There is an increasing demand for information on organic carbon (OC) in subsurface horizons, because subsurface horizons down to the bedrock can contribute to more than half of soil carbon stocks. In this study, we developed pedotransfer functions (PTFs) for predicting OC content in subsurface horizons of European soils. We used a dataset with a wide geographical coverage in Europe. The dataset was stratified sequentially into land‐cover and soil categories. For each category, PTFs were developed by multiple linear regression with the main soil and climatic factors of soil OC storage as predictor variables: OC in topsoil (0–20 cm), depth of subsurface horizons, texture and bulk density (BD) in subsurface horizons, and mean annual temperature and precipitation. Three land‐cover categories were separated: woodland, a combined category of grassland and non‐permanent arable land, and permanent arable land. For the combined land‐cover category, two soil categories were identified: (i) soils with clay‐rich subsoil and soils with little horizon development and (ii) organic‐rich soils and soils rich in Fe and Al compounds. The adjusted R2 of all PTFs was above 0.62. When PTFs were applied to independent data, the adjusted R2 was above 0.51 for all of them. The PTFs showed good prediction ability, with root mean square error (RMSE) values between 2.43 and 13.82 g C kg−1 soil. The adjusted R2 and RMSE of PTFs were better when BD was used as a predictor variable. The PTFs could be implemented easily for applications at the continental scale in Europe.
https://onlinelibrary.wiley.com/doi/full/10.1111/ejss.12464
