The potential to increase maize yields and mitigate climate change by adopting integrated soil fertility management across different regions in Kenya — A simulation study

To avoid soil fertility decline and increased greenhouse gas (GHG) emissions, it has been recommended to improve maize yields in sub-Saharan Africa with sustainable practices such as Integrated Soil Fertility Management (ISFM), instead of relying solely on mineral fertilizer. However, the yield responses and environmental trade-offs of ISFM likely depend on soil and climatic conditions. To explore this, we used the DayCent model to simulate 30-year average yields of maize monoculture across Kenya under 17 different ISFM scenarios, co-created with Kenyan smallholder farmers. We compared yields, changes in SOC stocks and N 2 O emissions against current baseline conditions (monocropping with minimal nutrient inputs). The scenario that best represented a ’feasible-input’ level consisted of 2 t C ha −1 yr −1 of farmyard manure and 60 kg N ha −1 season −1 of mineral fertilizer. Other scenarios included different amounts (0, 1, and 2 t C ha −1 yr −1 ) and types of organic inputs in combination with four rates of mineral N fertilizer (0, 30, 60, and 90 kg N ha −1 season −1 ). The uncertainty of model predictions was quantified through Monte Carlo simulations. The model results indicate a significant potential for yield improvements in the humid regions of western Kenya (from 3.7 to 8.1 t ha −1 yr −1 ) with the ‘feasible-input’ compared to the baseline scenario; GHG emissions per kg of yield were generally lower (the median value reduced from 0.9 to 0.5 kg CO 2 -eq kg −1 yield). However, in the semi-arid regions of eastern Kenya, maximum yields under any scenario were 1.1 t ha −1 yr −1 , reached at inputs of 1 t C farmyard manure ha −1 yr −1 or 60 kg mineral N ha −1 season −1 . The uncertainty analysis showed a high confidence in the ‘feasible-input’ scenario’s ability to increase yields and reduce SOC losses compared to the baseline, but a high uncertainty regarding its impact on GHG emissions. Specifically, the 95% credibility intervals for the combined CO 2 and N 2 O emissions ranged from reductions of up to 1000 kg CO 2 -eq ha −1 yr −1 to increases of up to 200 kg CO 2 -eq ha −1 yr −1 . These results strongly support the use of ISFM practices to enhance maize yields and mitigate soil fertility losses, particularly in areas with sufficient rainfall. However, due to the high uncertainty surrounding simulated N 2 O emissions, we cannot establish with certainty whether ISFM reduces GHG emissions on a per hectare basis or increases them. • We simulated the maize yields of combined manure and mineral N application in Kenya. • The simulated scenarios and baseline conditions were co-created with stakeholders. • Yield increases up to 200% were possible, but only in sites with sufficient rainfall. • Yield-scaled greenhouse gas emissions were highest in scenarios with low or no inputs. • Monte-carlo analysis showed a high uncertainty in projected N 2 O emissions.