Large uncertainty in soil carbon modelling related to method of calculation of plant carbon input in agricultural systems

The application of dynamic models to report changes in soil organic carbon (SOC) stocks, for example as part of greenhouse gas inventories, is becoming increasingly important. Most of these models rely on input data from harvest residues or decaying plant parts and also organic fertilizer, together referred to as soil carbon inputs (C). The soil C inputs from plants are derived from measured agricultural yields using allometric equations. Here we compared the results of five previously published equations. Our goal was to test whether the choice of method is critical for modelling soil C and if so, which of these equations is most suitable for Swiss conditions. For this purpose we used the five equations to calculate soil C inputs based on yield data from a Swiss long-term cropping experiment. Estimated annual soil C inputs from various crops were averaged from 28 years and four fertilizer treatments; the average was 3.6 Mg C ha(-1) and covered a surprisingly large range from 2.1 to 5.3 Mg C ha(-1) year(-1) among the five equations. For single crop species, differences reached 6.6 Mg C ha(-1) year(-1). The variation between estimated soil C inputs depended on crop type and increased with yield. Simulations with the model C-TOOL showed that calculated SOC stocks were affected strongly by the choice of the allometric equation. With four equations, a decrease in SOC stocks was simulated, whereas with one equation there was no change. This considerable uncertainty in modelled soil C is attributable solely to the allometric equation used to estimate the soil C input. We identify the evaluation and selection of allometric equations and associated coefficients as critical steps when setting up a model-based soil C inventory for agricultural systems. Highlights Uncertainty analyses in soil C modelling have focused mainly on model parameters and structures. The uncertainty related to estimation of soil C inputs from crop residues was calculated. Simulated soil C stocks and stock changes depend strongly on the method of estimating C input. The method to derive soil C inputs is critical for model-based soil C inventories.