From prairie to crop: Spatiotemporal dynamics of surface soil organic carbon stocks over 167 years in Illinois, U.S.A.

Abstract

Quantifying spatiotemporal dynamics of soil organic carbon (SOC) stocks is needed to understand the impact of land use change and can help target carbon sequestration efforts. In the recently and radically transformed landscapes of the state of Illinois, U.S.A., we evaluated surface SOC stocks under land use change using a space-for-time substitution method over 167 years. Additionally, we determined SOC stocks for the A horizon vs 0-30 cm depth to evaluate pedogenically-informed vs more commonly used fixed depth approaches. Legacy soil datasets from 1980 to 2012 were combined with environmental covariates using a random forest algorithm. To more accurately estimate pre-agricultural land use SOC stocks (i.e., pre-1845), SOC observations collected from soils under native prairie and forest were extracted from peer-reviewed publications. The model was validated on 25 % of the total 627 test data (RA-hor2: 0.59 and R0-302: 0.56; RMSEA-hor: 20.5 and RMSE0-30:19.3 Mg/ha) independent of the 75 % of data for calibration (R2: 0.91; RMSEA-hor:10.1 and RMSE0-30:9.6 Mg/ha). SOC stocks were largest under prairie (A horizon: 156.1 Mg/ha; 0-30 cm: 152.4 Mg/ha) and lowest under pasture (A horizon: 33.2, 0-30 cm: 44.6 Mg/ha). SOC stocks varied less by soil order than by land use. Between 1845 and 2012, surface SOC stocks decreased for most of Illinois, with greatest losses in central (-16.3 Mg/ha) and east-central Illinois (-47.0 Mg/ha) where approximately 80 % of prairie was converted to cropland. A slight increase in surface SOC stocks occurred in the unglaciated northwest region and the less recently glaciated south region, as well as in alluvial corridors. This study (i) highlights how estimating spatiotemporal dynamics of surface SOC stocks over centennial timescales can benefit from including measures of SOC under native land use not usually contained in legacy pedon datasets, and (ii) illustrates the potential of identifying localized hotspots of historical SOC loss and thus deficits that can be prioritized for carbon sequestration efforts.