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

Understanding interactions between soil structure, microbial communities, and greenhouse gas dynamics is critical for predicting carbon losses from drained agricultural peatlands. This study tested the hypothesis that land use alters soil structure and microbial communities, thereby shaping CO 2 flux, using high-resolution XCT, microbial profiling, and gas and soil measurements across winter wheat, sugar beet, and bare soil treatments on a productive UK farm on peatland. Bare soil exhibited the highest pore connectivity and gas diffusivity (Dp/D 0 : 0.08–0.10 in dry conditions), declining to near-zero during wet periods in October. Fungal alpha-diversity (Shannon index: 2.8–3.2) was significantly higher in cropped soils compared to bare soil (2.0–2.5), with sugar beet supporting the most diverse fungal communities. Sordariomycetes dominated fungal assemblages (50–75 % relative abundance), while Actinobacteria and Vicinamibacteria consistently comprised 20–30 % of bacterial communities. Soil moisture strongly regulated diffusivity (R 2 = 0.93, p < 0.001), driving seasonal shifts in gas transport and microbial dynamics. Fungal communities showed stronger treatment differentiation (R 2 = 0.24–0.49) than bacterial communities, with distinct assemblages observed in sugar beet at 20 cm depth (R 2 = 0.489, p = 0.011). An XGBoost machine learning model explained 82 % of the variance in CO 2 concentrations, identifying key fungal (OTU_15_F, OTU_6_F) and bacterial (OTU_901, OTU_5115) taxa as top predictors. These results highlight that crop selection can alter microbial diversity by up to 60 % and drive tenfold changes in soil gas diffusivity, underscoring the importance of integrating soil structural and microbial metrics into greenhouse gas models. Such insights can guide sustainable peatland management strategies that balance productivity with carbon conservation. • Integrates microbial and soil data to predict CO 2 emissions from arable peatlands. • Key taxa, soil temperature, and land use are top predictors via XGBoost (R 2 = 0.82). • Vegetation cover strongly influences microbial-driven CO 2 fluxes in peatlands. • Microbial biomarkers can indicate soil carbon changes and GHG emissions early. • Findings guide sustainable management and climate mitigation in organic soils.