How accurately can soil organic carbon stocks and stock changes be quantified by soil inventories?

Precise determination of changes in organic carbon (OC) stocks is prerequisite to understand the role of soils in the global cycling of carbon and to verify changes in stocks due to management. A large dataset was collected to form base to repeated soil inventories at 12 CarboEurope sites under different climate and land-use, and with different soil types. Concentration of OC, bulk density (BD), and fine earth fraction were determined to 60 cm depth at 100 sampling points per site. We investigated (1) time needed to detect changes in soil OC, assuming future re-sampling of 100 cores; (2) the contribution of different sources of uncertainties to OC stocks; (3) the effect of OC stock calculation on mass rather than volume base for change detection; and (4) the potential use of pedotransfer functions (PTF) for estimating BD in repeated inventories. <br><br> The period of time needed for soil OC stocks to change strongly enough to be detectable depends on the spatial variability of soil properties, the depth increment considered, and the rate of change. Cropland sites, having small spatial variability, had lower minimum detectable differences (MDD) with 100 sampling points (105 ± 28 gC m<sup>−2</sup> for the upper 10 cm of the soil) than grassland and forest sites (206 ± 64 and 246 ± 64 gC m<sup>−2</sup> for 0–10 cm, respectively). Expected general trends in soil OC indicate that changes could be detectable after 2–15 yr with 100 samples if changes occurred in the upper 10 cm of stone-poor soils. Error propagation analyses showed that in undisturbed soils with low stone contents, OC concentrations contributed most to OC stock variability while BD and fine earth fraction were more important in upper soil layers of croplands and in stone rich soils. Though the calculation of OC stocks based on equivalent soil masses slightly decreases the chance to detect changes with time at most sites except for the croplands, it is still recommended to account for changing bulk densities with time. Application of PTF for the estimation of bulk densities caused considerable underestimation of total variances of OC stocks if the error associated with the PTF was not accounted for, which rarely is done in soil inventories. Direct measurement of all relevant parameters approximately every 10 yr is recommended for repeated soil OC inventories