Cereal yields in Ethiopia relate to soil properties and N and P fertilizers

There is an urgent need to increase cereal yields in the Ethiopian Highlands to ensure national food security. A major crop response-to-fertilizer program was set up in 2017–2019 as part of the CASCAPE project in the Ethiopian Highlands. It covered 33 experiments on maize, teff and wheat in five reference soil groups (Nitisols, Luvisols, Vertisols, Leptosols and Andosols). Five levels of multi-nutrient fertilizer (50–300 kg NPSZnB ha^{− 1} and 100 kg urea ha^{− 1}) were applied. At the lower fertilizer level, average yields were 5500, 1500 and 3300 kg ha^{− 1} for maize, teff and wheat, respectively. At the highest rate, yields were 7900, 2100 and 5000 kg ha^{− 1}. Maize and wheat yields were strongly correlated to the reference soil groups, but not to rainfall differences. Wheat yields were also positively correlated to soil organic carbon levels, underpinning the need to apply integrated soil fertility management. Comparing NPSZnB fertilizers with NPS and DAP fertilizers revealed a lack of statistically significant advantage for the fertilizers including Zn and B. As B was present in fine-granular form in the fertilizer bags, being prone to segregation, firm conclusions on the need for this micronutrient cannot be drawn. The use of ‘recommendation windows’ is suggested to group results into concrete packages at district levels and below, preferably combined with soil maps since soil types were correlated with maize and wheat yields. The windows could then be the starting point to develop ‘last mile’ fertilizer use policies, relevant to farmers and the way they manage their fields in the landscape.