Assessment of variability and uncertainty of soil organic carbon in a mountainous boreal forest (Canadian Rocky Mountains, Alberta)

•Mountain soil organic carbon stocks characterized by large spatial variability•Spatial variability causes large uncertainty of mountain SOC stocks.•The coarse fraction introduces the largest uncertainty in mountain SOC stocks.•Knowledge on spatial distribution of geomorphic processes improves SOC stocks.

Mountain environments are heterogeneous and dynamic geomorphic environments sensitive to land use and climate change. Heterogenic environmental conditions result in a large variability of mountain soil properties, and thus in large uncertainties of inventories of soil organic carbon (SOC). In this study we analyzed the variability of soil properties associated with the calculation of a SOC inventory in a mountain environment in the Canadian Rocky Mountains (Alberta). Therefore, we calculated the analytical uncertainty and spatial variability of SOC stocks using Gaussian error propagation and Taylor series expansion along seventeen 36 m long transects to identify major sources of uncertainty. SOC stocks in the upper 10 cm and 30 cm are 2.4 ± 0.7 kg C m− 2 and 6.4 ± 5.6 kg C m− 2, respectively. The bulk densities generated the largest uncertainty associated with the analytical precision (10.0%). However, analytical uncertainties (ranging between 2.3 and 24.2%) are much smaller than the uncertainty introduced by the spatial variability, for instance of the coarse fraction (63.8%) and SOC concentration (40.1%). This study contributes to insufficiently considered analysis of uncertainties in SOC stocks and demonstrate the high potential of nested sampling approaches to identify sources of uncertainties of SOC stocks. To reduce the uncertainties associated with heterogeneous mountain environments, we propose to apply more sophisticated statistics (e.g. regression analysis considering frequency distributions of measured coarse fractions in different geomorphic environments) rather than simple mean per unit approaches, as frequently applied in regionalization studies of soil properties.