Linking soil structure and microbial communities to predict CO2 emissions from drained arable peatlands

Understanding the interactions between soil structure, microbial communities, and greenhouse gas dynamics is critical for predicting carbon losses from drained peatlands under agricultural use. This study investigates CO₂ emissions across winter wheat, sugar beet, and bare soil treatments on a productive UK peat farm, integrating high-resolution X-ray Computed Tomography (XCT), microbial community profiling, and in situ gas and soil measurements. Soil structure differed between treatments, with bare soil exhibiting the highest pore connectivity and gas diffusivity. These structural conditions aligned with higher in situ CO₂ concentrations, despite reduced root inputs and microbial diversity. In contrast, cropped soils supported more diverse microbial communities, especially fungi, but exhibited lower gas diffusivity and CO₂ concentrations—likely reflecting restricted oxygen availability and plant–microbe competition. Relative gas diffusivity (Dp/D₀) was strongly regulated by soil moisture across all treatments, with a consistent inverse relationship (R² > 0.93). A machine learning model (XGBoost) accurately predicted CO₂ concentrations (R² = 0.83) using microbial and physical soil properties, identifying microbial taxa potentially linked to carbon cycling.

These findings demonstrate that subtle differences in pore architecture can shape microbial function and carbon loss, even in the absence of statistically significant structural differences. This highlights the need to integrate microbial ecology and soil physics in greenhouse gas modelling for sustainable management of agricultural peatlands.