Effects of soil management strategies on maize and soybean yields in northeast China: A meta-analysis

Soil management strategies are crucial for optimizing crop yields and sustaining soil health. However, regional heterogeneity influences their effectiveness. To address this issue, developing region-specific soil management strategies is essential to ensuring future food security. As a solution, a regional machine learning aided meta-analysis was performed to assess the influence of soil management practices, climatic conditions, and soil characteristics on crop yield, and to predict the region-specific applicability of various soil management strategies across Northeast China (NEC). A meta-analysis of 1104 observations from 105 field trials showed that soil management practices increased yield by 36.9 %. Combined chemical and organic fertilization led to the highest yield increase, at 42.2 %, followed by organic fertilization at 41.1 % and chemical fertilization at 28.7 %. No tillage increased yield by only 2.2 %, while rotation tillage and subsoiling raised yields by 17.1 % and 18.8 %, respectively. Rotation tillage emerged as the most effective practice, while subsoiling provided even higher yield benefits in certain localized areas. Additionally, both crop monocropping and crop rotation significantly improved yield. Random forest analysis showed that agricultural management was the main driver of yield variation, contributing 38.47 %, followed by soil properties at 33.87 % and climate factors at 27.26 %. Structural equation modeling indicated that yield gains were mainly driven by increased soil organic carbon. Temperature, precipitation, and clay content also affected regional responses. This study underscores the importance of adapting soil management practices to the environmental conditions of NEC to achieve sustainable yield improvements. The proposed region-specific strategies may also inform agricultural planning in other areas facing similar constraints, thereby contributing to broader agricultural sustainability.