Biochar's impact on carbon sequestration and yield-scaled global warming potential in China: A machine learning-enhanced meta-analysis

Biochar has shown strong potential to mitigate greenhouse gas (GHG) emissions, yet most meta-analyses have examined isolated factors, overlooking the interactive effects of climate, soil, and management on agricultural productivity and environmental sustainability. This study integrated meta-analysis with machine learning (ML) to quantify biochar's effects on GHG emission, soil organic carbon (SOC), and yield-scaled global warming potential (GWP). A total of 1207 paired observations from 64 field studies conducted in China (2000–2024) were analysed. Biochar application significantly increased crop yield by 5 % and SOC by 34 %, while reducing N 2 O emissions by 25 %, resulting in a 25 % decrease in yield-scaled GWP. Weakly alkaline biochar was most effective in reducing CO 2 emissions in alkaline soils and CH 4 emissions in acidic to neutral soils ( p < 0.05). Low C/N ratio biochar significantly decreased CO 2 and CH 4 emissions across soils with varying C/N ratios. Machine learning models identified biochar pH, application rate, soil pH, and soil C/N ratio as the primary determinants of these effects. CO 2 emission responses were positively correlated with biochar application rate and negatively correlated with soil pH and C/N ratio ( p < 0.05), whereas CH 4 emissions were positively regulated by soil pH, soil C/N ratio, and biochar pH. Pathway analysis indicated that biochar pH indirectly reduced CO 2 emissions through enhanced SOC, while soil pH and C/N ratio directly influenced CH 4 dynamics ( p < 0.05). Overall, this study demonstrates biochar's potential to enhance crop productivity while mitigating GHG emissions, providing a scientific basis for its application in carbon-neutral agroecosystems development.