirect drivers are more important than the direct drivers for all countries in SDG 14 progress and non-biodiversity hotspot countries in SDG 15 progress. However, direct drivers have a profound impact on SDG 15 progress for biodiversity hotspot countries. It is critical to examine the hidden factors and associated mechanisms that drive environmental and socioeconomic changes. The integrated metacoupling framework (Liu, 2017, Liu, 2023) has helped to identify important natural and social drivers domestically and internationally for SDG progress in this study, and it would be of great use to further demonstrate interactions among different endogenous (domestic land competition between forest and agriculture, population (Stern et al., 1992, Dietz and Rosa, 1994, Ehrlich and Holdren, 1971, da Silva et al., 2021) and exogenous (tourism, trade (Xu et al., 2020, Zhao et al., 2020)) factors and analyze system feedbacks within and beyond countries, placing the foundation for building complex system dynamics models. Furthermore, understanding the mechanisms that cause SDG 15 progress for biodiversity hotspot countries is urgently needed. Intense land competition between forest and agricultural activities significantly influences countries’ SDG 15 progress. Further study could explore the possibility of releasing agricultural land use pressure of those hotspot countries by satisfying domestic agricultural needs through international trade or from countries with less land use competition. 33 Synergies and trade-offs among SDGs and their Targets should be carefully evaluated and incorporated into decision making. Drivers that promote synergic effects among SDGs and Targets should be emphasized and those creating conflicts should be given special attention. For example, Fish export creates the opportunity to improve Targets 14.5, 14.7 and SDG 14 simultaneously, which could be an effective leverage and promotion for future marine resources management and sustainable development, considering almost half a billion people depend at least partially on small-scale fisheries (Sachs et al., 2022). But key questions on Fish export including the portion of wild capture vs. aquaculture, direct export and re-export, should be cautiously examined prior to implementing policies at a global scale. The trade-offs should be realized to inform policymaking. For instance, Forest Land Area plays such an imperative role in contributing to SDG 15 and most of its Target progress, but attention must be drawn to investigate the mechanism of how it negatively impacts Target 15.5 as a measure of trends in overall extinction risk (Red List Index). The discussion of biodiversity habitat quality versus quantity should be adequately considered in future conservation planning and policy agenda. 3.6 Conclusions Using a metacoupling framework, we explored the relationship between drivers for SDG progress within and among countries. In particular, we deployed the machine learning based statistical approach (LASSO) to learn the significant variables that impact countries’ SDG progress. Our study highlights the expected and unexpected impacts of multiple factors that affect SDG progress for countries at different income levels and biodiversity hotspot statuses. Our results further illustrate the synergies and trade-offs between SDGs and their Targets, calling for careful decision making in the future to maximize the common benefits of multiple socioecological sustainability goals while minimizing the conflicting interests. Our study provides an exploratory 34 example of integrating the metacoupling framework and the LASSO statistical approach, paving the way for more pathway studies. 35 CHAPTER 4: SUSTAINING LIFE ON LAND THROUGH A METACOUPLING APPROACH: SIMULATING SPAIN’S SDG 15 PROGRESS 4.1 Abstract Anthropogenic activities such as natural resources harvest, trade, and population growth have substantial impacts on the environment and become a major challenge to socioecological sustainability. Lack of understanding in achieving environmental and socioeconomic sustainability within and across boundaries is a bottleneck in conservation and sustainability science. The metacoupling framework that integrates interactions across multiple scales and borders, together with System Dynamics Model, is a powerful tool to analyze system interactions and simulate responses both qualitatively and quantitatively. We first applied the metacoupling framework to understand the interactions among environmental and socioeconomic variables and their impact on countries’ SDG 15 progress. Then we used Spain as an example and developed a System Dynamics Model to explore how SDG 15 progress responded to forest and population change. Our results show that Net Forest Import has the dominant impact on SDG 15 progress, while other variables like Forestry Production, Forest Regeneration Rate, and Human Population also have impact on SDG 15 progress to different extents. SDG 15 progress, resonating with Forest Area variation, is likely to reach the peak in mid 2030s and depreciate in the long run with the increase of forest harvest. Future natural resources management and conservation planning should be aware of and set up the baseline for potential minimum sustainable forest regeneration and maximum sustainable harvest. The modeling outcome not only served such purposes for providing important information to natural resources management but can also be utilized by broader stakeholders for communication with different communities and learning feedback for model refinement. 36 4.2 Introduction A major challenge for conservation science is to achieve environmental and socioeconomic sustainability within and across boundaries. Integrated studies of coupled human and natural systems (CHANS, Liu et al., 2007a, Liu et al., 2007b) have generated important findings on complex patterns and processes that studies through a single lens of physical or social sciences cannot obtain. The holistic metacoupling framework (human-nature interactions within a CHANS as well as between adjacent and distant CHANS, Liu, 2017) integrates interactions across multiple scales across borders. This framework could provide a useful conceptual platform for stakeholder coordination and decision-making to achieve conservation and sustainable goals beyond boundaries. However, quantification of the framework is needed to make coordination and decision-making more effective. System dynamics modeling (SDM) is used to simulate and understand complex system patterns and processes with quantification features (Meadows, 2008). By identifying the stocks and flows, SDM represents the key feedback structures in the system. SDM can also show scenarios based on different policy interventions. For example, different extents of resource consumption would be the specific scenario analysis in the system dynamics model. Through the feedback loops in the system, potential problems and solutions could be found in terms of conservation and sustainable development goals. Besides, SDM could also help identify the delayed effect of policy intervention, which is significant and informative for future planning. Therefore, SDM is an appropriate approach used in this study to evaluate different policy scenarios and analyze potential strategies for achieving sustainable development goals. We have identified significant variables that impact SDG progress for countries at different income levels and biodiversity hotspot statuses in the previous chapter. However, the mechanisms 37 through which those variables impact SDG progress differ from country to country and therefore remain unknown. Intermediate converters among variable interactions were not fully understood. Here, we first apply the metacoupling framework to delineate the problem (system processes and interactions among forest, land use for anthropogenic activities, governance, economy, and SDG progress), and then use SDM to model the problem by identifying and quantifying the interactions among system components. The modeling outcomes aim to inform future conservation planning, natural resource management, and community sustainable development. Research questions: (1) How does the change of forest area impact a country’s SDG 15 progress? (2) What parameter has the largest impact on a country’s SDG 15 progress? 4.3 Methodology 4.3.1 Metacoupling framework To systematically understand the human-nature interactions within and across multiple scales and borders, metacoupling framework is used to understand the problem of this study. Three types of human-nature interactions (couplings) are delineated under the complete metacoupling framework: (1) within a coupled system (intracoupling), (2) between distant coupled systems (telecoupling), and (3) between adjacent coupled systems (pericoupling). Here, we adopted the metacoupling framework and followed the six general procedures for operationalizing this framework (Liu, 2017) including setting research goal, defining focal system, reviewing literature and conducting additional studies on flows, agents, causes, and effects, identifying couplings and sending, receiving, and spillover systems, conducting further studies on metacoupling components and interrelationship, and publishing and communicating final results. The preliminary system identification and definition (CHANS, metacoupling) can be found in Figure 4.1. 38 Figure 4.1. Simplified conceptual metacoupling framework: focal country (coupled human and natural system), adjacent country, distant country, and relationships between forest harvest, agricultural land transformation, population, economy development, SDG 15 performance, and international forest trade within the three coupled countries. Components are categorized under human system (brown rectangle) and natural system (green rectangle), including agricultural land, population, economy, forest, temperature, pollution, as causes and SDG 15 progress as effect. The interactions within a coupled human and natural system are shown in light blue arrows. The forestry trade is another component (i.e., flow) between the focal country and adjacent/distant countries. Each country was confined with a black line of rectangle. The interactions between different coupled human and natural systems (e.g., between focal system and distant system) are shown in black arrows (the solid line indicates a direct/observable interaction; the dash line indicates an indirect/unobservable or potential interaction). 4.3.2 Conceptual Framework based on Metacoupling With the focus on only one country (system), the Casual Loop Diagram below shows the interactions among forest, agriculture, population, economy, governance, and trade. There are two balancing feedback loops (labelled as brown B in Figure 4.2) and five reinforcing feedback loops (labelled as green R in Figure 4.2). The first balancing loop (B1) is that the larger total forest size will provide more forest harvest, but more harvest will lead to a decreasing forest size. The second balancing loop (B2) is that more population will have more population deaths holding the death rate constant, and the more deaths will cause a smaller size of population. Therefore, the population is balanced out through this loop. On the other hand, the reinforcing feedback loops include: R1. 39 More domestic forest demand will require more forest product, and more forest product will meet more domestic demand; R2. More economic growth (development activities) will boost domestic forest demand, and more domestic demand will satisfy economic needs; R3. More population will create more economy (productivity), and more economy will support larger population; R4. More population will drive more agriculture (activity, products), and more agriculture will support more population; R5. More population will drive more population births holding the birth rate constant, and more population births will contribute to a larger population size. Figure 4.2. Causal loop diagram with feedback loops for the forest system. The system components are in blue text, and the interactions among components are connected through pink arrows. The positive sign shows a positive effect, and the negative sign shows a negative effect. Feedback loops are labelled as unclosed ½ circles with arrows. The brown feedback loops with letter “B” are negative (balancing) feedback loops; the green feedback loops with letter “R” are positive (reinforcing) feedback loops. 4.3.3 System Dynamics Model (SDM) 4.3.3.1 Geographic foci, data sources, and assumptions This study used Spain as an example, to illustrate the interactions among forest, population, and SDG 15 progress. Spain is one of countries that made tremendous progress in SDG 15 between 2010 and 2020, with scores increasing from 29.74 to 77.31 and ranks emerging from the 60th to the 15th worldwide over those ten years (Zhang et al., 2023). Understanding on how Spain achieved 40 their SDG 15 progress and what variables have large impacts is useful for future conservation and development planning and as a reference for other countries. Multiple sources of data were used in this study. Population, Population Growth Rate, Forest Area, Forestry Production, and Net Forest Import data were collected through World Bank Group (World Bank Country and Lending Groups, 2021), while Forest Regeneration Rate, Carrying Capacity, and SDG 15 Progress were referred from literature review (Bolin et al., 2000, Instituto Nacional de Estadística, 2022, Zhang et al., 2023). The formula used to define relationships between components in the model were not all available. Several trials were made in calibration and validation processes to best match the real-world data, such as Forestry Production, Net Forest Import, and Population Growth Rate. The other assumption was also made especially for generating the graphical function for Effect of Carrying Capacity on the ratio of Population Growth Rate/Carrying Capacity (the effect increases with the ratio increase at a diminishing rate of return, range from 0 to 1). Namely, when the population approaches the carrying capacity, the effect is more profound leading to a lower population growth rate (Cohen, 1995, Vandermeer, 2010, Meadows et al., 2018). 4.3.3.2 Model description To simplify the complex system, only limited components and their interactions were included in the model from the causal loop diagram (Figure 4.2). The model simulated the human and natural system interactions (e.g., population, land use change, forest change through production, trade, and regeneration, and SDG progress) from 2000 to 2050 with DT = 1 year. The model was developed with Stella Architect V2.1.5 (ISEE System 2022), with initial settings listed in Table 4.1. The complex system is a coupled human and natural system at a country scale, where there are two major sub-systems: population and forest. 41 The model has two stocks (Population, and Forest Area), and four flows (Net Growth, Forest Regeneration, Net Forest Harvest, and Land use change). The Net Growth is a bi-flow, meaning it can be an inflow towards or outflow from the stock of Population (contributing to the increase or decrease of the stock) depending on the positive or negative values of Net Growth. Namely, if the Net Growth is positive, there will be more population added towards the stock of Population; if the Net Growth is negative, there will be a removal from Population. The Net Growth is determined by the Population Growth Rate, which is dependent on Effect of Carrying Capacity. The Effect of Carrying Capacity relies on Population and Carrying Capacity. The Carrying Capacity is a user-defined value, and it varies from country to country (for Spain, it is set as 53 million people, Instituto Nacional de Estadística, 2022). Land Use Change is an outflow, meaning that the stock Forest Area may be taken away by Land Use Change depending on both Population and Carrying Capacity. Forest Regeneration is an inflow towards the stock of Forest Area, and it is the multiplication of Forest Area and Forest Regeneration Rate. Forest Harvest is an outflow of Forest Area, which is impacted by Forestry Production and Net Forest Import. There is one balancing feedback loop (labelled as brown B), one reinforcing feedback loop (labelled as green R), and one mixed (reinforcing and balancing) feedback loop (labelled as yellow R/B in Figure 4.3) in the system. The mixed loop (R1/B1) between Population and Net Growth is dependent on whether Net Growth is positive or negative. If positive (inflow), it is a reinforcing loop because larger population will have more net population growth, which in turn contributes to a larger population size. If negative (outflow), it is a balancing loop because larger population will have less Net Growth, as a source of reducing population size. The balancing loop (B2) is that larger Population, stronger Effect of Carrying Capacity, lower Population Growth Rate, lower Net Growth, smaller Population, in this case, the Population is constrained through this loop. The 42 reinforcing loop (R2) is that more Forest Area, more Forest Regeneration holding the Forest Regeneration Rate constant, contributing to additional Forest Area, in this case, Forest Area is reinforced through the loop. Figure 4.3. System dynamics model with feedback loops for the population and forest system. The system components are in blue text, and the interactions among components are connected through pink arrows. Feedback loops are labelled as unclosed ½ circles with arrows. The brown feedback loops with letter “B” are negative (balancing) feedback loops; the green feedback loops with letter “R” are positive (reinforcing) feedback loops; the yellow feedback loops with letters “R/B” are mixed (both reinforcing and balancing) feedback loops. 43 Flow Converter Table 4.1. Summary of system components and initial values or formula. Component Name Stock Population Forest Area Net Growth Land Use Change Forest Harvest Forest Regeneration Population Growth Rate Effect of Carrying Capacity Carrying Capacity Forestry Production Net Forest Import Forest Regeneration Rate SDG 15 Progress Initial values or formula [unit] 40,567,864 [person] 170,939.3 [km2] Population_Growth_Rate/100*Population (Population/Carrying_Capacity)*1000 [km2] (Forestry_Production- Net_Forest_Import)/12000 [km2] Forest_Area*Forest_Regeneration_Rate [km2] (1-Effect_of_Carrying_Capacity)*43 Population/Carrying_Capacity (in graphical function) 53000000 [person] 98018*(TIME-2000) + 15921000 [m3] -266374*(TIME-2000) + 3354634 [m3] 0.016 0.000528131*Forest_Area - 20.77720428 4.3.3.3 Model calibration and validation To calibrate the model, I ran the model for 10 years with initial settings (DT=1) and plotted the simulated Population, Forest Area, Forestry Production, and Net Forest Import results against the real-world data from 2000 to 2010. The parameters I changed to fit the model with reality were Forest Harvest, Land Use Change, Population Growth Rate, and Effect of Carrying Capacity in graphical function. With the final calibrated model, I changed the runtime to 20 years from 2000 to 2020. After exporting simulated data from 2010 to 2020, I validated the model results by plotting them against real-world data. Because SDG 15 Progress data was not available before 2010 and the major data jumps between 2011 and 2012, 2015 and 2016 could mislead the prediction, only data from 2016 to 2020 were used for this variable in the iterative model calibration process to avoid noises. 44 4.3.3.4 Reference model prediction and sensitivity analysis The reference model prediction was based on the existing variables and their relationships with no interventions after model validation, between 2020 and 2050. The key variables of interest are Forest Area and SDG 15 Progress. I changed the runtime to 50 years from 2000 to 2050. The sensitivity analysis was also performed during the same period after model validation. The key variables of interest are Forest Area and SDG 15 Progress. To perform the five runs of sensitivity analysis, Forest Regeneration Rate, Initial Forest Area, Population, Forestry Production, and Net Forest Import were changed one at a time ranging from 10% lower and 10% higher than the baseline values or formula while holding other input variables constant. Sensitivity is estimated by the index of Sx, within the following formula. 𝑆𝑥 = ∆𝑋 𝑋 ⁄ ∆𝑃 𝑃 Where X is the variable under the original condition, ΔX is either the difference of the variable at 10% lower value between the original variable value or the difference of the original variable and the variable at 10% higher value. For example, for Forest Regeneration Rate analysis, X is the Forest Regeneration Rate at the baseline of 0.016; ΔX is either the difference between 0.0144 (lower 10%) and 0.016 (baseline) or the difference between 0.016 (baseline) and 0.0176 (higher 10%). P represents the value of either Forest Area or SDG 15 Progress under the original condition and ΔP is the difference in the data value of either Forest Area or SDG 15 Progress between the original and modified conditions. For example, for Forest Regeneration Rate analysis, P is either Forest Area or SDG 15 Progress (when Foerst Regeneration Rate is set at 0.016); ΔP is either the difference between Forest Area values (when Forest Regeneration Rate is 0.0144 and 45 0.016; or when Forest Regeneration Rate is 0.016 and 0.0176) or the difference between SDG 15 Progress values under the above conditions. Sx refers to the change in the Forest Area or SDG 15 Progress due to the change in the following variables at a time (Forest Regeneration Rate, Initial Forest Area, Population, Forestry Production, and Net Forest Import). The larger the value, the more sensitive Forest Area or SDG 15 Progress are to those variables of change. For Initial Forest Area, the baseline is 170,939.3, lower bound is 153,845.37, and upper bound is 188,033.23. For Population, the baseline is 40,567,864, lower bound is 36,511,077.6, and upper bound is 44,624,650.4. For the formula of Forestry Production and Net Forest Import, a coefficient of 0.9 or 1.1 was multiplied to its original formula to represent the 90% or 110% of variable levels. I ran each of the five sensitivity analyses individually and then exported the changed Forest Area and SDG 15 Progress in separate Excel files. For each sensitivity analysis, two Sx values were produced – one showed the difference between the baseline value and its 10% lower value, the other showed the difference between the baseline value and its 10% higher value. Those Sx values would depict how input variable sensitivity affects the key output variables of interest and show which variable among those five changed variables is more significant to the output variable variation. Table 4.2. Summary of modified system component values for five runs of sensitivity analysis. Component Name Lower/higher values or formula Converter Forest Regeneration Rate Forest Area Population Forestry Production Net Forest Import 0.0144, 0.0176 153,845.37, 188,033.23 36,511,077.6, 44,624,650.4 0.9* (98018*(TIME-2000) + 15921000), 1.1* (98018*(TIME- 2000) + 15921000) 0.9* (-266374*(TIME-2000) + 3354634), 1.1* (-266374*(TIME- 2000) + 3354634) 46 4.4 Results 4.4.1 Calibration The model calibration indicated when Forest Harvest, Land Use Change, Population Growth Rate, and Effect of Carrying Capacity in graphical function were set as current formula summarized in Table 4.1, the model best fit the real-world data from 2000 to 2010 (Table 4.3), especially for the key stocks of interest (Forest Area and Population). Then I plotted the simulated data against the real-world data for Forest Area, Forestry Production, Net Forest Import, and Population, which generated the R2 values of 0.998, 0.211, 0.606, and 0.987 (Figure 4.5). Table 4.3. Real-world and modeled data for Forest Area, Population, Forestry Production, and Net Forest Import between 2000 and 2010. A Real-world vs. Modeled Forest Area from 2000 to 2010 l e d o M 180000 179000 178000 177000 176000 175000 174000 173000 172000 171000 170000 y = 0.5572x + 75879 R² = 0.9981 170000 172000 174000 176000 178000 180000 182000 184000 186000 188000 Real-world Figure 4.4. Plots of modeled against real-world data for four variables between 2000 and 2010. (A) Forest Area, (B) Forestry Production, (C) Net Forest Import, and (D) Population. 47 Year20002001200220032004200520062007200820092010Real-world Forest Area170939.3172390.71173842.12175293.5176744.9178196.4179647.8181099.2182550.6184002185453.4Modeled Forest Area170939.3171861.7002172758.6346173630.2174476.6175297.9176094.2176865.3177611.2178331.7179026.7Real-world Forestry Production1592100016986000178280001813500018345000177110001732300016510000196273741606003521209399Modeled Forestry Production1592100016019018161170361621505416313072164110901650910816607126167051441680316216901180Real-world Net Forest Import335463436410003059000287100026390003287000332500033320981576000944904392626Modeled Net Forest Import335463430882602821886255551222891382022764175639014900161223642957268690894Real-world Population4056786440850412414315584218764542921895436531554439731945226803459541064636294646576897Modeled Population4056786441090352.0441584035.814204967342488118429023374330084143683868440516854440458944742902 Figure 4.4 (cont’d) B Real-world vs. Modeled Forestry Production from 2000 to 2010 17000000 16800000 16600000 l e d o M 16400000 16200000 16000000 15800000 y = 0.0955x + 1E+07 R² = 0.2114 0 5000000 10000000 15000000 20000000 25000000 Real-world C Real-world vs. Modeled Net Forest Import from 2000 to 2010 y = 0.6244x + 409308 R² = 0.606 l e d o M 4000000 3500000 3000000 2500000 2000000 1500000 1000000 500000 0 0 500000 1000000 1500000 2000000 2500000 3000000 3500000 4000000 Real-world 48 Figure 4.4 (cont’d) Real-world vs. Modeled Population from 2000 to 2010 y = 0.6174x + 2E+07 R² = 0.987 D l e d o M 45000000 44500000 44000000 43500000 43000000 42500000 42000000 41500000 41000000 40500000 40000000 40000000 41000000 42000000 43000000 44000000 45000000 46000000 47000000 Real-world With the initial values or formula of stocks, flows and converters in Table 4.1, the simulation results from 2000 to 2010 are shown in Figure 4.5. Forest Area constantly increased from 171 to 179 thousand square kilometers, while the three major flows all increased – Forest Regeneration increased from 2.74 to 2.86 thousand square kilometers, Forest Harvest increased from 1.05 to 1.35 thousand square kilometers, and Land Use Change slightly increased from 765 to 844 square kilometers. Forestry Production increased from 15.9 to 16.9 million cubic meters, while Net Forest Import decreased from 3.35 million to 691 thousand cubic meters. Population drastically increased from 40.6 to 44.7 million over 2000 and 2010. 49 A C B D Figure 4.5. Model calibration results. (A) Forest Area, (B) Forest Harvest, Forest Regeneration, Land Use Change, (C) Forestry Production, Net Forest Import, and (D) Population estimates between 2000 and 2010. 4.4.2 Validation Keeping the initial values and formula set in calibration and increasing the runtime for another 10 years (till 2020), I found the model fit the real-world data well (Table 4.4). Then I plotted the modeled data against the real-world data, which generate the R2 values of 0.902, 0.004, 0.640, and 0.280 (Figure 4.6). Table 4.4. Real-world and modeled data for Forest Area, Population, Forestry Production, and Net Forest Import between 2010 and 2020. 50 Year2011201220132014201520162017201820192020Real-world Forest Area185465.1185476.8185488.4185500.1185511.8185552.4185593185635.9185678.8185721.7Modeled Forest Area179696180339.6180957.3181548.9182114.2182653.1359183165.4254183650.8929184109.3261184540.5019Real-world Forestry Production19327772176867951899429820104343219503611917160119179531224697821863558617881367Modeled Forestry Production16999198170972161719523417293252173912701748928817587306176853241778334217881360Real-world Net Forest Import23361-116431-625677-1162310-1456442-1621376-855882-873275-1586331-1972841Modeled Net Forest Import424520158146-108228-374602-640976-907350-1173724-1440098-1706472-1972846Real-world Population46742697467730554662004546480882464448324648406246593236467977544713483747365655Modeled Population450669684537714745673817459573664622819146486695.5146733288.8146968381.0247192382.6947405702.71 A Real-world vs. Modeled Forest Area from 2010 to 2020 l e d o M 186000 185000 184000 183000 182000 181000 180000 179000 y = 17.097x - 3E+06 R² = 0.9018 185450 185500 185550 185600 Real-world 185650 185700 185750 B Real-world vs. Modeled Forestry Production from 2010 to 2020 18000000 17900000 17800000 17700000 17600000 l e d o M 17500000 17400000 17300000 17200000 17100000 17000000 16900000 y = 0.012x + 2E+07 R² = 0.004 0 5000000 10000000 15000000 20000000 25000000 Real-world Figure 4.6. Plots of modeled against real-world data for four variables between 2010 and 2020. (A) Forest Area, (B) Forestry Production, (C) Net Forest Import, and (D) Population. 51 Figure 4.6 (cont’d) C Real-world vs. Modeled Net Forest Import from 2010 to 2020 y = 0.979x + 229023 R² = 0.6401 1000000 500000 -2500000 -2000000 -1500000 -1000000 -500000 0 0 500000 l e d o M -500000 -1000000 -1500000 -2000000 -2500000 Real-world D Real-world vs. Modeled Population from 2010 to 2020 48000000 47500000 47000000 l e d o M 46500000 46000000 45500000 45000000 44500000 y = 1.3901x - 2E+07 R² = 0.2799 46200000 46400000 46600000 46800000 47000000 47200000 47400000 47600000 Real-world With the initial values or formula of stocks, flows and converters in Table 4.1, the simulation results from 2000 to 2020 are shown in Figure 4.7. Forest Area kept increasing from 179 to 185 thousand square kilometers, with Forest Regeneration increasing from 2.86 to 2.95 thousand square kilometers. Land Use Change 52 increased from 844 to 894 square kilometers, but Forest Harvest increased at a slower rate from 1.35 to 1.65 thousand square kilometers. This is mainly due to the change in Net Forest Import from 16.9 to -1.97 million cubic meters, while Forest Production increasing from 16.9 to 17.9 million cubic meters. Population increased from 44.7 to 47.4 million between 2010 and 2020. A C B D Figure 4.7. Model validation results. (A) Forest Area, (B) Forest Harvest, Forest Regeneration, Land Use Change, (C) Forestry Production, Net Forest Import, and (D) Population estimates between 2000 and 2020. 4.4.3 Reference model prediction and sensitivity analysis 4.4.3.1 Reference model prediction With the initial values or formula of stocks, flows, and converters in Table 4.1, I ran the reference model between 2000 and 2050. Forest Area constantly increased and reached the peak of 187,549 km2 in 2034, then slightly decreased to 183,071 km2 in 2050. SDG 15 Progress followed the same 53 pattern: it reached the peak of 78.27 in 2034 and then slightly dropped to 75.91 by the end of 2050. Land Use Change gradually increased from 765 to 966 km2 between 2000 and 2050. Forest Regeneration increased and reached the peak of about 3,000 km2 in 2034 and then decreased to 2,929 km2 in 2050, while Forest Harvest continuously increased from 1,047 to 2,565 km2 over those 50 years. This is primarily because Net Forest Import dropped from 3,354,634 to -9,964,066 m3, which became a net export, while Forest Production kept increasing from 15,921,000 to 20,821,900 m3 between 2000 and 2050. Population increased from 40,567,864 to 51,196,843, but the growth rate became slower over the years (Figure 4.8). A C B D Figure 4.8. Reference model results. (A) Forest Area, (B) SDG 15 Progress, (C) Forest Harvest, Forest Regeneration, Land Use Change, (D) Forestry Production, Net Forest Import, and (E) Population estimates between 2000 and 2050. 54 Figure 4.8 (cont’d) E 4.4.3.2 Sensitivity analysis of Forest Area Among the five runs of sensitivity analysis by changing one variable at a time, Forest Area is most sensitive to Net Forest Import and least sensitive to Initial Forest Area. Both 10% lower and higher values of Net Forest Import had major impact on Forest Area, with Sx values over 200 at the first few years of study. Although the Sx values dropped down to 80 around 2015, they bounced and peaked over 1700 in 2028. The absolute value of Sx remained as high as 700 in 2029 and shrank to about 10 at the end of 2050. Population and Forestry Production also had a large impact on Forest Area, but their Sx values (absolute) were not as high as Net Forest Import and decreased over time. Forest Regeneration Rate had a smaller impact on Forest Area, with an initial Sx value of 60 and diminishing towards 0 in 2050. Initial Forest Area has the minimal impact on Forest Area, regardless of 10% lower or higher of its baseline. 55 Sensitivity Analysis of Forest Area to Forest Regeneration Rate over Years Year 90% of Forest Regeneration Rate 110% of Forest Regeneration Rate Sensitivity Analysis of Forest Area to Initial Forest Area over Years A x S 70 60 50 40 30 20 10 0 B 1.2 1 0.8 x S 0.6 0.4 0.2 0 Year 90% of Initial Forest Area 110% of Initial Forest Area Figure 4.9. Sensitivity estimates of Forest Area to five variables between 2000 and 2050. The lower 10% of baseline value (e.g., 90% of variable) estimate is orange, and the higher 10% of baseline value (e.g., 110% of variable) estimate is in green. (A) Forest Regeneration Rate, (B) Initial Forest Area, (C) Population, (D) Forestry Production, and (E) Net Forest Import. 56 Figure 4.9 (cont’d) C Sensitivity Analysis of Forest Area to Population over Years x S 0 -50 -100 -150 -200 -250 Year 90% of Population 110% of Population D Sensitivity Analysis of Forest Area to Forestry Production over Years x S 0 -20 -40 -60 -80 -100 -120 -140 Year 90% of Forestry Production 110% of Forestry Production 57 Figure 4.9 (cont’d) E x S Sensitivity Analysis of Forest Area to Forest Import over Years 2500 2000 1500 1000 500 0 -500 -1000 Year 90% of Forest Import 110% Forest Import 4.4.3.3 Sensitivity analysis of SDG 15 Progress SDG 15 Progress is also most sensitive to Net Forest Import and least sensitive to Initial Forest Area. SDG 15 Progress is highly sensitive to both 10% lower and higher values of Net Forest Import, with Sx values over 150 for the first four years of study. The Sx values decreased to 60 in 2015, but they immediately increased by 1000 in 2028. The absolute value of Sx stayed as high as 550 in 2029 and eliminated to 9 in 2050. Population and Forestry Production had smaller but still noticeable impact on SDG 15 Progress, with an initial Sx value over 100 and reducing to single digits by the end of 2050. Forest Regeneration Rate had an even smaller impact on SDG 15 Progress, with an initial Sx value of about 50 and dropping to 0 in 2050. Initial Forest Area has the smallest impact on SDG 15 Progress throughout the whole study period between 2000 and 2050. 58 Sensitivity Analysis of SDG 15 Progress to Forest Regeneration Rate over Years Year 90% of Forest Regeneration Rate 110% of Forest Regeneration Rate Sensitivity Analysis of SDG 15 Progress to Initial Forest Area over Years A 60 50 40 x S 30 20 10 0 B x S 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Year 90% of Initial Forest Area 110% of Initial Forest Area Figure 4.10. Sensitivity estimates of SDG 15 Progress to five variables between 2000 and 2050. The lower 10% of baseline value (e.g., 90% of variable) estimate is orange, and the higher 10% of baseline value (e.g., 110% of variable) estimate is in green. (A) Forest Regeneration Rate, (B) Initial Forest Area, (C) Population, (D) Forestry Production, and (E) Net Forest Import. 59 Figure 4.10 (cont’d) C Sensitivity Analysis of SDG 15 Progress to Population over Years x S 0 -20 -40 -60 -80 -100 -120 -140 -160 -180 -200 Year 90% of Population 110% of Population D Sensitivity Analysis of SDG 15 Progress to Forestry Production over Years 0 -20 -40 x S -60 -80 -100 -120 Year 90% of Forestry Production 110% of Forestry Production 60 Figure 4.10 (cont’d) Sensitivity Analysis of SDG 15 Progress to Forest Import over Years E 2000 1500 1000 x S 500 0 -500 -1000 Year 90% of Forest Import 110% Forest Import 4.5 Discussion Our model results document how forest, SDG 15 progress, and population interacted within the couple human and natural system using Spain as an example. With the existing data from 2000 and 2010, the model is well calibrated by adjusting the initial values or formula for each parameter. The model fit well with the real-world Forest Area for those years. The calibration method is valid because only the endogenous factors in the model were changed which generates its own system behavior. Further, through the calibration and validation processes, the model well explains the trajectory of Forest Area (key stock of interest) patterns and sufficiently delineates Net Forest Import with real-world data from 2010 to 2020, despite that Forestry Production and Population variables are sparsely fit. The simulation result of SDG 15 Progress between 2020 and 2050 reflects the joint impact of forest and population systems. SDG 15 Progress, resonating with Forest Area dynamics, is likely to gain moderately by 2050 compared to 2000. However, it is noticeable 61 that the peak of SDG 15 Progress would reach in 2034, and that SDG 15 Progress may collapse due to loss of Forest Area and overharvest in the long term. Forest Area and SDG 15 Progress are sensitive to different parameters to various extents. The Net Forest Import, as a result, has the largest impact on SDG 15 Progress, because the sensitivity index (Sx) has the highest values compared to other variables of change (Population, Forestry Production, Forest Regeneration Rate, and Initial Forest Area) between 2000 and 2050. Many Targets (United Nations, n.d.) under SDG 15 are directly associated with Forest or Protected Area. Adding such Forest Area would have a direct impact on SDG 15 Progress, and such a linear relationship between Forest Area and SDG 15 Progress was defined in the model. The growth of Net Forest Import seems to have a profound impact on Forest stock in Spain. Besides, Forestry Production and Population can also have a large impact on SDG 15 Progress especially during early years when SDG 15 Progress was at a relatively low level. Forest Regeneration Rate has a smaller impact on SDG 15 Progress, but it should not be ignored. This provides potential insights for domestic sustainable forest harvest and international trade. For example, considering population growth and domestic demand for forestry, a baseline for sustainable forest harvest and trade should be set for Spain and the international community, to achieve both domestic and international forest conservation and sustainable development. Challenges such as lack of data and bounded rationality of picturing system structure existed when building the model. Several assumptions have been made to indicate the limit of this model and under what conditions the model worked. It is difficult to overcome existing limitations in this study such as the underestimation of Forestry Production and simplification of Net Forest Import trend. Our goal is to train the model with best fit to as many variables as possible, but modeled Forest Area (key stock of interest) reliably fit real-world data at the cost of sparsely fit of 62 Forestry Production with a simple time-dependent function defined in the model. The change of Forestry Production and Net Forest Import are highly dependent on the market (involving both domestic and international supply and demand) which is not necessarily correlated with time or maybe there is a delay in market response reflected in the change at specific years. External variables could also shape the dynamics of markets such as global economy, transportation delays, pandemic (Li et al., 2017, Amrouss et al., 2017, Golar et al., 2020). Another limitation is although Forest Area and Population (stocks) generally fit the real-world data during calibration and validation processes, Forestry Production and Net Forest Import could be under/overestimated, which could add uncertainty for prediction outcomes. To obtain a better and realistic estimate result and represent a more complete system patterns and processes, future study should consider more elements of both natural and human factors including the elasticity of the forestry market, differentiation between neighboring and distant trade partners, domestic and global economy, domestic and international policies' intersections and interactions, and socioecological shocks (e.g., pandemic, natural disaster, climate change) (Frieden and Martin, 2002, Michinaka et al., 2011, Xu et al., 2020, White and Wulfing, 2024). The modeling approach integrating human and natural systems can be generalized and applied to other countries and SDG Progress simulations at different scales. Despite the limitations, the information offered by the System Dynamics Model in this study would still assist in better adaptive management for Forest management, natural resources policy-making, and sustainable development in the future. To disseminate the modeling results, future work should extend to stakeholder engagement. In this case, by sharing the findings with stakeholders (natural resources management, demographics, and development planning governments, research institutes, associate NGOs, and public communities), obtaining feedback 63 on the model and additional real-world data could help include important parameters and refine the model. The modeling outcomes would also be utilized to facilitate communications among community members, governments and NGOs. Those findings would be informative to stakeholders such as the public and decision makers on land use and management. For instance, the public might have a better understanding of the policy impact – how it would regulate their agricultural and urban land, how it would enhance forest conservation and trade, and whether it would bring them more environmental and economic benefits in a sustainable way. Winning public support is a significant part of conducting sustainable development work. With the outcomes from the model, decision-makers would have more information of benefits and losses at specific time to make cost-effective policies for natural resources management and conservation planning. 64 CHAPTER 5: SYNTHESIS This dissertation focuses on the current challenge for conservation science and the knowledge gap in the United Nations’ SDGs 14 and 15 assessments on a global scale. Adding to the knowledge of SDG assessment, socioecological driver exploration, and mechanism disentanglement, this dissertation broadly contributes to public research and development. Analyzing environmental and social drivers for SDGs 14 and 15 variations offers helpful information for domestic and international development and decision-making. With operationalizing the metacoupling framework, this dissertation explores complex system dynamics (interactions and processes) and conducts scenario analysis to better inform future conservation planning and natural resources sustainable management. The main conclusions of each chapter are summarized below. Chapter 2 evaluated countries’ SDGs 14 and 15 performances between 2010 and 2020, based on the indicator selection and guidance from the United Nations. This delineates how countries did in SDGs 14 and 15 over the past decade, and that through comparisons, which countries did well or poorly. This evaluation step also provides significant data for the following chapters. Global biodiversity conservation and sustainable development made positive progress, but ocean sustainability progress surprisingly slowed after the United Nation Member States adopted SDGs in 2015. Low-income countries lagged in SDGs 14 and 15 progress, and the gap between low-income and high-income countries became wider over time. Chapter 3 identified the important direct and indirect socioecological drivers for countries' SDG variation with multivariate regressions. This chapter further sheds light on the understanding of mechanisms that drive SDGs 14 and 15 variations and places the foundation for the following modeling work. Multiple drivers have mixed expected and unexpected impacts on SDG progress 65 for countries across different income and biodiversity hotspot groups. Fish production has the most profound negative impact on SDG 14 progress for all countries, while the positive and negative impact of drivers on SDG 15 progress varies for countries of different income levels and biodiversity statuses. Synergies and trade-offs between SDGs and their Targets call for special attention for policy making to maximize the common benefits of multiple socioecological sustainability goals while minimizing the conflicting interests. Chapter 4 investigated the drivers for SDG 15 in Spain and framed the interactions among forest, land transformation, population, and SDG 15 with the metacoupling framework. System Dynamics modeling is exercised to simulate the stocks (e.g., forest, population) change over interactions and time. This chapter deepens understanding of drivers for SDG 15 and provides useful policy implications for decision-makers with scenario analysis. SDG 15 progress, associated with forest area, is likely to reach the peak in the mid-2030s and depreciate in the long run as forest harvest increases. Forestry trade and production as well as human population have a major impact on SDG 15 progress. Future natural resources management and conservation planning should be aware of and set up the baseline for potential minimum sustainable forest regeneration and maximum sustainable harvest. In summary, achieving sustainable development everywhere is the goal that requires every country to actively participate and make enormous efforts. To know where countries stand in SDGs 14 and 15 is the first important step. By understanding the drivers for those SDGs variations, countries would make better-informed and collaborative decisions for future sustainable development and conservation planning. Following sustainable practices in natural resources management and holding socioecological baselines are the cornerstone for a prosperous society and a sustainable planet, now and into the future. 66 REFERENCES Aisen, A., & Veiga, F. J. (2013). How does political instability affect economic growth?. European Journal of Political Economy, 29, 151-167. Ali, A. C. A. R. (2019). The effects of political stability on economic growth of the presidental government system. Uluslararası Ekonomi ve Siyaset Bilimleri Akademik Araştırmalar Dergisi, 3(9), 18-31. Amrouss, A., El Hachemi, N., Gendreau, M., & Gendron, B. (2017). Real-time management of transportation disruptions in forestry. Computers & Operations Research, 83, 95-105. Becker, R. A., Chambers, J. M. and Wilks, A. R. (1988) The New S Language. Wadsworth & Brooks/Cole. Bolin, B., Sukumar, R., Ciais, P., Cramer, W., Jarvis, P., Kheshgi, H., ... & Steffen, W. (2000). Global Perspective. In land Use, Land-Use change and Forestry, RT Watson, IR Noble, B Bolin, NH Ravindranath, DJ Verardo, DJ Dokken (eds.) A Special Report of the IPCC. Burns, T. J., Kick, E. L., Murray, D. A., & Murray, D. A. (1994). Demography, development and deforestation in a world-system perspective. International Journal of Comparative Sociology, 35(3-4), 221-239. Burns, T. J., Kick, E. L., & Davis, B. L. (2003). Theorizing and rethinking linkages between the natural environment and the modern world-system: Deforestation in the late 20th century. Journal of World-Systems Research, 357-390. Chung, M. G., & Liu, J. (2022). International food trade benefits biodiversity and food security in low-income countries. Nature Food, 3(5), 349-355. Cohen, J. E. (1995). How many people can the earth support?. The sciences, 35(6), 18-23. da Silva, R. F. B., Viña, A., Moran, E. F., Dou, Y., Batistella, M., & Liu, J. (2021). Socioeconomic and environmental effects of soybean production in metacoupled systems. Scientific Reports, 11(1), 18662. Dewey, J., Husted, T. A., & Kenny, L. W. (2000). The ineffectiveness of school inputs: a product of misspecification?. Economics of education review, 19(1), 27-45. Díaz, S., Settele, J., Brondízio, E. S., Ngo, H. T., Agard, J., Arneth, A., ... & Zayas, C. N. (2019). Pervasive human-driven decline of life on Earth points to the need for transformative change. Science, 366(6471), eaax3100. Didham, R. K., Tylianakis, J. M., Hutchison, M. A., Ewers, R. M., & Gemmell, N. J. (2005). Are invasive species the drivers of ecological change?. Trends in ecology & evolution, 20(9), 470-474. 67 Dietz, T., & Rosa, E. A. (1994). Rethinking the environmental impacts of population, affluence and technology. Human ecology review, 1(2), 277-300 Dietz, T., Frank, K. A., Whitley, C. T., Kelly, J., & Kelly, R. (2015). Political influences on greenhouse gas emissions from US states. Proceedings of the National Academy of Sciences, 112(27), 8254–8259. Dietz, T. (2017). Drivers of Human Stress on the Environment in the Twenty-First Century. Annual Review of Environment and Resources, 42(1), 189–213. Ehrlich, P. R., & Holdren, J. P. (1971). Impact of Population Growth: Complacency concerning this component of man's predicament is unjustified and counterproductive. Science, 171(3977), 1212-1217. Europa Publications (2002). Western Europe 2003. Regional surveys of the world, ISSN 0953- 6906. Psychology Press. p. 559. Feng, Y. (1997). Democracy, political stability and economic growth. British journal of political science, 27(3), 391-418. Fox, J., & Monette, G. (1992). Generalized collinearity diagnostics. Journal of the American Statistical Association, 87(417), 178-183. Freijeiro‐González, L., Febrero‐Bande, M., & González‐Manteiga, W. (2022). A critical review of LASSO and its derivatives for variable selection under dependence among covariates. International Statistical Review, 90(1), 118-145. Frieden, J., & Martin, L. L. (2002). International political economy: Global and domestic interactions. Political science: The state of the discipline, 118-146. Golar, G., Malik, A., Muis, H., Herman, A., Nurudin, N., & Lukman, L. (2020). The social- economic impact of COVID-19 pandemic: implications for potential forest degradation. Heliyon, 6(10). Grossman, G. M., & Krueger, A. B. (1991). Environmental impacts of a North American free trade agreement. Instituto Nacional de Estadística. (2022). Population projections. 2022-2072. https://www.ine.es/en/prensa/pp_2022_2072_en.pdf ISEE System (2022). Stella Architect V2.1.5. https://www.iseesystems.com/ Jorgenson, A. K., & Givens, J. E. (2013). The emergence of new world-systems perspectives on global environmental change 1. In Routledge international handbook of social and environmental change (pp. 31-44). Routledge. 68 Jorgenson, A. K., Fiske, S., Hubacek, K., Li, J., McGovern, T., Rick, T., ... & Zycherman, A. (2019). Social science perspectives on drivers of and responses to global climate change. Wiley Interdisciplinary Reviews: Climate Change, 10(1), e554. Kuznets, S. (2019). Economic growth and income inequality. In The gap between rich and poor (pp. 25-37). Routledge. Leal, P. H., & Marques, A. C. (2022). The evolution of the environmental Kuznets curve hypothesis assessment: A literature review under a critical analysis perspective. Heliyon, 8(11). Li, L., Liu, J., Long, H., de Jong, W., & Youn, Y. C. (2017). Economic globalization, trade and forest transition-the case of nine Asian countries. Forest Policy and Economics, 76, 7-13. Liu, J., Dietz, T., Carpenter, S. R., Folke, C., Alberti, M., Redman, C. L., ... & Provencher, W. (2007a). Coupled human and natural systems. AMBIO: a journal of the human environment, 36(8), 639-649. Liu, J., Dietz, T., Carpenter, S. R., Alberti, M., Folke, C., Moran, E., ... & Taylor, W. W. (2007b). Complexity of coupled human and natural systems. science, 317(5844), 1513- 1516. Liu, J., Hull, V., Batistella, M., DeFries, R., Dietz, T., Fu, F., ... & Zhu, C. (2013). Framing sustainability in a telecoupled world. Ecology and Society, 18(2). Liu, J. (2017). Integration across a metacoupled world. Ecology and Society, 22(4). Liu, J. (2023). Leveraging the metacoupling framework for sustainability science and global sustainable development. National Science Review, 10(7), nwad090. Mace, G., H. Masundire, and J. Baillie. (2005). Biodiversity in ecosystems and human well- being: current state and trends. Hassan, H., Scholes, R. & Ash, N. Island Press: Washington DC 77: 122 Meadows, D. H. (2008). Thinking in systems: A primer. chelsea green publishing. Meadows, D. H., Meadows, D. L., Randers, J., & Behrens, W. W. (2018). The limits to growth. In Green planet blues (pp. 25-29). Routledge. Michinaka, T., Tachibana, S., & Turner, J. A. (2011). Estimating price and income elasticities of demand for forest products: cluster analysis used as a tool in grouping. Forest policy and economics, 13(6), 435-445. Muthukrishnan, R., & Rohini, R. (2016, October). LASSO: A feature selection technique in predictive modeling for machine learning. In 2016 IEEE international conference on advances in computer applications (ICACA) (pp. 18-20). Ieee. Nelson, G. C. (2005). Drivers of ecosystem change: summary chapter. Ecosystems. 69 Nelson, G. C., Bennett, E., Berhe, A. A., Cassman, K., DeFries, R., Dietz, T., ... & Zurek, M. (2006). Anthropogenic drivers of ecosystem change: an overview. Ecology and Society, 11(2). O’brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality & quantity, 41, 673-690. Rockström, J., Steffen, W., Noone, K. et al. (2009). A safe operating space for humanity. Nature 461, 472–475 . Sachs, J. D., Kroll, C., Lafortune, G., Fuller, G., & Woelm, F. (2022). Sustainable development report 2022. Cambridge University Press. Shortreed, S. M., & Ertefaie, A. (2017). Outcome-adaptive lasso: variable selection for causal inference. Biometrics, 73(4), 1111-1122. Stern, P. C., Young, O. R., & Druckman, D. E. (1992). Global environmental change: Understanding the human dimensions. National Academy Press. Stern, D. I. (2004). The rise and fall of the environmental Kuznets curve. World development, 32(8), 1419-1439. Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology, 58(1), 267-288. United Nations (n.d.). The 17 Goals. https://sdgs.un.org/goals Vandermeer, J. (2010). How populations grow: the exponential and logistic equations. Nature Education Knowledge, 3(10), 15. Wang, S., Ji, B., Zhao, J., Liu, W., & Xu, T. (2018). Predicting ship fuel consumption based on LASSO regression. Transportation Research Part D: Transport and Environment, 65, 817-824. White, E. R., & Wulfing, S. (2024). Extreme events and coupled socio-ecological systems. Ecological Modelling, 495, 110786. World Bank Country and Lending Groups (2021). World Bank Country and Lending Groups – World Bank Data Help Desk. https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-worldbank-country- and-lending-groups. Wu, X., Liu, J., Fu, B., Wang, S., & Wei, Y. (2021). Integrating multiple influencing factors in evaluating the socioeconomic effects of payments for ecosystem services. Ecosystem Services, 51, 101348. 70 Xing, Q., Wu, C., Chen, F., Liu, J., Pradhan, P., Bryan, B. A., ... & Xu, Z. (2024). Intranational synergies and trade-offs reveal common and differentiated priorities of sustainable development goals in China. Nature Communications, 15(1), 2251. Xu, Z., Chau, S.N., Chen, X. et al. (2020). Assessing progress towards sustainable development over space and time. Nature 577, 74–78. Xu, Z., Li, Y., Chau, S. N., Dietz, T., Li, C., Wan, L., ... & Liu, J. (2020). Impacts of international trade on global sustainable development. Nature Sustainability, 3(11), 964- 971. York, R., Rosa, E. A., & Dietz, T. (2003). STIRPAT, IPAT and ImPACT: analytic tools for unpacking the driving forces of environmental impacts. Ecological economics, 46(3), 351-365. Zhang, Y., Li, Y., & Liu, J. (2023). Global decadal assessment of life below water and on land. Iscience, 26(4). Zhao, Z., Cai, M., Connor, T., Chung, M. G., & Liu, J. (2020). Metacoupled tourism and wildlife translocations affect synergies and trade-offs among Sustainable Development Goals across spillover systems. Sustainability, 12(18), 7677. Zhao, Z., Cai, M., Wang, F., Winkler, J. A., Connor, T., Chung, M. G., ... & Liu, J. (2021). Synergies and tradeoffs among Sustainable Development Goals across boundaries in a metacoupled world. Science of the Total Environment, 751, 141749. 71 APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 2 Please see Supplementary Material section in: Zhang, Y., Li, Y., & Liu, J. (2023). Global decadal assessment of life below water and on land. Iscience, 26(4). https://doi.org/10.1016/j.isci.2023.106420 The supporting information includes 4 tables in Excel: Table S1. SDG 14 scores & targets. This spreadsheet contains calculations for SDG 14 scores and target values for all countries between 2011 and 2019, and analysis by country’s income level. Table S2. SDG 15 scores & targets. This spreadsheet contains calculations for SDG 15 scores and target values for all countries between 2010 and 2020, and analysis by country’s biodiversity- hotspot and income category. Table S3. SDG data source. This spreadsheet has a detailed description of SDG, target, indicator, sub-indicator, data characteristics, and sources. Table S4. Country class/category. This spreadsheet includes the categorized country (by income, and by biodiversity-hotspot and income) information used in SDGs 14 and 15 analyses. 72 APPENDIX B SUPPORTING INFORMATION FOR CHAPTER 3 Table A3.1. Full regression results for SDG 14. 73 Table A3.2. Full regression results for Target 14.1. 74 Table A3.3. Full regression results for Target 14.5. 75 Table A3.4. Full regression results for Target 14.7. 76 Table A3.5. Full regression results for SDG 15. 77 Table A3.6. Full regression results for Target 15.1. 78 Table A3.7. Full regression results for Target 15.4. 79 Table A3.8. Full regression results for Target 15.5. 80 Table A3.9. Full regression results for Target 15.6. 81 Table A3.10. Full regression results for Target 15.8. 82 Table A3.11. Full regression results for Target 15.9. 83 Supplementary Methods: Raw Data Analysis 1. Raw data preliminary analysis for SDG 14 (1) Drivers’ correlation (including all variables) Figure A3.1. Correlation among independent variables for SDG 14. (2) Form specification analysis Below, From left to right, there are four plots for SDG 14, and five plots for SDG 15. The first plot is (1) original form vs. SDG, and the second plot (2) uses a nonlinear detection package (nlcor) provided by R (Ranjan and Najari, 2020). Although the second plots produced several piecewise lines that were not convenient to incorporate into one regression, they confirmed whether both plot (1) and plot (2) were the same, then a linear relationship between that independent variable and SDG should be used in the regression. The third plot (3) is log form vs. SDG, which is only shown if the original form plot was non-linear, or the distribution of independent variables is obviously skewed (close to zero because the numeric scale range is large). 84 The fourth plot (4) is original or log form (depending on whether plot (1) or (3) is a better fit) of independent variable vs. SDG separated by income level. For SDG 15, plot (4) was separated by biodiversity hotspot, and plot (5) was separated by income level. In total, according to the number of independent variables, there are 21 subplots for SDG 14 and 25 subplots for SDG 15. After this step, we added interaction terms of high-income * independent variable (for SDGs 14 and 15) and biodiversity-hotspot * independent variable (for SDG 15 only, because biodiversity hotspot was based on terrestrial lands, Zhang et al., 2023). The high-income and biodiversity-hotspot were dummy variables, meaning that when a country was a high-income country, the high-income value would be 1 and there would be a coefficient estimate for that independent variable; otherwise, when a country was a low-income country, the high-income value would be 0 and there would not be a coefficient estimate for that independent variable. We only included interaction terms when a major difference between lines was observed in the plots (4) and (5). For instance, if in plot (4) or (5) the lines were significantly different from each other, we generated the interaction terms. For example, when one coefficient is positive, while the other coefficient is negative, we used the interaction terms; if the two lines were paralleled, we considered there was no need to include such interaction terms. 85 Raw vs. SDG 14 Nonlinear detection a. Log vs. SDG 14 Separated raw/log vs. SDG 14 Figure A3.2. Plots of each independent variable vs. SDG 14. From the left to right, the independent variables are in the form of (1) original, (2) with nonlinear detecting result, (3) log, and (4) separated original or log form depending on (1) or (2) by income level. (a) Decision on variable form: Corrupt vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (b) Decision on variable form: log.Crop_ani_exp vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (c) Decision on variable form: log.Crop_ani_imp vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (d) Decision on variable form: log.Fish_exp vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (e) Decision on variable form: log.Fish_imp vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (f) Decision on variable form: log.Fish_prod vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (g) Decision on variable form: log.GDP vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (h) Decision on variable form: GDP_Capita vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (i) Decision on variable form: Gov_effec vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (j) Decision on variable form: Pol_stab vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (k) Decision on variable form: log.Pop_0_14 vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (l) Decision on variable form: log.Pop_15_64 vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (m) Decision on variable form: log.Pop_65 vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (n) Decision on variable form: log.Pop_den vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (o) Decision on variable form: Pop_grow_rate vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (p) Decision on variable form: log.Pop_total vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (q) Decision on variable form: Reg_quali vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (r) Decision on variable form: Rule_law vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (s) Decision on variable form: Temp_change vs. SDG 14 SCORE, 86 Figure A3.2 (cont’d) linear form is the best fit based on the plots above and will be used in the overall regression. (t) Decision on variable form: log.Tourist vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (u) Decision on variable form: Voic_acc vs. SDG 14 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. b. c. d. 87 Figure A3.2 (cont’d) e. f. g. 88 Figure A3.2 (cont’d) h. i. j. 89 Figure A3.2 (cont’d) k. l. m. 90 Figure A3.2 (cont’d) n. o. p. 91 Figure A3.2 (cont’d) q. r. s. 92 Figure A3.2 (cont’d) t. u. 93 2. Raw data preliminary analysis for SDG 15 (1) Drivers’ correlation (including all variables) Figure A3.3. Correlation among independent variables for SDG 15. (2) Form specification analysis Raw vs. SDG 15 Nonlinear detection Log vs. SDG 15 Separated hotspot raw/log vs. SDG 15 Separated income raw/log vs. SDG 15 a. Figure A3.4. Plots of each independent variable vs. SDG 15. From the left to right, the independent variables are in the form of (1) original, (2) with nonlinear detecting result, (3) log, (4) separated original or log form depending on (1) or (2) by biodiversity hotspot status, and (5) separated original or log form depending on (1) or (2) by income level. (a) Decision on variable form: Separate Agri_perc vs. SDG 15 SCORE, linear form is the best fit based on the plots above and 94 Figure A3.4 (cont’d) will be used in the overall regression. (b) Decision on variable form: Separate log.Agri_skm vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (c) Decision on variable form: Separate Corrup vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (d) Decision on variable form: Separate log.Crop_ani_exp vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (e) Decision on variable form: Separate log.Crop_ani_imp vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (f) Decision on variable form: Separate Fore_rents_perc_GDP vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (g) Decision on variable form: Separate log.Fore_skm vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (h) Decision on variable form: Separate Forest_exp vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (i) Decision on variable form: Separate Forest_imp vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (j) Decision on variable form: Separate Forestry_prod vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (k) Decision on variable form: Separate log.GDP vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (l) Decision on variable form: Separate GDP_capita vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (m) Decision on variable form: Separate Gov_effec vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (n) Decision on variable form: Separate Pol_stab vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (o) Decision on variable form: Separate log.Pollution vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (p) Decision on variable form: Separate log.Pop_0_14 vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (q) Decision on variable form: Separate log.Pop_15_64 vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (r) Decision on variable form: Separate log.Pop_65 vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (s) Decision on variable form: Separate Pop_den vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (t) Decision on variable form: Separate Pop_grow_rate vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (u) Decision on variable form: Separate log.Pop_total vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (v) Decision on variable form: Separate Reg_quali vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (w) Decision on variable form: Separate Rule_law vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (x) Decision on variable form: Separate Temp_change vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. (y) Decision on variable form: Separate Voic_acc vs. SDG 15 SCORE, linear form is the best fit based on the plots above and will be used in the overall regression. 95 Figure A3.4 (cont’d) b. c. d. e. 96 Figure A3.4 (cont’d) f. g. h. i. 97 Figure A3.4 (cont’d) j. k. l. m. 98 Figure A3.4 (cont’d) n. o. p. q. 99 Figure A3.4 (cont’d) r. s. t. u. 100 Figure A3.4 (cont’d) v. w. x. y. 101 Figure A3.5. LASSO process for SDG 14. 102 Figure A3.5 (cont’d) 103 Figure A3.6. LASSO process for SDG 15. 104 Figure A3.6 (cont’d) 105 REFERENCES Ranjan, C., & Najari, V. (2020). Package ’nlcor’: Compute Nonlinear Correlations. In:Research Gate. doi:10.13140/RG.2.2.33716.68480. Zhang, Y., Li, Y., & Liu, J. (2023). Global decadal assessment of life below water and on land. Iscience, 26(4). 106 APPENDIX C SUPPORTING INFORMATION FOR CHAPTER 4 107 Table A4.1. Sensitivity Analysis of Forest Area and SDG 15 Progress to Forest Regeneration Rate between 2000 and 2050. 108 Forest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 Progress2000170939.369.50113917170939.369.50113917#DIV/0!#DIV/0!170939.369.50113917#DIV/0!#DIV/0!2001171861.700269.98828733171588.197469.8438419862.837254348.45312609172135.203170.1327326857.2156857344.139205542002172758.634670.46198618172206.214670.1702360331.27305724.15148249173311.929970.7541985528.4759899222.012072162003173630.239270.92230755172793.450470.480373520.7495916116.04816524174469.686271.3656455718.8944380914.633949082004174476.637871.36931689173349.990970.7742997715.4863645511.99449795175608.667571.9671768814.1024781610.943148552005175297.940271.80307217173875.90771.0520523712.327274969.560742822176729.056372.5588889311.226397588.7273305222006176094.20272.22360273174371.214971.313639810.22028527.936982934177830.982473.14085039.3082884357.2490442222007176865.319172.63085355174835.770271.558985868.7145138026.776102515178914.417573.713045977.9376112686.1922315052008177611.181173.02476641175269.422271.788010937.5845203325.904543598179979.328274.275458316.9090965965.3988631382009178331.67173.40527944175672.012872.000631536.7050598985.225884645181025.675474.828066676.1087055654.7811446252010179026.664173.77232687176043.376272.196760056.0009851894.682272197182053.413775.370847125.4680144944.2863987212011179696.027974.12583862176383.33872.376304395.4244747834.236889883183062.490975.903772084.9434806423.8811030062012180339.62174.4657401176691.714772.539167724.943647333.86519295184052.847476.426810074.5060751243.5429086962013180957.293274.79195193176968.313772.685248194.5364307453.550188405185024.415776.939925434.1357006533.2563430872014181548.884675.10438968177212.932172.814438754.1870588993.279737957185977.120277.443078173.8180016823.0103538232015182114.225475.40296372177425.35772.926626953.883969633.044939795186910.876277.936223673.5424504982.7968354472016182653.135975.68757902177605.36573.021694733.6184910122.839117189187825.590478.419312623.3011510242.6097085082017183165.425475.95813501177752.721673.099518343.3839912922.657164065188721.160278.892290773.0880648172.4443236852018183650.892976.21452543177867.181573.159968133.1753121742.495108719189597.473379.355098882.8984950672.2970645372019184109.326176.45663824177948.487873.202908552.9883810832.349815306190454.407879.807672552.7287335392.1650757022020184540.501976.68435552177996.372673.228197972.8199397322.218774883191291.831980.249942182.5758134392.0460726332021184944.185576.89755336178010.55673.235688652.6673502552.099956156192109.603780.681832842.4373328851.9382057032022185320.131177.09610187177990.746673.225226692.5284542061.991696927192907.571381.103264282.3113265251.8399613652023185668.081177.27986504177936.641273.196651982.4014683481.892623868193685.572581.514150812.1961706461.7500890972024185987.732977.44868306177847.891373.149780422.2849060671.801591929194443.401181.914383572.0905116431.6675462542025186278.745577.60237584177724.109873.084407572.1775181761.7176388195180.813882.303834091.9932121031.5914565942026186540.769777.74075898177564.901473.000324652.0782476111.639950129195897.560582.682370231.9033095481.52107862027186773.448477.86364378177369.862372.897318451.9861938461.567832122196593.383883.049856121.8199840361.4557805232028186976.415677.97083708177138.580472.775171311.9005849441.500690008197268.019583.406152121.7425327051.3950207882029187149.297378.06214126176870.63572.633661041.8207553771.438010973197921.195783.751114711.6703495881.3383324372030187291.710678.13735415176565.596772.472560861.7461282931.379350492198552.633384.084596481.6029094571.2853107032031187403.26478.19626895176223.027372.291639341.6762012151.324321323199162.045684.406446041.5397548071.2356029732032187483.55778.23867419175842.479772.090660351.6105344131.27258455199749.138484.716507951.480485271.1889006422033187532.180478.26435366175423.497871.869383031.5487414031.223842268200313.609785.014622711.4247489621.1449324482034187548.715578.27308637174965.616471.627561661.4904811131.177831546200855.149485.300626641.3722353481.103458982035187532.734778.26464644174468.361171.364945711.43545141.134319437201373.439985.574351891.3226693411.0642681282036187483.801178.2388031173931.248171.081279721.3833836391.093098813201868.155185.835626321.2758063761.0271712962037187401.468378.19532057173353.784470.776303251.3340381951.053984876202338.960986.08427351.2314282880.9920002212038187285.280378.13395807172735.467470.449750851.2872006121.016812221202785.515186.320112611.1893398390.9586043032039187134.771478.05446969172075.784870.101352031.2426783830.981432352203207.466986.542958411.1493657770.926848342040186949.466577.95660439171374.214869.730831161.2002982080.947711567203604.45786.752621191.1113483420.8966106032041186728.880177.8401059170630.225769.337907441.1599036490.915529157203976.117686.948906691.0751451310.8677811982042186472.517277.70471268169843.275968.922294871.1213531290.884775859204322.072387.131616061.0406272730.8402606552043186179.872377.55015786169012.813968.483702161.08451820.855352526204641.935687.30054581.0076778550.8139587162044185850.4377.37616917168138.278268.021832731.0492820530.827168978204935.313487.455487720.9761905620.7887932942045185483.664477.18246889167219.096967.536384611.0155382260.800143007205201.802487.596228850.9460685020.7646895612046185079.039276.96877378166254.688167.027050410.9831894770.774199511205440.990387.722551390.9172231850.7415791632047184636.007576.73479502165244.459366.493517260.9521468080.749269738205652.455587.834232670.8895736250.7193995312048184154.011876.48023814164187.807765.935466790.9223286040.725290626205835.766887.93104510.8630455620.6980932792049183632.483776.20480299163084.119865.352575020.8936598780.702204222205990.484188.012756050.8375707740.6776076762050183070.843975.9081836161932.771664.744512340.8660716140.679957174206116.15788.079127860.8130864780.657894184Estimates under 90% of Forest Regeneration RateSx_90Sx_110Estimates under 110% of Forest Regeneration RateYearForest AreaSDG 15 Progress Table A4.2. Sensitivity Analysis of Forest Area and SDG 15 Progress to Initial Forest Area between 2000 and 2050. 109 Forest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 Progress2000170939.369.50113917153845.3760.4733048210.769853949188033.2378.5289735110.7907763172001171861.700269.98828733154494.267460.816007630.989563060.763041356189229.133179.160567020.9905118730.7845830512002172758.634670.46198618155113.322861.142950010.9790625210.756108088190403.946479.781022350.9809659280.778280082003173630.239270.92230755155702.602461.45416680.9685060060.749062666191557.875980.39044830.9713690970.7718751512004174476.637871.36931689156262.158861.749685890.9579007880.741913249192691.116780.988947890.9617279890.7653756812005175297.940271.80307217156792.029662.029527080.9472537910.734667631193803.850881.576617270.9520489010.7587887552006176094.20272.22360273157292.196862.293680910.9365713920.727333045194896.207282.153524550.9423376290.752120952007176865.319172.63085355157762.481862.542052990.9258588990.719915644195968.156482.719654120.9325989990.7453778582008177611.181173.02476641158202.698462.774545030.9151214130.712421359197019.663883.274987780.9228376480.7385648722009178331.67173.40527944158612.652662.991054530.9043638340.704855906198050.689483.819504360.9130580310.7316871882010179026.664173.77232687158992.141463.191474350.8935908610.697224791199061.186884.353179380.9032644190.724749812011179696.027974.12583862159340.952863.375692460.8828070010.68953331200051.10384.875984780.893460910.7177575542012180339.62174.4657401159658.864763.543591610.8720165670.681786556201020.377385.38788860.8836514250.7107150512013180957.293274.79195193159945.644863.695049060.8612236880.673989426201968.941685.88885480.8738397170.7036267512014181548.884675.10438968160201.049863.829936360.8504323120.66614662202896.719386.378842990.8640293740.6964969272015182114.225475.40296372160424.825363.948119150.8396462070.658262653203803.625686.857808290.8542238250.6893296842016182653.135975.68757902160616.705464.049456940.8288689760.650341855204689.566487.32570110.8444263410.6821289592017183165.425475.95813501160776.41264.133802970.818104050.64238838205554.438887.782467050.8346400460.6748985272018183650.892976.21452543160903.655364.201004080.8073547040.63440621206398.130588.228046780.8248679130.6676420092019184109.326176.45663824160998.132764.250900550.7966240550.626399159207220.519688.662375930.8151127770.6603628722020184540.501976.68435552161059.529464.283326020.7859150710.618370882208021.474489.085385010.8053773380.6530644392021184944.185576.89755336161087.517464.29810740.7752305770.610324879208800.853689.496999330.7956641610.645749892022185320.131177.09610187161081.756364.295064770.7645732560.602264498209558.505889.897138970.7859756870.6384222712023185668.081177.27986504161041.892364.274011350.7539456590.594192945210294.269890.285718740.7763142350.6310844952024185987.732977.44868306160967.525164.234735710.7433500730.58611315211007.940690.662630420.7666818850.6237392272025186278.745577.60237584160858.214464.177005320.7327885660.578027815211699.276691.027746350.7570805150.6163889232026186540.769777.74075898160713.510164.100582540.7222631150.569939541212368.029391.380935420.7475119230.6090359462027186773.448477.86364378160532.952664.005224510.7117756090.561850831213013.944191.722063040.7379778260.6016825742028186976.415677.97083708160316.071963.890683110.7013278520.553764094213636.759392.050991050.7284798650.5943309952029187149.297378.06214126160062.388163.756704830.6909215670.545681648214236.206592.36757770.7190196060.5869833162030187291.710678.13735415159771.410963.603030730.6805583970.537605721214812.010492.671677570.7095985420.5796415642031187403.26478.19626895159442.639563.429396350.6702399080.529538455215363.888592.963141540.7002180980.5723076862032187483.55778.23867419159075.562563.235531630.6599675910.521481909215891.551693.241816740.6908796280.5649835542033187532.180478.26435366158669.657963.021160830.6497428660.513438061216394.702893.50754650.6815844230.5576709642034187548.715578.27308637158224.392762.786002450.6395670820.505408809216873.038393.760170290.6723337110.5503716442035187532.734778.26464644157739.222862.529769180.6294415210.497395977217326.246793.999523710.6631286550.5430872522036187483.801178.2388031157213.59362.25216780.61936740.489401313217754.009394.22543840.6539703640.5358193762037187401.468378.19532057156646.936861.952899110.6093458730.481426496218155.999894.437742040.6448598840.5285695422038187285.280378.13395807156038.676361.631657860.5993780330.473473135218531.884294.636258280.6357982120.5213392132039187134.771478.05446969155388.221861.288132680.5894649140.46554277218881.321194.820806710.6267862850.5141297912040186949.466577.95660439154694.97260.922005990.5796074930.457636878219203.960994.99120280.6178249930.5069426172041186728.880177.8401059153958.313860.532953930.5698066920.449756875219499.446595.147257880.6089151750.4997789772042186472.517277.70471268153177.621860.120646270.5600633820.441904113219767.412695.288779090.600057620.4926401032043186179.872377.55015786152352.258659.684746390.5503783790.434079887220007.48695.415569330.5912530720.485527172044185850.4377.37616917151481.574559.224911120.5407524540.426285434220219.285695.527427230.5825022310.4784413042045185483.664477.18246889150564.907258.740790710.5311863280.418521938220402.421795.624147070.5738057530.471383582046185079.039276.96877378149601.581958.232028740.5216806750.410790528220556.496695.705518810.565164250.4643550252047184636.007576.73479502148590.910957.698262060.5122361280.403092282220681.104295.771327970.5565782980.457356622048184154.011876.48023814147532.193657.139120660.5028532740.395428228220775.830195.821355630.5480484310.4503892982049183632.483776.20480299146424.716456.554227620.493532660.387799347220840.251195.855378350.5395751450.4434539522050183070.843975.9081836145267.752355.943199030.4842747940.380206573220873.935595.873168170.5311589040.43655143Estimates under 90% of Initial Forest AreaSx_90Estimates under 110% of Initial Forest AreaSx_110YearForest AreaSDG 15 Progress Table A4.3. Sensitivity Analysis of Forest Area and SDG 15 Progress to Population between 2000 and 2050. 110 Forest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 Progress2000170939.369.50113917170939.369.50113917#DIV/0!#DIV/0!170939.369.50113917#DIV/0!#DIV/0!2001171861.700269.98828733171938.243470.02871213-224.5292-173.1320349171785.157169.94786252-204.026546-157.30185032002172758.634670.46198618172908.973570.54138481-114.912794-88.74458073172607.97370.38241708-104.151473-80.413151382003173630.239270.92230755173851.95371.03940151-78.3127614-60.56871668173407.643470.80474781-70.8204365-54.753398852004174476.637871.36931689174767.311271.52283057-60.0249578-46.49052585174184.055471.2147951-54.1212265-41.897470322005175297.940271.80307217175655.181671.99174244-49.0698822-38.05743974174937.087271.61249454-44.0716117-34.160497212006176094.20272.22360273176515.697772.44620765-41.7784143-32.44474632175666.607471.99777677-37.3477399-28.983613972007176865.319172.63085355177348.989872.88629505-36.5673004-28.43345966176372.47572.37056732-32.5333343-25.276566682008177611.181173.02476641178155.181973.31207007-32.6490691-25.41727673177054.53972.73078645-28.9160064-22.490953632009178331.67173.40527944178934.388673.72359331-29.58793-23.06066038177712.638373.07834908-26.0982886-20.320799122010179026.664173.77232687179686.713174.12091919-27.1232395-21.16292348178346.601473.41316467-23.8409301-18.581922372011179696.027974.12583862180412.244574.50409484-25.0896188-19.59672712178956.246573.73513715-21.9912943-17.156818642012180339.62174.4657401181111.056574.87315911-23.3771483-18.27743418179541.381174.04416485-20.4474157-15.966989612013180957.293274.79195193181783.05375.22806129-21.9140345-17.14981565180101.801974.34014047-19.1385835-14.957997322014181548.884675.10438968182428.098875.56872996-20.6489932-16.17442901180637.283774.62294502-18.0139935-14.090724762015182114.225475.40296372183046.051675.89508999-19.5437988-15.32187336181147.540274.8924273-17.0354923-13.335782182016182653.135975.68757902183636.761476.20706218-18.5693764-14.56978494181632.277975.14843226-16.1746543-12.671272822017183165.425475.95813501184200.070176.50456296-17.7032193-13.90085082182091.195775.39080101-15.4098748-12.08055422018183650.892976.21452543184735.810676.78750409-16.9276339-13.30145968182523.985475.61937068-14.7244476-11.550757492019184109.326176.45663824185243.806277.05579235-16.2285196-12.76076336182930.331175.83397442-14.1052589-11.071781332020184540.501976.68435552185723.870977.30932936-15.5945022-12.2700104183309.909176.03444133-13.541878-10.635591662021184944.185576.89755336186175.807977.54801131-15.0163062-11.82206372183662.38876.22059636-13.0259113-10.235724122022185320.131177.09610187186599.410277.77172883-14.4862937-11.41104578183987.428576.39226032-12.5505345-9.8669237782023185668.081177.27986504186994.459877.98036677-13.9981196-11.03207348184284.683276.54924977-12.1101484-9.5248798162024185987.732977.44868306187360.727578.17380409-13.5461369-10.68079394184553.796876.691377-11.700395-9.2062425992025186278.745577.60237584187697.972778.35191375-13.1253641-10.35336232184794.405676.81844997-11.3178205-8.9083620742026186540.769777.74075898188005.943378.51456254-12.7316502-10.04657546185006.137776.93027225-10.9594612-8.6289643652027186773.448477.86364378188284.375278.66161109-12.3615149-9.757748557185188.612877.02664298-10.6227603-8.3660877032028186976.415677.97083708188532.992778.79291369-12.0120242-9.484619311185341.44277.1073568-10.3055005-8.1180310522029187149.297378.06214126188751.507978.90831833-11.6806931-9.225272653185464.227877.17220384-10.0057503-7.8833125492030187291.710678.13735415188939.610979.00766135-11.3654763-8.978134898185556.564377.2209696-9.72181924-7.6606356862031187403.26478.19626895189096.927879.09074532-11.0649623-8.742127957185618.036677.253435-9.45222253-7.4488615452032187483.55778.23867419189223.076479.15736826-10.7778946-8.516292492185648.220877.26937621-9.19565063-7.2469859142033187532.180478.26435366189317.665279.20732355-10.5031518-8.299772344185646.684277.26856472-8.95094465-7.0541202622034187548.715578.27308637189380.294279.24039986-10.2397301-8.09180134185612.985177.25076718-8.71707568-6.8694758632035187532.734778.26464644189410.554279.25638111-9.98672869-7.891692091185546.672577.21574542-8.4931274-6.6923504712036187483.801178.2388031189408.026879.25504634-9.74333736-7.698826413185447.286177.1632564-8.27828139-6.5221170612037187401.468378.19532057189372.284579.23616968-9.50882544-7.512647119185314.356477.09305208-8.07180475-6.3582143072038187285.280378.13395807189302.889979.19952029-9.2825328-7.332650949185147.404377.00487948-7.87303949-6.2001384472039187134.771478.05446969189199.396679.14486225-9.06386175-7.158382465184945.941176.89848052-7.68139347-6.0474363452040186949.466577.95660439189061.348179.07195455-8.85227014-6.98942875184709.468576.77359203-7.49633267-5.8996995192041186728.880177.8401059188888.278278.98055096-8.64726532-6.825414794184437.478576.62994568-7.31737444-5.7565590032042186472.517277.70471268188679.710778.87040003-8.44839886-6.66599947184129.453176.46726792-7.14408174-5.6176808862043186179.872377.55015786188435.159678.74124498-8.25526198-6.510872006183784.864376.28527989-6.97605809-5.4827624572044185850.4377.37616917188154.128378.59282364-8.06748142-6.359748884183403.17476.08369743-6.81294318-5.3515288252045185483.664477.18246889187836.110278.42486841-7.88471595-6.212371109182983.83475.86223096-6.65440904-5.2237299832046185079.039276.96877378187480.588178.23710617-7.70665316-6.068501795182526.285675.62058544-6.50015668-5.0991382192047184636.007576.73479502187087.034378.02925822-7.53300669-5.92792403182029.959675.35846031-6.34991315-4.9775458452048184154.011876.48023814186654.910477.80104022-7.36351375-5.790438971181494.276475.0755494-6.20342889-4.8587631972049183632.483776.20480299186183.667377.55216213-7.1979329-5.655864158180918.645674.77154092-6.06047548-4.742616862050183070.843975.9081836185672.744977.28232813-7.03604211-5.524032180302.465974.44611734-5.92084355-4.628948099Sx_110YearForest AreaSDG 15 ProgressEstimates under 90% of PopulationSx_90Estimates under 110% of Population Table A4.4. Sensitivity Analysis of Forest Area and SDG 15 Progress to Forestry Production between 2000 and 2050. 111 Forest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 Progress2000170939.369.50113917170939.369.50113917#DIV/0!NA170939.369.50113917#DIV/0!NA2001171861.700269.98828733171994.375270.05835711-129.5358223-99.88370026171729.025269.91821755-117.6689951-90.712449342002172758.634670.46198618173026.924270.60367824-64.39258744-49.72895869172490.34570.32029411-58.44781244-45.117233162003173630.239270.92230755174037.1371.13719924-42.67243798-33.00374563173223.348370.70741586-38.702207-29.912496872004174476.637871.36931689175025.164371.65901079-31.80823973-24.63611315173928.111271.079623-28.82566774-22.305558052005175297.940271.80307217175991.185572.1691965-25.28656823-19.61166379174604.694971.43694785-22.89688007-17.737876622006176094.20272.22360273176935.298372.66781175-20.9362708-16.25892291175253.105771.77939371-18.94206554-14.689929932007176865.319172.63085355177857.448873.15482802-17.82683418-13.86152527175873.189472.10687909-16.11530335-12.510477722008177611.181173.02476641178757.577673.63021395-15.49299716-12.06128719176464.784672.41931887-13.99363388-10.873897362009178331.67173.40527944179635.619474.09393502-13.6762831-10.65921516177027.722672.71662387-12.34207506-9.5992866032010179026.664173.77232687180491.50274.54595318-12.22160209-9.535912318177561.826272.99870055-11.01963849-8.5781020152011179696.027974.12583862181325.146474.9862266-11.03026151-8.615396821178066.909473.26545064-9.936601091-7.7412698142012180339.62174.4657401182136.465475.41470932-10.03646295-7.847013209178542.776673.51677089-9.033148239-7.0427393382013180957.293274.79195193182925.363975.83135107-9.194654225-7.195691171178989.222573.75255279-8.267867359-6.4506283082014181548.884675.10438968183691.73876.236097-8.472296068-6.636379226179406.031173.97268235-7.611178024-5.9421628952015182114.225475.40296372184435.47576.62888755-7.845525458-6.150705454179792.975974.17703989-7.041387035-5.500641292016182653.135975.68757902185156.452677.00965821-7.296445295-5.724889997180149.819174.36549983-6.542222828-5.1135363562017183165.425475.95813501185854.539377.37833942-6.81136729-5.348394527180476.311574.5379306-6.101243053-4.7712677452018183650.892976.21452543186529.593577.73485646-6.379645504-5.013021764180772.192374.6941944-5.70876859-4.466383432019184109.326176.45663824187181.463778.0791293-5.992873788-4.712299507181037.188674.83414719-5.357158124-4.1929995672020184540.501976.68435552187809.988178.41107255-5.644327282-4.441049359181271.015674.95763848-5.04029745-3.9464084882021184944.185576.89755336188414.994978.73059537-5.328560754-4.195078612181473.376275.06451135-4.753237099-3.7227987452022185320.131177.09610187188996.301579.03760144-5.041119149-3.970956424181643.960675.15460229-4.491926446-3.5190512832023185668.081177.27986504189553.715279.33198889-4.778321274-3.765848012181782.446975.2277412-4.25301927-3.3325891122024185987.732977.44868306190086.998979.61363255-4.537098387-3.577389836181888.466875.28373358-4.03372573-3.16126352025186278.745577.60237584190595.878479.88238756-4.314871716-3.403595475181961.612575.32236411-3.831701477-3.00326862026186540.769777.74075898191080.072280.13810533-4.109458879-3.242783793182001.467375.34341263-3.644962674-2.8570761782027186773.448477.86364378191539.291980.38063349-3.919000872-3.093522529182007.604875.34665406-3.471818917-2.721384112028186976.415677.97083708191973.241780.60981584-3.741903589-2.954583732181979.589575.33185832-3.310821478-2.595076122029187149.297378.06214126192381.618580.82549228-3.576792984-2.824908628181916.976275.29879025-3.160720921-2.4771896682030187291.710678.13735415192764.111681.02749876-3.422477805-2.703579396181819.309675.24720954-3.020434408-2.3668903592031187403.26478.19626895193120.402981.21566724-3.277920417-2.589796428181686.125175.17687066-2.889018575-2.2634512932032187483.55778.23867419193450.166581.3898256-3.142212628-2.482859882181516.947675.08752277-2.765647888-2.1662362512033187532.180478.26435366193753.068781.54979764-3.014556311-2.382154564181311.29274.97890969-2.649596605-2.074685972034187548.715578.27308637194028.76881.69540297-2.894246868-2.287137499181068.66374.85076976-2.5402244-1.9883068162035187532.734778.26464644194276.914881.82645702-2.780660231-2.197327584180788.554674.70283586-2.436963864-1.9066614422036187483.801178.2388031194497.151781.94277095-2.673241532-2.112297038180470.450674.53483525-2.339310509-1.8293609412037187401.468378.19532057194689.112982.04415159-2.57149572-2.031664163180113.823774.34648956-2.24681429-1.7560583342038187285.280378.13395807194852.424482.13040146-2.474979709-1.955087325179718.136274.13751469-2.159072442-1.6864430242039187134.771478.05446969194986.703982.20131863-2.383295727-1.882259779179282.83973.90762076-2.07572341-1.6202361672040186949.466577.95660439195091.560782.25669675-2.296085778-1.812905347178807.372373.65651203-1.996441598-1.5571866822041186728.880177.8401059195166.595582.29632497-2.213026527-1.746774669178291.164773.38388684-1.920933212-1.4970678832042186472.517277.70471268195211.400582.31998787-2.133825469-1.683642026177733.633873.0894375-1.848932229-1.4396745712043186179.872377.55015786195225.559182.32746545-2.058217107-1.62330259177134.185672.77285027-1.780197373-1.3848205372044185850.4377.37616917195208.645982.31853307-1.985960066-1.565570054176492.214272.43380527-1.714509155-1.3323364142045185483.664477.18246889195160.226782.29296139-1.9168343-1.510274577175807.102272.0719764-1.651667551-1.2820677972046185079.039276.96877378195079.858282.25051631-1.850638824-1.457260995175078.220271.68703125-1.591489845-1.2338736312047184636.007576.73479502194967.088282.19095898-1.787189675-1.406387264174304.926971.27863106-1.533808807-1.1876247852048184154.011876.48023814194821.455282.11404566-1.726318158-1.357523094173486.568570.84643063-1.478471062-1.1432028142049183632.483776.20480299194642.488482.01952774-1.667869265-1.310548758172622.479170.39007823-1.425335703-1.1004988692050183070.843975.9081836194429.707781.90715165-1.611700323-1.265354023171711.980269.90921555-1.374273031-1.059412748Sx_110YearForest AreaSDG 15 ProgressEstimates under 90% of Forestry ProductionSx_90Estimates under 110% of Forestry Production Table A4.5. Sensitivity Analysis of Forest Area and SDG 15 Progress to Net Forest Import between 2000 and 2050. 112 Forest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 ProgressForest AreaSDG 15 Progress2000170939.369.50113917170939.369.50113917170939.3NA170939.369.50113917170939.3NA2001171861.700269.98828733171833.74569.97352328171833.745474.045193171889.655570.00305138171889.6555431.04120422002172758.634670.46198618172704.496570.43339418172704.4965246.4395574172812.772770.49057817172812.7727224.12691972003173630.239270.92230755173551.719270.88083871173551.7192171.0255413173708.759170.9637764173708.7591155.5686512004174476.637871.36931689174375.565571.31593751174375.5655133.7020274174577.7171.42269628174577.71121.63818812005175297.940271.80307217175176.174671.73876401175176.1746111.6546854175419.705871.86738034175419.7058101.5951542006176094.20272.22360273175953.631972.14936327175953.631997.28465068176234.772272.29784219176234.772288.531500382007176865.319172.63085355176707.863272.54769623176707.863287.34149929177022.77572.71401088177022.77579.492263772008177611.181173.02476641177438.789172.93372087177438.789180.20685305177783.573173.11581195177783.573173.006233732009178331.67173.40527944178146.323773.30739181178146.323774.98933273178517.018273.50316708178517.018268.263025022010179026.664173.77232687178830.374173.66866001178830.374171.16288191179222.954273.87599372179222.954264.784443282011179696.027974.12583862179490.839774.0174724179490.839768.40308779179901.21674.23420483179901.21662.275535852012180339.62174.4657401180127.612274.35377168180127.612266.50601855180551.629874.57770853180551.629860.550921852013180957.293274.79195193180740.574374.67749599180740.574365.34562783181174.01274.90640786181174.01259.496027792014181548.884675.10438968181329.600174.98857877181329.600164.85087396181768.16975.22020059181768.16959.046252622015182114.225475.40296372181894.554175.28694849181894.554164.99402144182333.896775.51897894182333.896759.176386592016182653.135975.68757902182435.291375.57252854182435.291365.78641033182870.980575.8026295182870.980559.896737342017183165.425475.95813501182951.656675.84523706182951.656667.28034641183379.194376.07103296183379.194361.254861512018183650.892976.21452543183443.484876.10498678183443.484869.57774596183858.30176.32406409183858.30163.343403862019184109.326176.45663824183910.600376.35168497183910.600372.84826946184308.05276.56159151184308.05266.316605072020184540.501976.68435552184352.81776.58523333184352.81777.36346291184728.186776.7834777184728.186770.421335872021184944.185576.89755336184769.938176.8055279184769.938183.56117348185118.432976.98957882185118.432976.055609622022185320.131177.09610187185161.755977.01245901185161.755992.1729638185478.506377.17974472185478.506383.884519062023185668.081177.27986504185528.051877.20591125185528.0518104.4975055185808.110377.35381884185808.110395.088633782024185987.732977.44868306185868.562977.3857457185868.5629123.0567645186106.902977.51162043186106.9029111.96067992025186278.745577.60237584186182.988377.55180349186182.9883153.4482337186374.502777.65294818186374.5027139.58933022026186540.769777.74075898186471.019777.70392184186471.0197211.0390592186610.519777.77759612186610.5197191.9445912027186773.448477.86364378186732.341477.84193393186732.3414358.6558928186814.555377.88535362186814.5553326.14182132028186976.415677.97083708186966.629977.96566891186966.62991508.674059186986.201477.97600525186986.20141371.6127542029187149.297378.06214126187173.553678.07495177187173.5536-609.3601273187125.04178.04933076187125.041-553.873152030187291.710678.13735415187352.773578.16960333187352.7735-242.2925459187230.647878.10510497187230.6478-220.17503492031187403.26478.19626895187503.942178.24944015187503.9421-147.0650777187302.58678.14309774187302.586-133.60459772032187483.55778.23867419187626.70478.31427451187626.704-103.4898758187340.410178.16307387187340.4101-93.99078652033187532.180478.26435366187720.695478.3639143187720.6954-78.6097326187343.665378.16479302187343.6653-71.372487262034187548.715578.27308637187785.544378.39816303187785.5443-62.58009128187311.886678.14800971187311.8866-56.800082212035187532.734778.26464644187820.870278.41681971187820.8702-51.4312698187244.599378.11247317187244.5993-46.664792172036187483.801178.2388031187826.283978.41967886187826.2839-43.25554846187141.318478.05792733187141.3184-39.232315852037187401.468378.19532057187801.387778.40653041187801.3877-37.02257474187001.548977.98411074187001.5489-33.565978292038187285.280378.13395807187745.775178.37715966187745.7751-32.12723999186824.785477.89075648186824.7854-29.115673052039187134.771478.05446969187659.030778.33134724187659.0307-28.19097108186610.512277.77759215186610.5122-25.537247332040186949.466577.95660439187540.730178.26886905187540.7301-24.96491453186358.202877.64433974186358.2028-22.604468272041186728.880177.8401059187390.4478.18949619187390.44-22.27884043186067.320277.49071561186067.3202-20.162582522042186472.517277.70471268187207.717978.09299496187207.7179-20.01242832185737.316577.31643041185737.3165-18.102207862043186179.872377.55015786186992.111877.97912673186992.1118-18.07827169185367.632877.12118899185367.6328-16.34388352044185850.4377.37616917186743.160877.84764796186743.1608-16.41137873184957.699376.90469039184957.6993-14.828526262045185483.664477.18246889186460.39477.69831008186460.394-14.96244768184506.934876.6666277184506.9348-13.511316142046185079.039276.96877378186143.331577.53085952186143.3315-13.6934223184014.746976.40668804184014.7469-12.357656612047184636.007576.73479502185791.483277.34503755185791.4832-12.5744751183480.531876.12455248183480.5318-11.340431822048184154.011876.48023814185404.349777.14058031185404.3497-11.58191027182903.67475.81989597182903.674-10.438100342049183632.483776.20480299184981.421376.91721873184981.4213-10.69667598182283.546275.49238724182283.5462-9.6333416652050183070.843975.9081836184522.178676.67467843184522.1786-9.903287157181619.509375.14168878181619.5093-8.912079244Sx_110YearForest AreaSDG 15 ProgressEstimates under 90% of Net Forest ImportSx_90Estimates under 110% of Net Forest Import