Fig 1.
Schematic diagram of the agent-based model.
The central cycle (thick black arrows) represents the farm production process, with each cycle occurring over one year (i.e. one timestep). Agents (farming households), corporate agriculture (represented as an exogenous forcing term) and the environment (1ha plots) are shown in grey and green. Agent decisions are shown in blue. Scenario options are shown in orange. Health-related outcomes are shown in red. See text for further details.
Table 1.
Key environment and agent factors, their initial values, and how they change over timea.
Table 2.
Prices for key factors, their initial values, and how they change over timea.
Table 3.
Agricultural policy scenarios and associated settings for proportion preferring agroecology and global price transmission.
Fig 2.
Total food production and local food price in the absence of climate change and agricultural policies.
The plots show total food production (top two plots) and local food price (bottom two plots) (y-axes) under combinations of global price transmission (x-axes) and proportion preferring agroecology (line colour) after 25 (left plots) and 50 years (right plots), in the absence of climate change or specific agricultural policies. For the y-axes, total food production and local food price are shown as the mean result across 250 runs under each combination of factors. For the x-axes, global price transmission is an elasticity such that a value of 0.6, for example, means that a 1% rise in global food price would cause a 0.6% rise in local food price. For the line colours, a value of 0.2, for example, means that 20% of orphan farmers prefer to develop via agroecology and 80% via entrepreneurial farming.
Fig 3.
Income slopes by farming style in the absence of climate change and agricultural policies.
The plots show income slopes for orphan, agroecology, and entrepreneurial farms (top to bottom plots, respectively) (y-axes; scale differs for each style) under combinations of global price transmission (x-axes) and proportion preferring agroecology (line colour) after 25 (left plots) and 50 years (right plots), in the absence of climate change or specific agricultural policies. For the y-axes, the income slopes are the gradient (units = $ per year) of mean farm net income by farming style over the previous ten years, shown as the mean result across 250 runs under each combination of factors. For the x-axes, global price transmission is an elasticity such that a value of 0.6, for example, means that a 1% rise in global food price would cause a 0.6% rise in local food price. For the line colours, a value of 0.2, for example, means that 20% of orphan farmers prefer to develop via agroecology and 80% via entrepreneurial farming.
Fig 4.
Total food production, local food price, and converted and abandoned farms, under the agricultural policy and climate scenarios.
The plots show time-series for total food production, local food price, and the number of converted and abandoned farms (top to bottom plots, respectively) under the four agricultural policy scenarios, for no, low, and high climate change (left to right plots, respectively). For the y-axes, all results are shown as the mean value across 250 runs. For the coloured lines, the four scenarios are: (i) ‘Entre’ in which agricultural policy favour entrepreneurial farming, 25% of farmers prefer agroecology, and global price transmission is 0.75; (ii) ‘Entre eroding’ is as for ‘Entre’ except policy support erodes over time; (iii) ‘Peasant’ in which policy favours peasant farming, 75% of farmers prefer agroecology, and global price transmission is 0.25; and, (iv) ‘None’ in which policy favours neither farming style, 50% prefer agroecology, and global price transmission is 0.5 (Table 3).
Fig 5.
Nutrition, labour, income inequality, net farm income, and real land productivity under the agricultural policy and climate scenarios.
The plots show time-series for nutrition in orphan farming households, labour, income inequality, net farm income, and real land productivity (top to bottom plots, respectively) under the four policy scenarios, for no, low and high climate change (left to right plots, respectively). For the y-axes, all results are shown as the mean value across 250 runs. For the coloured lines, the four scenarios are: (i) ‘Entre’ in which agricultural policy favour entrepreneurial farming, 25% of farmers prefer agroecology, and global price transmission is 0.75; (ii) ‘Entre eroding’ is as for ‘Entre’ except policy support erodes over time; (iii) ‘Peasant’ in which policy favours peasant farming, 75% of farmers prefer agroecology, and global price transmission is 0.25; and, (iv) ‘None’ in which policy favours neither farming style, 50% prefer agroecology, and global price transmission is 0.5 (Table 3). The outcomes are as follows—‘Orphan nutrition’: the average proportion of a basic diet being consumed in orphan farming households; ‘Labour’: the total number of full-time equivalent workers across all farming households; ‘Income Gini’: income inequality, where a higher value means greater inequality; ‘Mean net farm income’: average net farm income across all households over the previous five years;. ‘Real land productivity’: an indicator of farming intensity after removing the contribution of purchased inputs to Gross Value Product.
Table 4.
Farm development processes and their implications for nutrition, under the agricultural policy and climate scenarios at 25 and 50 years.
Fig 6.
Total food production and local food price at year 50: Sensitivity analysis.
The plots show total food production (top row) and local food price (bottom row) (y-axes) after 50 years, under no, low, and high climate change (left to right plots, respectively), under different yield ratio assumptions (x-axes), for the peasant and entrepreneurial policy scenarios with different climate sensitivity assumptions (line colour). For the y-axes, values are shown as the mean result across 250 runs under each combination of factors (Note that the y-axis scale for local food price differs in each plot). For the x-axes, the numbers show the ratio of agroecology to entrepreneurial maximum production (e.g. 1:4 means maximum production for agroecology is 25% that of entrepreneurial farming). For the coloured lines, ‘entre’ and ‘peasant’ refer to the entrepreneurial and peasant scenarios, respectively, and, the numbers refer to agroecology climate sensitivity relative to entrepreneurial farming, where: 0.9 means agroecology losses are 10% lower, 1 means losses are equal, and 1.1 means agroecology losses are 10% higher.
Fig 7.
Alternative climate-undernutrition model structures based on different standpoints.
Panel A shows the general structure underlying previous global-level climate-undernutrition models, which link together a series of component models. A pathway is traced from climate, to crops, to trade, to nutrition-related health outcomes. Production and consumption are separated, with the upstream component models calculating food availability (i.e. production). The health-impact model then combines the latter with socioeconomic variables to estimate consumption-related outcomes in entire populations. Panel B shows the model structure adopted in this paper. Constellations of producer-consumer farmers practicing different styles of farming develop over time under given climate, policy, style preference, and price transmission scenarios. Different facets of the farm development process give rise to patterns of hunger and other health-supporting conditions in the rural community.
Table 5.
The four key patterns seen in the results of the ‘Style preference and globalization’ model runs, their implications, and comments on the underlying mechanisms.
Table 6.
The four key patterns seen in the results of the ‘Climate change and agricultural policy’ model runs, their implications, and comments on the underlying mechanisms.