Intensifying rice production to reduce imports and land conversion in Africa

The study addresses how much scope exists to raise rice yields in Africa and to what extent yield intensification can reduce reliance on imports and limit cropland expansion by 2050. Rice demand in Africa has surged due to rapid population growth and dietary shifts toward rice, with imports now supplying about 40% of consumption and cropland expanding ~0.4 M ha per year. Yet average yields (≈2.4 Mg ha⁻¹) have remained low and stagnant compared with other rice-producing regions. Given the economic and geopolitical risks of heavy import dependence and the environmental costs of land conversion and greenhouse gas emissions, the paper quantifies current yield gaps across major rice environments (irrigated, rainfed lowland, rainfed upland) and models scenarios to 2050 to inform policies and R&D priorities for sustainable intensification and improved self-sufficiency.

Prior work documents Africa’s growing rice demand and import dependence, yield stagnation, and large yield gaps relative to Asia (e.g., Seck et al. 2012; van Ittersum et al. 2016; Arouna et al. 2021). Studies highlight biophysical and management constraints—insufficient nutrient supply, pest/weed/disease pressure, and water management—especially in rainfed systems (Saito et al. 2013; Tanaka et al. 2017). Modeling frameworks such as the Global Yield Gap Atlas and ORYZA have been applied to estimate yield potential and gaps in various regions (Bouman et al. 2001; van Wart et al. 2013; Li et al. 2017). Scenario assessments suggest sustainable intensification can reduce import needs and land conversion but must navigate environmental and socioeconomic constraints (Cassman & Grassini 2020; Yuan et al. 2021, 2022; Hasegawa et al. 2021).

Scope: Fifteen African rice-producing countries covering ≈65% of harvested rice area and 80% of production (2018–2020): Egypt; Burkina Faso; Côte d’Ivoire; Ghana; Mali; Niger; Nigeria; Senegal; Ethiopia; Kenya; Madagascar; Rwanda; Tanzania; Uganda; Zambia. Three environments were analyzed separately: irrigated, rainfed lowland, and rainfed upland. Site selection and upscaling: Following the Global Yield Gap Atlas (GYGA) protocol, representative weather stations (RWS) and climate zones (CZs) were selected using SPAM 2010 crop distribution maps and expert elicitation. Circular buffers (100 km radius) around RWS, clipped to CZ boundaries, were chosen to reach ≥50% coverage per country–water regime while minimizing overlap. Final coverage: 45 RWS (irrigated), 45 (rainfed lowland), 26 (rainfed upland); buffers cover 38% of area but represent CZs containing ≈54% (rainfed) and 71% (irrigated) area. Weather data: Daily solar radiation, Tmax/Tmin, precipitation, vapor pressure deficit, and wind speed for 20 years (2000–2019) in 10 countries; 11 years (1995–2005) in Egypt, Côte d’Ivoire, Senegal, Madagascar, Rwanda. Data underwent QA/QC and gap-filling; sources included measured and propagated gridded datasets. Crop management and calendars: Local agronomists provided dominant cropping systems (ecosystem, water regime), establishment method, sowing/transplanting dates, varieties, maturity, and cropping intensity per buffer; calendars compiled (Supplementary Fig. 1). Yield potential simulation: ORYZA v3 model with DRATE v2 for phenology calibration was used to simulate yield potential (irrigated) and water-limited yield potential (rainfed). Generic genetic parameters for African rice varieties were adopted from prior calibrations (ORYZA2000v2n14). Assumptions included: base development temperature 14 °C, lower growth threshold 12 °C, critical sterility temperature 35.6 °C, no photoperiod sensitivity unless specified. Rainfed lowland simulations used non-puddled clay loam with bund height 25 cm and two groundwater-depth scenarios (40 cm and 100 cm) split 50:50 to represent favorable vs drought-prone conditions. Rainfed upland assumed sandy loam without bund and deep groundwater (1000 cm). Soil hydraulic parameters followed prior studies; sensitivity analyses tested soil water holding, hardpan, bund height, groundwater depth. Validation: Cross-validation against well-managed experimental yields where available and comparison with GYGA yield potentials for analogous climate zones in other regions. Yield gap estimation: For each country–water regime, actual farmers’ yields (2018–2020) were compiled from national statistics, literature, and expert input. Attainable yield was set at 80% of simulated yield potential for irrigated and 70% for rainfed rice. The exploitable yield gap equals attainable yield minus actual yield. Area-weighted aggregation considered cropping intensity and area shares of cycles. Demand, SSR, and deficit: Baseline demand (2018–2020) used production, imports, exports, stock changes. Future (2050) demand combined UN medium-variant population projections with country-specific per-capita rice demand trajectories from IFPRI’s IMPACT model; milled-to-paddy conversion applied using country milling rates (0.63–0.69). SSR = production/demand; deficit = projected demand minus extrapolated production under historical yield trends (capped at attainable yield) on current area. Scenario assessment (to 2050): Explored yield intensification from current to full closure of exploitable yield gap alongside three area expansion rates: 0.2, 0.4, 0.6 M ha yr⁻¹ (±50% around historical ≈0.4 M ha yr⁻¹). Assumptions: no change in irrigated area share or cropping intensity; regional trade balances within Africa; rice price US$289 Mg⁻¹ paddy (World Bank Thai 5% equivalent); imports/exports valued accordingly. Coverage extended continent-wide by scaling demand to include data-poor countries and assuming average yield gap patterns for non-modeled producers. Sensitivity analysis: Tested −5%/−10% reductions in yield potential and land suitability (as climate-change proxies) under current area expansion, recomputing production and SSR outcomes.

• Yield potential and actual yields: Area-weighted average yield potential across sites, environments, and water regimes is ≈8 Mg ha⁻¹ (≈4 Mg ha⁻¹ in rainfed upland West Africa to ≈11 Mg ha⁻¹ in irrigated Nile Delta, Egypt). Irrigated potential averages 9.9 Mg ha⁻¹; rainfed 7.0 Mg ha⁻¹. Actual area-weighted farmer yield is 2.9 Mg ha⁻¹ (36% of potential).
• Exploitable yield gaps: Attainable yield defined as 80% (irrigated) and 70% (rainfed) of potential. Area-weighted exploitable gap is 52% of attainable yield overall; by regime: irrigated 47%, rainfed lowland 53%, rainfed upland 58%. Absolute exploitable gaps: irrigated 3.7 Mg ha⁻¹, rainfed lowland 5.7 Mg ha⁻¹, upland 2.4 Mg ha⁻¹. Egypt and Senegal (irrigated) are near closure; East Africa shows the largest irrigated gaps; large gaps are widespread in rainfed areas.
• Current self-sufficiency: None of the 15 analyzed countries are presently self-sufficient; SSR ranges ≈0.85 (Mali, Tanzania) to <0.20 (Ethiopia, Kenya, Niger). Area-weighted average SSR across the 15 is 0.67; continent-wide SSR ≈0.57 (due to low-producing consumers in North Africa).
• 2050 demand and deficit under historical yield trends: Population growth (1.3 → 2.5 billion) and higher per-capita consumption (48 → 60 kg cap⁻¹) increase demand by ≈135% to ≈150 Mt (paddy) by 2050. With continuation of historical yield trends on current area, Africa-wide SSR falls to ≈0.26 (country range ≈0.03–0.47). The resulting rice deficit is ≈67 Mt by 2050—equivalent to ≈US$20 billion in imports at current prices or ≈23 M ha of new rice area at current yields. Country deficits exceed 6 Mt in Côte d’Ivoire, Mali, Madagascar, and reach ≈16 Mt in Nigeria.
• Scenario outcomes (area expansion constrained to 0.2–0.6 M ha yr⁻¹): • Current area expansion (0.4 M ha yr⁻¹) + full closure of current exploitable yield gap eliminates import needs by 2050, but requires sustained yield gains of ≈104 kg ha⁻¹ yr⁻¹—challenging given past stagnation. • Halving the exploitable yield gap (≈52 kg ha⁻¹ yr⁻¹ gains) + current area expansion (0.4 M ha yr⁻¹) raises SSR from 0.57 to ≈0.82 by 2050; imports remain near today’s levels due to higher absolute demand. • Halving the gap + faster expansion (0.6 M ha yr⁻¹) can achieve self-sufficiency by 2050 but implies >doubling rice area (≈15 → 33 M ha), with high investment and environmental costs. • Halving the gap + slower expansion (0.2 M ha yr⁻¹) increases import dependence roughly two-fold, costing ≈US$16 billion yr⁻¹ by 2050.
• Comparative context: Africa’s yield potential per crop is somewhat lower and less stable than Southeast Asia’s (≈8 vs ≈9 Mg ha⁻¹), but the relative gap to attainable yield is much larger in Africa (≈50% vs ≈36% in SE Asia and ≈14% in China), indicating substantial scope for management-driven gains.

The findings show that Africa’s growing rice demand cannot be met sustainably via imports and land expansion alone. Large exploitable yield gaps—especially in irrigated and rainfed lowland systems—offer the primary lever to reduce import dependence and curb cropland conversion. Scenario analysis indicates that a realistic intensification pathway—halving current exploitable yield gaps through improved management (fertility, water, pest/weed/disease control, varietal choice, establishment) combined with continuation of historical area expansion—substantially increases the self-sufficiency ratio while avoiding drastic import growth. Achieving full self-sufficiency without extreme land expansion requires even larger yield gains, which may be difficult to sustain. The results emphasize targeted investments in agricultural R&D, extension, and inputs to close gaps, prioritizing regions with high absolute and relative gaps, and favoring productivity-enhancing strategies such as expanding irrigated area where environmentally and socioeconomically feasible and increasing cropping intensity. The analysis underscores the need to maintain rice export surpluses in other regions to buffer African demand, while highlighting the risks of status quo: escalating imports, exposure to price shocks, and environmentally costly land conversion.

By mapping yield potential and exploitable yield gaps across major rice environments in 15 African countries and projecting 2050 outcomes under alternative intensification and area-expansion scenarios, the study demonstrates that substantial yield intensification is essential to improve rice self-sufficiency and to avoid large increases in imports and land conversion. A pragmatic pathway is to halve current exploitable yield gaps while maintaining current rates of area expansion, which elevates SSR markedly and stabilizes import needs despite surging demand. Policy and R&D should prioritize: closing nutrient-related yield gaps; strengthening extension and access to inputs and finance; rehabilitating and, where sustainable, expanding irrigation; improving agronomic practices; and cautiously raising cropping intensity. Future work should refine subnational targeting of interventions, integrate dynamic climate risks and water constraints, and evaluate environmental trade-offs and socio-economic feasibility of irrigated expansion and intensification at scale.

• Input data constraints: Limited availability and coverage of high-quality weather, soil, management, and experimental yield data required assumptions and upscaling; nonetheless, cross-validation supports robustness of yield potential estimates.
• Climate change not explicitly simulated in baseline scenarios: Sensitivity analysis with −5%/−10% reductions in yield potential and land suitability suggests minor impacts on 2050 outcomes relative to current gap sizes, but localized effects could be larger.
• Genetic gains not included: Potential improvements from breeding or hybrid adoption were excluded; expected gains are smaller than current gaps and face adoption barriers.
• Socioeconomic feasibility: Assumed rates of gap closure (notably 50% by 2050) may be optimistic given constraints in finance, markets, and governance, though precedents exist (e.g., Egypt, Senegal) and Asia’s Green Revolution rates are used as benchmarks.
• Area expansion assumptions: Fixed irrigated share and cropping intensity; real-world changes could alter outcomes. Environmental impacts of expansion (ecosystem conversion, methane, water extraction) are acknowledged but not quantified.
• Continental scaling: Demand scaled for data-poor countries; production for non-modeled producers assumed to follow average gaps, introducing uncertainty.