Modeling global anthropogenic erosion in the Holocene

A large proportion of natural vegetation has been converted to agricultural use, and this typically accelerates erosion by one to two orders of magnitude. Quantification of this accelerated erosion is important to understand the impact of human activities on soil ecosystem service given that soil erosion induces soil degradation and changes in soil organic carbon (SOC) stocks. Until now, few studies have evaluated the accumulated impact of agricultural erosion, since the start of agriculture (ca. 6000 BC), on the soils system and the carbon cycle. In this study, we mainly focused on the enhanced water erosion by conversion of natural vegetation to crops, while wind erosion on the cropland is not assessed. We first evaluated and constrained existing anthropogenic land cover change (ALCC) scenarios by comparing observed cumulative erosion for the agricultural period under a wide range of global agro-ecological conditions with model simulations. An optimized land-use scenario that makes the best fit between the simulation and the observation was derived in the model calibration. We further applied a spatially distributed erosion model, which was modified based on Revised Universal Soil Loss Equation (RUSLE), under the optimized land-use scenario across globe to estimate the total anthropogenic cumulative erosion and characterize their spatial variability. Simulations suggest that conversion from natural vegetation to cropland has caused a global cumulative agricultural erosion of 27,187 ± 9030 Pg for the period of agriculture. This results in an average cumulative sediment mobilization of 1829 ± 613 kg m−2 on croplands, corresponding to a soil truncation of ca. 1.34 ± 0.45 m. Regions of early civilization, particularly with high cropland fractions such as South Asia, Southeast Asia, and Central America have higher area-averaged anthropogenic erosion than other regions. This results in spatial variability in soil truncation rates because of erosion, which would further affect the soil production rate. Our study shows that observations of long-term anthropogenic erosion at the catchment scale can be used to constrain the reconstructed land-use scenarios.