Contrasting rainfall and heat risks to southern African maize during El Niño

El Niño poses substantial risks to maize production in southern Africa, but the relative contributions of rainfall deficits and heat, and the opportunities to anticipate these hazards, vary geographically. We combine historical climate–yield analysis across ten southern African countries with verification of forecasts from two models to assess these risks and their predictability. National and sub-national maize yield records and gridded climate data are used to characterise climate–yield relationships and statistically decompose El Niño-associated yield differences into rainfall and heat contributions. These analyses reveal that the estimated rainfall contribution is larger in Zimbabwe and Mozambique, whereas the estimated heat contribution is larger in South Africa, Namibia and Zambia. For both temperature and rainfall, forecast skill is generally greatest for the first month after initialisation. At longer leads, rainfall skill approaches climatological levels in most regions, while temperature and compound dry–hot forecasts retain skill in some areas. In the northern part of the domain, rainfall forecast skill for March at one-month lead improves during El Niño, suggesting enhanced opportunities for early warning. Overall, the findings of this study demonstrate the importance of matching agricultural early warning to regional differences in climate–yield sensitivity and forecast skill. In particular, temperature forecasts offer opportunities to anticipate agricultural hazards where heat contributes substantially to yield reductions, even when rainfall forecast skill is limited.