Soil organic carbon in the colombian paramos: an opportunity for climate mitigation

Abstract

Soils store more carbon than the atmosphere and vegetation combined, yet regional maps that resolve how climate, topography, substrate, and land use shape soil organic carbon (SOC) remain scarce. We assembled > 1,800 field observations of SOC (0–40 cm) across Colombian páramos and modeled SOC density with a random forest that allowed non-linear responses to elevation, temperature, precipitation, soil moisture, terrain position, land cover and a longitudinal proxy for parent material. Model performance, assessed with spatially blocked 5-fold cross-validation, explained 65% of the independent variance. We mapped pixel-wise uncertainty using a cluster bootstrap, producing confidence intervals and national stock estimates with uncertainty. SOC peaked at intermediate elevations under cool, wet conditions and locally concave to planar geomorphologies. Relative to natural páramo (165 Mg C ha⁻¹), disturbed areas stored substantially less SOC (63 Mg C ha⁻¹). Mean SOC density across Colombian páramos was 168 Mg C ha⁻¹, yielding an approximate national stock of 4.2 × 10² Tg C. The typical pixel-level 95% confidence interval (half-width) was 7 Mg C ha⁻¹. Because SOC losses from drainage, tillage and overgrazing are avoidable and reversible, our maps and uncertainty estimates provide a spatial basis to target soil-focused mitigation, conservation, and restoration in high-elevation landscapes of the Colombian paramos that current biomass-based mechanisms often overlook.