Tropentag, September 16 - 18, 2026, Göttingen "Towards multi-functional agro-ecosystems promoting climate-resilient futures"

Impact of environment and crop management on the variance of maize (Zea mays L.) yield in different agroecological zones in Kenya

¹University of Helsinki, Dept. of Geosciences and Geography, Finland
²University of Helsinki, Dept. of Agricultural Sciences, Finland
³University of Helsinki, Dept. of Forest Sciences, Finland

Abstract

Addressing the limited understanding of how environmental factors and crop management affect maize (Zea mays L.) grain yields across agroecological zones (AEZs) is essential in improving decision making and climate change adaptation planning in sub-Saharan Africa (SSA).
This study evaluated how environmental (seasonal climate factors, soil type) and crop management practices (irrigation, planting dates) affected rainfed and irrigated maize yields in semi-arid, sub-humid and humid agroecological zones (AEZs) in Kenya and identified critical thresholds that demarcated the transition between yield gains and losses.
We conducted field experiments in Maktau, Kishenyi and Kabete over four growing seasons (2024–2025), covering both long (LR) and short (SR) rains, applying full irrigation in LR2024 and SR2024 and growth-stage based deficit irrigation in LR2025 and SR2025. AquaCrop was calibrated and validated with the observed data and used to simulate four other early and delayed planting (local planting date ±15 and ±30 days) and corresponding rainfed alongside full and deficit irrigation strategies. We employed XGBoost models to estimate yield responses to seasonal environmental and crop management factors and utilized SHAP (Shapley Additive Explanations) values to quantify the relative importance and influence of each predictor across the AEZs.
Across the three AEZs, SHAP-derived yield responses to seasonal irrigation, seasonal rainfall and seasonal mean maximum temperature followed patterns consistent with their respective moisture regimes. In semi-arid Maktau, yield gains were associated with seasonal irrigation above 585 mm and yields declined at seasonal rainfall below ~110mm, while temperature effects were weak and inconsistent across the 29–31°C range. In sub-humid Kishenyi, the yield gains corresponded to seasonal irrigation above 547 mm and seasonal rainfall above ~270–315 mm range, with temperature contributions turning negative above ~24°C. In humid Kabete, yield gains strengthened sharply beyond seasonal irrigation of ~495 mm, seasonal rainfall above ~ 600 mm, and temperature followed a clear inverted-U pattern, with yield gains between ~23–25°C and losses outside this range. Early planting (local planting date -15 and -30 days) consistently produced positive yield contributions across all AEZs, while delayed planting (local planting date +30 days) produced increasingly negative yield contributions.
This study highlights the complex AEZ-specific interplay between environmental, crop management and maize yield. The thresholds identified and interaction effects enables transition from broad interventions to targeted AEZs aligned strategies, that improve yield and enhance climate resilience in maize cropping systems.

Keywords: Agroecological zones, AquaCrop, irrigation, planting dates, smallholder, sub-Saharan Africa, XGBoost model

Contact Address: Morice Odhiambo, University of Helsinki, Dept. of Geosciences and Geography, Gustaf hällströmin katu 2 A (Physicum), 00560 Helsinki, Finland, e-mail: morice.odhiambohelsinki.fi