African food system and biodiversity mainly affected by urbanization via dietary shifts

Africa’s urban population has more than doubled since the early 2000s and is projected to double again by 2050. Urban area expansion (historically ~5% yr−1) exceeds population growth and directly converts cropland, grassland, and natural habitats to urban use, with global analyses suggesting food production losses, biodiversity declines and increased land-use change emissions. However, most assessments focus on direct conversion and overlook indirect land-use displacement (agricultural expansion elsewhere) and urbanization-related dietary shifts. In Africa, urbanization is closely associated with rising rice consumption due to sociocultural food environments, convenience and market access. Standard SSP-based demand trajectories may thus understate future African rice demand and its environmental footprint (water use, methane). This study aims to integrate urban land-use displacement and dietary shifts into a unified assessment to quantify their individual and combined effects on land-use change, staple crop production, biodiversity (BII), rice availability, trade, methane emissions and water use to 2050.

Prior work often treats urban expansion as a generic land-cover conversion, emphasizing direct effects on land and emissions while neglecting indirect displacement dynamics. Emerging local-scale studies recognize urbanization’s complexity, including peri-urban shifts toward higher-value products and comparative advantage changes, but continent-scale analyses remain sparse. Urbanization typically correlates with higher consumption of animal-based foods and lower cereal shares globally; in Africa, it specifically drives strong growth in rice demand due to convenience and food retail changes. Common SSP-based demand projections anticipate decelerating rice demand growth in developing regions, potentially missing Africa’s urbanization-driven surge. Rice cultivation imposes notable environmental costs (irrigation water, methane emissions), so underestimating demand leads to underestimation of environmental impacts. Biodiversity assessments frequently use BII, but many studies do not incorporate fragmentation and degradation effects, potentially understating biodiversity losses. These gaps motivate a holistic, systems approach integrating displacement and dietary dynamics in large-scale models.

The study uses an adapted version of the Global Biosphere Management Model (GLOBIOM), a spatially explicit partial equilibrium model for agriculture, forestry and bioenergy operating on 212,707 simulation units (5 arcmin resolution) differentiated by biophysical and geopolitical attributes. Demand is exogenously driven by population and GDP per capita via income elasticities (SSP-specific) and endogenously responds to own-price elasticity; cross-price effects are not modeled. A wide set of food/feed crops and livestock commodities is included; fruits and vegetables are not explicitly modeled. Trade assumes nonlinear costs and homogeneous goods. Urban expansion: Spatially explicit projections (1 km) of urban expansion to 2050 are taken from published SSP-consistent maps. Using Google Earth Engine, these maps are overlaid with CGLS-LC100 land-cover data to estimate, for each simulation unit and time step, the proportion of each land-cover class converted to urban area. The proportion is treated as exogenous, applied prior to each model’s land allocation so that displacement is endogenously simulated in the same time step. If projected conversion exceeds remaining area of a land-cover class, exogenous conversion is capped at the available extent. Direct effects are quantified from the exogenous urban conversions by land-cover class; combined direct+displacement effects are computed as the difference in modeled land-use change between runs with and without urban expansion. Production losses are computed as: (a) direct losses by multiplying exogenous crop area converted by predicted yields; and (b) direct+displacement losses by comparing production between urban-expansion and baseline runs. Dietary shift parameterization: Household microdata from the World Bank’s LSMS surveys across multiple African countries are used to estimate per capita rice consumption from 7-day recalls, scaled to annual values and aggregated separately for urban and rural households. The urban-to-rural consumption ratio (α) is computed per survey wave and country, then aggregated to regional and continental levels using population weights. This α is used to disaggregate the representative consumer in GLOBIOM’s demand framework for rice, imposing higher per capita rice demand in urban populations consistent with observed differences. Regions lacking LSMS coverage retain baseline demand trajectories. Biodiversity: Biodiversity impacts are assessed with the Biodiversity Intactness Index (BII) linked to land-use areas per simulation unit using land-use-specific BII values adapted from PREDICTS for GLOBIOM. BII captures shifts due to land-use composition changes but not within-class degradation or fragmentation. Outcomes assessed: land-use change (cropland, grassland, managed/primary forest, other natural land, wetlands, urban), staple crop production changes, rice-specific indicators (production, consumption per capita, imports, producer price), agricultural methane emissions, total agricultural water use, and BII. Scenarios follow SSP1, SSP2, and SSP3 baselines with added urban expansion and dietary demand effects.

• Urban expansion extent and sources (SSP2, Africa, to 2050): 3.28 Mha converts to urban area. Sources: cropland 50.4%; grassland 12.7%; primary forest 12.0%; managed forest 10.0%; other natural land 12.2%; wetlands 2.7%. Total conversion is similar under SSP1 (3.38 Mha) and SSP3 (3.07 Mha).
• Displacement effects: Urban expansion induces displacement to meet demand for crops, livestock and wood. Modeled expansions include cropland into grasslands/other natural lands (~1.05 Mha), grassland into primary forest/other natural land (~0.54 Mha), and managed forest into primary forest (~0.28 Mha). Indirect (displacement) land-use changes exceed direct effects for some natural land classes.
• Staple crop production losses: Cropland directly converted is small (0.63% of continental cropland under SSP2). Net production losses at the continental scale remain minor (<0.7%) for major staples when displacement is accounted for. If only direct effects are considered, losses are overestimated. Regional contrast: Northern Africa shows larger direct losses for rice (−1.88%), millet (−2.07%), and wheat (−4.53%), which are mostly compensated by displacement; maize exhibits a substantial loss that worsens with displacement (−7.29%). In sub-Saharan Africa, direct effects are <1% and largely offset by displacement.
• Natural lands and biodiversity: Relative decreases in natural lands are small at the continental scale (<0.5%). Under SSP2, direct effects on primary forest + other natural lands total about −0.09%; including displacement increases this to about −0.25%. Despite small continental averages, localized biodiversity impacts (BII) can be substantial; in the Lagos–Ibadan region, urbanization (expansion + demand) lowers BII by up to 4.5% under SSP1.
• Urbanization-driven rice demand: LSMS data show urban households consume 1.5× the rice of rural households. Incorporating this gap raises projected rice demand vs. baseline (e.g., +9.2% under SSP1; +8.4% under SSP2). Most of the additional demand is met by increased African production (70–80%), with the remainder via higher imports (up to +2 Mt). Continental rice production rises by about +8.0% (SSP2), increasing producer prices (+1.05% under SSP2) as production expands into less-productive areas.
• Environmental footprint: Agricultural CH4 emissions increase by ~2.4% (+12.2 MtCO2e yr−1 under SSP2), mostly from rice. Effects on total agricultural water use are small: urban land expansion slightly reduces water use by cropland abandonment; demand effects are minor (<0.5%) under SSP2/SSP3 and slightly positive under SSP1 depending on water constraints.
• Scenario sensitivity: Urbanization impacts (demand and expansion) are generally more pronounced under SSP1 than SSP2/SSP3 due to higher urbanization and lower baseline rice demand/production in SSP1.

By explicitly modelling both direct urban land conversion and indirect displacement, the study finds that continental-scale food production losses from Africa’s urban expansion to 2050 are limited (<1%) because displacement largely compensates for cropland lost to cities. However, displacement shifts pressure onto natural lands (primary forests and other natural lands), with indirect effects exceeding direct conversion in some cases. Considering urbanization-induced dietary shifts is pivotal: higher urban rice demand drives higher African rice production, imports, prices, and methane emissions, altering trade balances and environmental outcomes that are missed by standard demand trajectories. Biodiversity impacts may be modest on average but can be severe locally in rapidly urbanizing zones, suggesting that urban planning must integrate indirect land-use dynamics and biodiversity safeguards. The results underscore the importance of representing consumer heterogeneity and commodity-specific dietary contexts in global models to improve projections of land allocation, trade and environmental impacts, thereby supporting better-informed policy on food security, land-use planning and conservation.

The study provides a holistic assessment of African urbanization’s impacts by integrating land-use displacement and urban dietary shifts into a global economic-land-use model. It shows that while direct urban expansion minimally reduces continental staple production, displacement can intensify pressure on natural lands. Accounting for higher urban rice consumption substantially increases projected rice production, imports, prices and methane emissions, highlighting urbanization as a key driver of food-system and environmental change. Policy and planning should incorporate indirect land-use dynamics and evolving dietary patterns. Future research should improve survey coverage for underrepresented regions, represent consumer heterogeneity and cross-price effects, consider product differentiation (e.g., preferences for imported vs. local rice), explicitly model drivers of dietary change (income, education, retail environments), and integrate additional urbanization-linked resource demands (e.g., water for domestic/industrial use) and biodiversity fragmentation/degradation into assessment frameworks.

• Affordability and preference heterogeneity: Rice is modeled as a homogeneous good; known urban preferences for imported rice imply imperfect substitutability, potentially overstating domestic production responses and understating import increases.
• Dietary data coverage: LSMS coverage is uneven; some regions rely on single-country surveys (e.g., Ethiopia for Rest of Central-Eastern Africa; Malawi for Rest of Southern Africa), limiting representativeness.
• Demand modeling: Cross-price effects between foods are not modeled; consumer heterogeneity beyond urban–rural (e.g., income, education) is not fully captured; fruits and vegetables are not explicitly modeled.
• Biodiversity metric: BII does not capture within-class degradation or habitat fragmentation, potentially underestimating biodiversity losses.
• Urban expansion assumption: Exogenous urban conversion is applied independent of future land-cover trajectories within time steps (with capping), which may not capture feedbacks between land cover and urban growth.
• Trade representation: One-way bilateral trade with homogeneous goods may not capture quality differentiation and policy frictions.
• Supply response uncertainty: The capacity of African agricultural and supply chains to realize modeled displacement responses may be constrained by institutional and financial barriers.
• Water competition: Interactions with rising non-agricultural water demand linked to urbanization are not explicitly modeled under SSP2/SSP3 constraints.