Focus on global–local–global analysis of sustainability

The UN Sustainable Development Goals have been widely adopted for purposes of framing long run, global sustainability challenges. Of the 17 goals, eight are closely tied to essential food, land, water, climate, and biodiversity outcomes that are already under intense pressure. Can the future demands for food, fuel, clean water, biodiversity, climate change mitigation and poverty reduction be reconciled? Ensuring the long-term sustainability of land and water resources, even as we seek to meet the world economy's growing demands, requires informed management of the complex networks of policies, infrastructure and technologies that connect the food and resource nexus. In addressing this challenge, local, meso, and global scale perspectives are required (figure 1). The local scale refers to the socio-economic and environmental conditions in the neighborhood of individual agents, e.g., landowners with their surrounding environment (Johnson et al 2023b). The meso-scale can be thought of as regional-scale, with regions varying from small watersheds to entire river basins (e.g. the Mississippi River Basin), or from an individual county to entire countries. At the other end of the spectrum from local scale, the global scale encompasses the entire world. Processes interact across these local, meso, and global scales. For example, local decisions of individual agents can affect food, land, water, and other resources at larger meso-scales, and the aggregation of local decisions affects conditions all the way up to the global scale. Local conditions are in turn shaped by meso-scale and global-scale environmental processes such as climate change, as well as economic conditions as reflected in prices determined by global markets. One cannot understand the ramifications of climate change or other shocks to global markets without understanding the collection of local decisions, and one cannot formulate local decisions without understanding how local conditions are driven by larger meso-scale and global-scale forces. Informed decision-making requires multi-scale information that integrates from local to global scales.

Zoom In Zoom Out Reset image size

This special issue is the outcome of a workshop held at Purdue University in April 2022. It comprises thematic syntheses of five overarching dimensions of the global-to-local-to-global (GLG) challenge, including: climate change, ecosystems and biodiversity, governance, water resources and cyber-infrastructure (CI). In addition, there are eight applications of GLG analysis to land and water sustainability challenges, ranging from environmental stress in the Amazon River Basin to groundwater depletion in the United States. This overview article draws out the main themes and insights from these papers. We have divided this synthesis into four sections. The first focuses on global drivers of local sustainability stresses. The second section seeks to draw insights from these papers about how sustainability policies can generate spillovers wherein interventions in one locality (and sector) have impacts in other locations (or sectors)—both proximate and across the globe. We then turn to a theme that emerged from our workshop as an area needing special research attention—the 'missing middle' in GLG analyses, i.e. explicit representation of meso-scale processes, including both environmental and socio-economic phenomena, and how they mediate the interactions and feedbacks from local to regional to global scales and vice versa, global to regional to local. The final section of our review draws insights about GLG transdisciplinary analysis that can bridge domains and ultimately deliver the kind of comprehensive assessments required to inform sustainability policies.

Land and water sustainability challenges invariably require location-specific analysis. Differences in climate, soils, hydrology, culture, economic conditions, and governance dictate locally differentiated solutions. This is why the vast majority of studies undertaken on this topic refer to specific locations. However, providing long term policy advice to local and regional private and public sector decision makers, without a global context, can be misleading. This point is effectively illustrated in the special issue paper by Haqiqi et al (2023a) focusing on groundwater sustainability stresses in the Western United States. They show that the most important drivers of US groundwater demand emanate from overseas. US population, income and biofuels demands contribute less than one-third of the total demand drivers. Restricting any sustainability analysis to the national borders could result in a seriously flawed set of policy recommendations.

While Haqiqi et al use a single integrated framework to capture these GLG interactions, Mosnier et al (2023) offer an alternative approach to conveying global change to local analyses. Their FABLE framework allows national models to independently develop sustainable pathways driven by national goals. National exports and imports are then fed into an online platform which iteratively balances global supply and demand, thereby informing the national pathways and ensuring global consistency of these local decisions.

Climate change is an important global driver that impinges on land and water sustainability at local levels. The synthesis paper in this issue by Baldos et al (2023) points out that the global to local connection is generally well covered in the climate impacts literature. However, they highlight the need for improved analysis of local and global tipping points whereby the climate—or socio-economic—system shifts into a different regime. These can be both positive as well as negative in terms of planetary and societal well-being. The paper by Banerjee et al (2022) focuses on a specific regional tipping point of global significance—namely the potential demise of the Amazon rainforest ecosystem. The drivers behind this systemic risk are both global in nature (climate change, international trade) and local (fires and conversion of nature lands to agriculture). Their analysis employs a suite of models, which integrate regional economic-environmental modeling with a land change science model and a local ecosystem modeling approach. They conclude that the Amazon tipping point could be averted with a mix of policies, including reductions in deforestation, intensification of existing agriculture and improving fires management. This suite of policies would generate nearly US$340 billion in additional wealth for the region.

Instead of focusing on long run, global drivers such as population growth and climate change, Haqiqi et al (2023b) focus on global changes that are of shorter duration and generate immediate stress on local social and ecological systems. They focus on the confluence of two global shocks, the COVID-19 pandemic, and the presence of widespread drought, and address how these shocks interact. These global shocks reduce greenhouse gas emissions, as was widely observed during the pandemic (Diffenbaugh et al 2020), however, they also increased food insecurity globally (although not in every region). Haqiqi et al (2023b) find that market-mediated adaptations in cropland use, water withdrawals and international trade reduce the impact on undernourishment by about 40%. While global trade had the biggest adaptation impact, integrated adaptation is important as the changes in trade are dependent on adaptation at the other levels. The role of these adaptations is not uniformly distributed and the adaptations can, in some regions, exacerbate food insecurity.

This special issue includes a suite of papers that illustrate the importance of considering spillovers in evaluating the wider regional, national, or global impacts of local decisions, and in designing effective policies. Spillovers are situations whereby a particular driver or a policy has impacts on other regions or sectors. In the policy context, a spillover is positive when the overall benefit of the intervention is enhanced by the spillover, and is negative (sometimes called leakage) when the overall benefit is diminished by the spillover ⁵ . In the context of the GLG analysis that forms the theme of this special issue, the implication is that spillovers are generated by often poorly-understood wider impacts of local or sector-specific decisions and their subsequent interactions and feedbacks across scales. Compared to the global to local linkages discussed in the prior section, the local to global spillover effects are less well covered in the literature (Hertel et al 2019).

Several studies in this special issue (Liu et al 2023, Ray et al 2023, Haqiqi et al 2023a, 2023b) use an integrated modeling framework to simulate such local responses and feedbacks between scales. These studies integrate gridded biophysical models (hydrological models, crop models) with a global economic model that resolves grid-level responses in the region(s) of interest. In their integrated model, which simultaneously solves local and global responses, global or national-level policies or drivers lead to farmer responses at the pixel-level. These responses are aggregated back to the national or global level and fed to the global economic model, which calculates changes in input and commodity prices, and implications for both farmer and consumer responses, which further feedback to the pixel-level.

Liu et al (2023) use such a modeling framework to evaluate the influence of US national policies to reduce nitrogen (N) losses from fertilizer application on corn. They investigate the impact of four policies: a national tax on N loss; improving N-use efficiency; controlled drainage, and wetland restoration. They find that the spatially targeted policies—wetland restoration and controlled drainage—are the most effective. But these policies, by raising crop prices and lowering fertilizer prices, increase crop production in areas outside of the current production zones, i.e., spatial spillovers. The authors show that spillovers can be mitigated by additionally implementing policies with uniform national coverage such as the N tax. Indeed, a combined policy of wetland drainage, N loss tax and increased N-use efficiency can reduce N loss by 30% with a less than 2% increase in crop prices and marginal national impacts on yields.

Cisneros et al (2023) review the literature to identify challenges in designing effective sustainability policies. These authors review policies to mitigate losses in biodiversity and ecosystem services. In addition to the challenge of spatial spillovers, they additionally point to the importance of considering heterogeneity in local responses that can be difficult to predict, the issue of additionality (i.e., figuring out what would have happened in the absence of the policy), and the issue of unintended consequences (which are essentially unexpected spillovers into non-targeted sectors).

Integrating location-specific analysis within a global context is critical for guiding policies but is insufficient for adequately representing the interactions of social, economic, and environmental processes across scales. Many phenomena exist only at 'meso-scales' that operate in between local and global scales. They are too spatially variable to be considered global, but have influence that reach beyond the local level and cause locations or sectors to be interdependent. Examples include regional labor markets, which cause the location of firms and households to be linked, and hydrological connectivity, which causes the functional integrity of downstream waters to be dependent on upstream disturbances. In the absence of meso-scale models to capture these spatial dependencies, any representation of global-local interactions may be biased and generate misleading results.

Johnson et al (2023b) provide a careful exposition of the challenges of addressing 'the missing middle,' including model complexity, data needs, and uncertainties. They consider these challenges in the context of five types of models used in sustainability science and propose fruitful directions for future research to improve explicit representation of meso-level phenomena. Because incorporating meso-scale models greatly increases complexity, they advocate for constructing meso-scale models that are focused on specific regions and then embedding these more detailed models into global models to ensure consistency. This provides a feasible approach to improving the policy relevance of global models by representing nation or regional specific policies, economic sectors, or stakeholder preferences with more detail than otherwise could be included.

Cultice et al (2023) focus on spatial economic interactions that manifest at meso-scales, including the spatial flows and interactions of people, goods, and information that link economic activities across locations. These interactions arise due to distance, which creates similarities and differences in accessibility across space, or from spillovers that create spatial dependencies. Spatial equilibrium is an important framework for ensuring consistency at the meso-scale by explicitly accounting for how spatial interactions are mediated by and influence prices at local, meso, or global scales. However, solving spatial equilibrium models is computationally intensive and this approach has not been integrated into global integrated assessment models (IAMs) with more detailed representations of biophysical and land use processes.

Ray et al (2023) provide a compelling example of why accounting for a key meso-scale phenomenon—spatially delineated labor markets—is important. Most IAMs assume that workers are freely mobile across space, which ignores spatial frictions, e.g., from commuting or policies that restrict migration. Instead, Ray et al account for more realistic conditions in which the mobility of farmworkers is limited and compare these results to a model with perfect labor mobility. Varying availability of workers at specific locations alters the optimal producer response to crop price shocks and groundwater sustainability policies. They demonstrate that failure to account for such labor market rigidities leads to an overstatement of the effectiveness of conservation policies and an understatement of equity impacts.

Banjeree et al (2022) provide another example of the importance of incorporating local spatial dependencies at meso-scales. They use the Integrated Economic-Environmental Modeling (IEEM) Platform linked with spatial land use, land cover change, and ecosystem services modeling to approximate the economic, natural capital and ecosystem services impacts of an environmental tipping point in the Amazon. The IEEM Platform allocates equilibrium land uses to the meso-scale (states and departments). These quantities are downscaled to a local (grid cell) level through a spatial land use change model that uses suitability rules to allocate land use accounting for meso-level spatial interactions, including agglomeration, neighborhood effects, and spatial ecological functions.

Addressing multiple and interacting sustainability challenges requires a systems approach capable of integrating across a variety of domains and intellectual disciplines. For example, transdisciplinary integration is needed to analyze how agro-ecosystems can be managed to simultaneously contribute to food, forage, fiber, and fuel production, poverty alleviation, income generation and secure livelihoods, gender equality, local and regional air and water quality, biodiversity conservation, and climate change mitigation and adaptation. These topics range across the social and natural sciences. No single discipline has the breadth and expertise to do all of this on its own. Just as the previous sections have discussed how sustainability challenges require integrating across spatial scales from global to local, and from local to global, with special attention to models linking through the meso-scale, these challenges also require a transdisciplinary systems approach.

The papers in this special issue highlight important cross-disciplinary linkages for food, land, and water resources. Many of these papers stress the importance of integrating across climate, ecology, economics, hydrology, land use models, among others. Liu et al (2023) combine an economic model (SIMPLE-G) with an agro-ecosystem model (Agro-IBIS) to analyze the effects of policies to manage nitrogen runoff from agriculture on water quality and agricultural markets. Troy et al (2023) discuss the linkages of water quality and quantity with agricultural markets, land use, ecosystems, and health impacts, and describe an approach for improving the sustainability of agricultural water systems.

Several papers in this special issue integrate economic and biophysical models to analyze how changes in market conditions or government policies affect ecosystem services, climate change, and biodiversity, and how these changes in turn affect macroeconomic variables, such as GDP, employment, and trade. Banerjee et al (2022) use the IEEM Platform that integrates economic, land use, and ecosystem service models, to analyze whether the Amazon rainforest will likely cross a tipping point. The papers by Johnson et al (2023b), and Cisneros-Pineda et al (2023), though focused mainly on issues of modeling the meso-scale, and linkages across local-global and global-local scales, also illustrate the importance of linking economic, land use, and biophysical models. These papers discuss the development of an 'Earth-Economy model' that links the GTAP general equilibrium model of the economy with the InVEST model of ecosystem services showing how economic drivers can lead to ecosystem change, which changes the flow of ecosystem services, and how this then affects macroeconomic performance.

Because of the interlinked nature of socio-ecological systems, policies focused on achieving a policy goal in one domain can conflict with policies to achieve goals in different domains. Johnson et al (2023a) show that policies enacted independently without paying attention to interlinkages can cause 'policy collisions' in which mutual negative interference between policies causes each policy to fail to reach its intended targets. An example of potentially conflicting policies includes regulating water quality by reducing nitrogen runoff from agriculture in the Mississippi River Basin to reduce hypoxia ('dead zone') in the Gulf of Mexico, on the one hand, and regulating agricultural production dependent on ground water from the Ogalala Aquifer to prevent depletion of the aquifer, on the other. Stringent regulation of nitrogen to reduce hypoxia could increase cultivation on lands reliant on groundwater from the Ogalala Aquifer, and reduced production on lands dependent on the aquifer could increase cultivation further east worsening nitrogen runoff. Examples from Costa Rica and China show how more integrated planning at landscape levels aimed at jointly providing multiple benefits can at least partially overcome policy collisions and improve outcomes in multiple dimensions. On a related theme, Mosnier et al (2023) highlight the interconnectedness of food and land use systems and the need for more collaborative planning across sectors and scales. They show that coordinated planning could result in large gains in agricultural productivity that would allow greater natural forest cover recovery and increases in carbon sequestration.

This special issue brings together insights from a variety of different disciplines for the analysis of sustainability through a GLG lens. Whether the approach involves combinations of agronomy, hydrology, climate science, biodiversity, geography, economics, or other disciplines, the GLG perspective offers important insights. It is clear that, without fine-scale, local analysis, interventions focusing on land and water sustainability will likely be misguided. But formulating such policies without the broader, national/global context is also problematic—both from the point of view of the global drivers of local sustainability stresses, as well as to capture unanticipated spillovers. Finally, because local and global systems are connected to and mediated by meso-scale processes, accounting for key meso-scale phenomena is critical for characterizing GLG interactions and processes.

Sustainability policies, in the context of GLG analysis, should be effective, efficient, and equitable. Papers in this special issue touch on all aspects of these criteria. For example, Liu et al (2023) found that wetland restoration was the most effective in reducing nitrate loss in the Mississippi Basin, but installing controlled drainage was the most efficient (i.e., nitrate loss per $ invested). Both Liu et al (2023) and Haqiqi et al (2023b) find that coupled strategies, that are often synergistic, are much more effective in achieving sustainability goals compared to single strategies. The study by Ray et al (2023) finds that, under a restricted labor mobility situation, regions targeted for the sustainable groundwater use policy will see sharp decrease in wages, but the spillover effects will increase wages in non-targeted regions. Cisneros-Pineda et al (2023), in their review, point to a study examining different policies to reduce deforestation in Malaysia and Indonesia, which found that domestic policy action by Malaysia & Indonesia would be beneficial for those countries, but tariff policy action by the rest of the world would be harmful. Future GLG analyses should aim to explore all three of these policy dimensions: effectiveness, efficiency and equity.

As additional complexity is added to GLG analyses—particularly in the dimensions of time (dynamic modeling), spatial resolution, and uncertainty (stochastic analysis)—the computing and data technologies and overall cyber-infrastructure (CI) ecosystem must also evolve to enable these innovations. The digital revolution currently underway is generating vast amounts of data to support expansion of transdisciplinary sustainability analysis along these three dimensions. However, to realize its potential, it is necessary to harness the power of advanced CI. In their contribution to this volume, Song et al (2023) review recent developments in CI that can facilitate such advances—from low-hanging fruit such as creating more robust, normalized metadata for geographic location and integrating heterogeneous geospatial data to developments at the research frontier, including causal discovery and non-linear dynamic systems modeling using knowledge-guided artificial intelligence and machine learning methods. Additionally, they highlight the critical role that CI learning and workforce development play in greater accessibility and usability of data and modeling, as well as the need for serious work on human-centered user interfaces, that are adaptive to diverse backgrounds and needs, to ensure sustainable development that is ethical and just.

In reviewing the future research agendas laid out by the authors in this special issue, it is clear that there is great scope for increasing the complexity of GLG analysis. Baldos et al (2023) highlight the need for capturing likely tipping points through more elaborate biophysical modeling. Cultice et al (2023) emphasize the need for explicit modeling of spatial equilibria. Haqiqi et al (2023a) and Ray et al (2023) highlight the role of behavioral heterogeneity in GLG modeling. Troy et al (2023) highlight the need to capture local water rights and allocation rules as well as monitoring local water quality. But incorporating all of these complexities into any one model is likely to be prohibitively complex and might well preclude any understanding of how the entire framework operates. Combining local-, meso-, and global-scales of analysis, integrating across disciplines, as well as consideration of dynamics and uncertainty, greatly increases the complexity of analysis. Increased complexity can outstrip data and modeling capabilities, slow down research, make results more difficult to understand and interpret, and complicate effective communication with decision-makers and other users of the analyses. As Johnson et al (2023b) illustrate in their figure 2, a limited computational budget requires making trade-offs about where to opt for more versus less detail, e.g. in spatial resolution, number of economic sectors, or biophysical processes. For this reason, it is important to identify when, and where, additional complexity is required. We believe that each GLG framework needs to be 'fit for purpose'—designed to capture complexity that is key to the problem at hand. Ray et al (2023) offer an instructive example of this by showing how the incorporation of 'meso-scale' labor market interactions—in the form of commuting zones—changes key outcomes of a groundwater sustainability policy. Future guidance regarding required complexity will be most useful as the emerging field of GLG analysis of sustainability matures.