March 3, 2026Open Access

Identifying limiting factors for maize yield in China's major maize-producing regions using random forest

Key Points

Maize yield is constrained by climate, with nighttime warming particularly affecting spring maize yields.
Random forest models explained over 80% of yield variation, highlighting planting density as a key factor.
Factors like solar radiation and minimum temperature significantly influence yield at different growth stages.
Adaptation strategies are essential, as climate warming could reduce maize yields by up to 11.8% without intervention.

Abstract

• Spring maize yield is mainly constrained by nighttime warming during the grain-filling stage, whereas summer maize is most sensitive to high minimum temperatures during the establishment stage. • Climate explains over 50% of maize yield variability, while increasing planting density consistently enhances yields and represents the most effective management lever. • Stage-resolved, interpretable machine learning identifies clear climate- and management-sensitive targets for reducing maize yield losses under climate warming. Maize ( Zea mays L.) is an important food crop worldwide. Understanding yield-limiting factors is essential for optimizing maize productivity under varying agroclimatic conditions. In this study, the relative contributions of climate, soil, and management factors to yield variation in spring and summer maize across 34 sites in China during 2017-2020 were assessed. Random forest (RF) models explained more than 80% of the yield variation, and SHapley Additive exPlanations (SHAP) and Accumulated Local Effects (ALE) were employed to interpret the effects of key variables. Climate emerged as the dominant driver, accounting for nearly 50% of the total feature importance. For spring maize, solar radiation during the establishment stage (ES) had a strong positive effect, whereas the minimum temperature during the grain-filling stage (GFS) had a negative effect. In contrast, summer maize yield was constrained by elevated nighttime temperatures during ES but benefited from increased growing degree days (GDD) during GFS. Among all the variables, planting density (PD) was consistently important across both systems, and increasing PD represented a direct and effective pathway to enhance yield. The results of the yield component analysis further revealed that the significantly higher kernel number per ear (on average 68 kernels more than summer maize) was the main contributor to the superior performance of spring maize. Climate scenario simulations indicated that, without adaptive management, future warming could reduce spring and summer maize yields by 6.1–11.8% and 5.5–9.1%, respectively. These findings underscore the stage-specific climate sensitivity of maize and support the development of targeted adaptation strategies to sustain yields under future climate change.

Identifying limiting factors for maize yield in China's major maize-producing regions using random forest

Key Points

Abstract

Cite This Study