Projecting trends of arabica coffee yield under climate change: A process-based modelling study at continental scale

International audience Context: Climate change may lead to negative impacts on coffee production, such as reduced yields. Addressing this issue requires identifying climate risks and assessing the adaptation potential of agronomic practices across spatial and environmental gradients. OBJECTIVE This study aimed to evaluate climate change impacts on arabica coffee yields at continental scale and evaluate a specific adaptation measure, i.e. increasing shade tree density in agroforestry settings, by simulating the physiological links between coffee growth, climatic factors and agronomic management. Methods: After evaluating the performance of the process-based model DynACof in simulating arabica yields (using data from previous studies), we developed a new tool called G-DynACof, a modelling framework for spatializing DynACof on a regional scale using extensive climate projections and soil geodata. We used G-DynACof to simulate trends of potential coffee yields in Latin America and Africa using an ensemble of downscaled and bias-corrected climate projections for the period 2036–2065 compared to a historical period 1985–2014. RESULTS AND Conclusions: Despite considerable uncertainties due to the scarcity of information on agronomic management at the regional scale, our results indicate that potential yields could decrease between 23 % and 35 % in Latin America and between 16 % and 21 % in Africa, depending on the Shared Socioeconomic Pathway (SSP) considered (SSP1–2.6 and SSP5–8.5, respectively). Yield variations were very heterogeneous in space, with yields increasing at high altitudes and low latitudes, indicating a possible future shift of production areas. In our simulations, the effect of increased shade tree density on productivity was also spatially variable, and its potential for adaptation to climate change remains uncertain, requiring further investigation. SIGNIFICANCE Impact analyses and adaptation modelling of coffee agrosystems, together with socio-economic indicators, can delineate realistic, comprehensive, integrated risk assessments and support effective adaptation recommendations.