Peatland vulnerability to agricultural expansion: Conversion risk, inactivity, and carbon loss

Peatlands store nearly 30% of global soil carbon, yet face conversion pressure from agricultural expansion. Focusing on Alberta, Canada, we develop a spatial framework that integrates multi-source datasets with random forest modeling, permutation importance, prior-probability calibration, and multi-year inactivity detection to evaluate conversion risk, drivers, land-use stability, and carbon loss for cropland and pasture. From 1998–2024, 848 km² of peatlands were converted to cropland and 1001 km² to pasture, concentrated within the provincial “White Area” and along the White–Green boundary. Since the 2000s, agricultural expansion has shifted northward and outward from long-settled districts, placing peatlands near the White-Area margin in the Green Area at heightened future risk. Locational forces dominate: cropland conversion correlates more strongly with economic conditions and land endowment, whereas pasture expansion is more constrained by climate. Although pixel-level peak risk is higher for cropland, pasture occupies a larger medium-to-high risk footprint. Post-conversion agricultural use on peat is unstable: nearly half of peat-derived parcels later fall out of production—43.31% of peat cropland and 56.04% of peat pasture were inactive, defined as not classified as agriculture for ≥ 3 consecutive years during 2011–2024. Estimated cumulative peat-soil carbon losses were approximately 1.47 Tg C for cropland and 1.57 Tg C for pasture, with emission-factor-based 95% uncertainty ranges of 1.21–1.75 Tg C and 0.80–2.36 Tg C, respectively. Policies should make soil-carbon costs explicit in approvals that lower peat water tables, enforce peat type integrity in offsets, and target high-risk corridors; public funds are better directed to clustered retire-and-rewet of already inactive peat parcels.