Quantifying the potential of agricultural soils to store carbon. A data-driven approach illustrated for the Netherlands.

Improved soil and cropland management changes the soil carbon stocks and thereby mitigate climate change. However, spatially explicit insights on management impacts as well as critical thresholds for optimum SOC levels are lacking, which is crucial for actionable changes in farming practices. In this study we unravelled the contribution of soil texture, geohydrology and soil quality to changes in SOC in the Netherlands using a data-driven approach (using XGBoost) using 21.123 soil analyses done by agricultural laboratories. The current C stock of the 0-30cm soil layer is 119 ton C ha-1 and could be increased by 21 to 59 ton C ha-1 depending on soil type, land use and the agronomic measures taken. The SOC saturation capacity, expressed as the ratio between the actual and potential SOC stock varied from 85 to 93% in grassland soils, from 55 to 83% in arable soils and from 69 to 91% in other land uses. On average, the actual C saturation degree was 75%. The key factors controlling the potential of soils to sequester additional carbon within environmental limits for N and P included the crop sequence in the last decade, soil texture (i.e. oxide extractable aluminium, iron and phosphorus), the acidity, and groundwater depth. The data driven approach shows that spatially explicit recommendations for carbon farming are possible up to the farm and field scale, facilitating the implementation of carbon farming and the mitigation of climate change. When all agricultural fields are saturated with C, an equivalent of 257 Mton of CO2 can be stored.