4 March 2023 in Journal of Cleaner Production
19 September 2026 in SSRN Electronic Journal
OGOtávio Grassi Gonçalves
+8
Brazil’s biodiesel strategy relies heavily on soybeans, but climate risks to their productivity are rarely incorporated into energy planning. We estimate spatially explicit changes in soybean productivity across 1,131 Brazilian municipalities and assess implications for biodiesel supply. Using municipal agricultural records and gridded weather data for 2000–2022, we train extreme gradient boosting (XGBoost) models and project climate-conditioned productivity using a five-model climate ensemble under intermediate (SSP2-4.5) and high (SSP5-8.5) emissions pathways through 2100. The historical model identified a nonlinear temperature response, with predicted productivity declining above approximately 25.22°C during municipality-specific sowing windows. Between 2023–2030 and 2090–2100, mean productivity is projected to decline by 6.9% under SSP2–4.5 and 10.8% under SSP5–8.5. Losses exceeding 10% affect 40% of municipalities under SSP2-4.5 and 73% under SSP5-8.5 considering a long-term period, with persistent hotspots along the Amazon–Cerrado agricultural frontier and in southern producing regions. Given soybean oil share of 73.3% of national biodiesel feedstocks, these losses imply reductions of 5.1% and 7.9% in total feedstock availability, equivalent to 0.50 and 0.77 billion litres of biodiesel annually at 2025 consumption levels. Under current technology, restoring this output would require an additional 2.4 and 3.8 million hectares of soybeans. These estimates exclude future carbon dioxide (CO2) fertilisation, technological progress and additional adaptation measures. Climate vulnerability could therefore constrain higher biodiesel blending mandates, while increasing pressure on land, food, and feed markets. Brazil’s biodiesel strategy should combine climate-informed feedstock planning with targeted agricultural adaptation and diversification toward additional and climate-resilient feedstocks.