Coupled social and land use dynamics affect dietary choice and agricultural land-use extent

Global food demand tripled from 1961 to 2013, driven by population growth and increased per capita consumption. This massive increase in demand is directly linked to land use, as land is the primary source of global food supply. In 2013, an estimated 3.5 billion hectares of land were used for food production, representing 72% of agricultural land. Future food demands necessitate agricultural expansion or intensification, but these actions pose major ecological threats like deforestation, habitat fragmentation, and increased greenhouse gas emissions. Agricultural intensification's future is uncertain due to yield ceilings for major crops in some countries and decelerating yield growth globally. Existing mathematical models often focus on the supply side or use fixed demand trajectories, neglecting the coupled dynamics between land use and dietary choices influenced by social factors. Eco-conscious consumption evolves endogenously within populations, driven by social and economic factors, and land-use dynamics and socially-influenced dietary choices are coupled through two-way feedback. This paper introduces a modeling framework to couple country-level dynamics of eco-conscious dietary decision-making under social learning processes with country-level land-use dynamics, aiming to demonstrate how such coupled dynamics differ from isolated approaches and provide insight into how these processes alter projected global land use and dietary trends. The goal is not to create policy-ready projections, but rather to showcase the potential of this framework for future applications. A minimal model was chosen for easier data fitting and insight generation.

Mathematical models of sustainable food systems are increasingly researched, but those focusing on sustainable agricultural technologies tend to concentrate on the supply side. Demand-side models often assume future demand trajectories independent of model variable evolution. Sophisticated land system models used in IPCC reports, for instance, employ scenarios for homogenous dietary consumption patterns as inputs and do not study system-induced drivers of human consumption behaviors. The importance of incentivizing sustainable consumption has been acknowledged, highlighting the significant influence of dietary patterns on global land use trajectories. Studies indicate that individuals incorporate environmental factors into dietary decisions; however, there's limited research into how these dietary pattern shifts evolve within populations due to social and economic factors, especially their response to changing land use. This study leverages evolutionary game theory models of social learning dynamics to capture how individuals learn behaviors from one another, an approach previously applied to other coupled systems like the global climate and terrestrial ecosystems but not yet to the coupled dynamics of global land use for food production and human dietary decision-making.

A mathematical model is developed to describe a social learning process influencing dietary behavior. The model incorporates a two-way feedback loop between land use and dietary practices: dietary practices impact land use, and land-use trends subsequently influence dietary behavior. For each country, the model considers the proportion of the population below the poverty line (consuming a sustenance diet) and the proportion above the poverty line (choosing between eco-conscious and land-intensive diets). An eco-conscious diet is defined as requiring between *l(t)* and *cl(t)* hectares per capita, while a land-intensive diet requires more than *cl(t)* hectares. The model utilizes an evolutionary game theory framework to capture social learning and socio-economic incentives for dietary shifts. Individuals switch diets based on the difference in personal utility between eco-conscious and land-intensive diets, influenced by factors like average per capita income, social learning rate (describing how quickly behaviors are adopted), and global land use. The model is parameterized using country-level data (1961-2013) from a previously published model that accounts for food imports and exports. The parameters are estimated by fitting the model to data that accounts for per capita land used for dietary consumption, which is calculated accounting for differential yields of food sources and excludes the land equivalent of food wastage. For the 153 countries studied, projections of global land use until 2100 are made for 20 scenario combinations. These combinations involve five SSP (Shared Socioeconomic Pathway) scenarios providing population and income projections, and four future agricultural yield trajectories (defined by a parameter *f*, from 0.2 - high yield, to 0.8 - low yield). In the primary results, the SSP2 scenario (continuation of current trends) is assumed. Baseline projections utilize estimated parameters, while sensitivity analyses explore variations from these baseline values.

Global-level model projections reveal that the proportion of eco-conscious consumers is predicted to rise until mid-21st century in response to growing global land use, then decline as global land use decreases. The peak global land use falls between 3.25 and 4.5 billion hectares (depending on yield scenarios), and the proportion of eco-conscious consumers peaks between 45% and 53%. This pattern persists across most SSP scenarios, except in SSP3 at low yields, where eco-conscious consumption saturates. This behavior stems from the utility function governing dietary choices. Initially, the net utility favors land-intensive diets, but the growing global land use shifts the utility advantage towards eco-conscious diets (between 2020 and 2050). After 2050, declining land use reverses the utility advantage, leading to a decline in eco-conscious consumption. The global population decline in the latter half of the century further contributes to this decline. Social dynamics partially counteract benefits from higher yields. Higher yields reduce land pressure, lessening the perceived need for eco-conscious diets, thus resulting in lower proportions of eco-conscious consumers in high-yield scenarios. Regional analysis reveals heterogeneity; Africa is projected to continuously increase land use until 2100 across all yield scenarios. Asia experiences land-use increases until 2030 then continuous decline, reflecting changes in eco-conscious consumption and population size. Europe's land use peaks around 2040, showing continuous growth in eco-conscious consumption until 2100. Oceania and the Americas experience declining land use, with varying eco-conscious consumer proportion trends depending on yield scenarios. This heterogeneity reflects variations in baseline parameters, the impact of global land use, and differences in population projections. Sensitivity analyses demonstrate the substantial impact of socio-economic factors on peak global land use. For low-yield scenarios, peak land use and peak year are more sensitive to social dynamics than in high-yield scenarios. Changes in social parameters governing the pace and desirability of change have significant impacts on land use, especially in low yield scenarios. Synergistic effects are found where simultaneous changes to multiple socioeconomic parameters are needed to alter land use projections. Modest improvements in incentives for eco-conscious diets could yield significant land-use benefits, particularly in low-yield scenarios. Increasing both income and non-income net benefits of eco-conscious diets is optimal for reducing peak global land use and delaying the peak year.

The study demonstrates the significant impact of coupled social-and-land processes on land-use projections, highlighting the need for incorporating these dynamics into global land-use modeling alongside economic forces and climate change impacts. Individual diets are influenced by several factors (religion, health concerns, urbanization, etc.) subsumed into the model's parameters. The model shows how coupled dynamics can lead to giga-hectare impacts on land use, especially under low yield and high population scenarios. Analyses explored how social parameter changes minimize peak land use and year under various scenarios. Increasing the net benefits of adopting eco-conscious diets is crucial for reducing peak global land use, and reducing social learning rates can accentuate this effect. Social dilemmas, like the clash between individual and social optima in adopting eco-conscious diets, warrant further consideration, accounting for global inequities. The model's simplicity (e.g., binary diet classification, no institutional factors) presents potential limitations. Future extensions could incorporate population heterogeneity, social norms, and social learning between countries, and consider perceived climate change risk. The study also did not explicitly account for the effect of changing yields on food prices, which could influence dietary behavior. Finally, the assumption that populations respond to global land use rather than local land use could be culturally contingent and requires further investigation.

This research demonstrates the substantial impact of coupled social and land use dynamics on global land-use projections, highlighting the importance of including these feedbacks in future models. The findings suggest that increasing the net benefits (both income and non-income related) of adopting eco-conscious diets is a crucial strategy for mitigating land use pressures. Future research should focus on enhancing model complexity to capture more nuanced social processes, incorporate other relevant socio-economic factors, and explore the interaction of dietary choices with climate change dynamics.

The model's simplicity, such as the binary classification of dietary behavior and the assumption of homogeneous behavior within countries, represents a limitation. The limited availability of social data could lead to overfitting, highlighting the need for better data collection on social aspects of dietary choices. The model also does not explicitly address the influence of food prices, potentially influenced by yields and macroeconomic factors, on dietary behaviors, nor does it consider the possibility that spared land from reduced food production may be used for other resource extraction purposes. The assumption that populations are responsive to global rather than local land use is also crucial and may warrant further investigation.