Projected Climate Impacts on Crop Yields in Africa

This dataset compiles projections of crop yield changes in Africa under future climate scenarios, drawing from 72 peer-reviewed crop modeling studies. It represents a major update to earlier efforts by Challinor et al. (2014), Aggarwal et al. (2019), and Hasegawa et al. (2021), which originally covered 31 African-focused studies on rice, maize, and wheat. The current dataset expands coverage through additional systematic searches in Web of Science and Scopus and includes yield projections for grains, tubers, and cash crops with and without adaptation interventions. A forthcoming update will add projections for legumes based on 35 additional references. The dataset captures key variables including crop type, country, climate scenario, projected and baseline yield, CO₂ fertilization, and adaptation measures. Methodology:The dataset was developed through a systematic review following PRISMA guidelines. Peer-reviewed studies published before 2024 were identified via Scopus and Web of Science using structured search strings based on previous reviews (Aggarwal et al., 2019; Challinor et al., 2014; Carr et al., 2022). Included studies provided estimates of climate change impacts on crop yields in Africa using crop simulation models. Data were extracted on yield under baseline and future conditions, with and without adaptation. Climate impact was calculated as: YI (%) = ((Yf - Yb) / Yb) × 100, where Yf is projected yield and Yb is baseline yield. Adaptation effectiveness was assessed as: ΔY_adapt (%) = ((Yf_adapt − Yf_no_adapt) / Yf_no_adapt) × 100, where Yf_adapt is projected yield with adaptation, and Yf_no_adapt without. Additional variables captured include region, time period, CO₂ assumptions, and climate model used. The final dataset is harmonized to support meta-analysis and visualization of climate risks to African agriculture. Each geolocated study site within this bounding box was assigned gridded climate delta values using the ISIMIP3b NextGDDP dataset, which provides bias-adjusted daily climate projections derived from CMIP6 models. We extracted: ΔT (°C): Average change in daily mean temperature between historical (2005) and future midpoints (e.g., 2030s, 2050s, 2080s) ΔP (%): Percentage change in cumulative precipitation over the same periods These delta values were matched to the study sites based on their latitude and longitude coordinates, allowing a consistent and spatially explicit characterization of the local climate change signal associated with each yield projection.