Bulk density and coarse fragment content of the French soil monitoring network for better assessment of changes of soil organic carbon stocks

• Changes of bulk density (BD) and coarse fragment (CF) content were evaluated in the French soil quality monitoring network. • CF content did not change between campaigns while topsoil BD decreased significantly by −3.1 ± 0.9 %. • The change of methods between campaigns in a substantial number of sites biased the estimates of CF and BD. • The impact of the changes of BD and CF on the evolution of soil organic carbon stocks (SOCS) was simulated. • Estimation of the evolution of SOCS is difficult if campaign-specific values of CF are considered, due to rocky soils. Bulk density (BD) and coarse fragments (CF) are often missing in soil monitoring networks. In the French Soil Quality Monitoring Network BD and CF have been measured in two campaigns in sites distributed in a grid over mainland France. The objective of this work is to evaluate i) how BD and CF have changed, ii) how volumetric methods and land use affect the observed trends, and iii) to simulate the impact of BD and CF changes on the estimations of changes of soil organic carbon stocks (SOCS). The results showed no significant change of (Δ) CF between campaigns in either the topsoil or subsoil, while there was a significant decrease of BD only in the topsoil (−3.1 ± 0.9 %). When methods were constant in both campaigns, BD decreased in croplands and grasslands topsoil and it did not change in forests. We could explain a rather low part of the ΔBD variance (R 2 = 0.21), mostly linked to the within-site variability of CF and to changes in some methodological aspects between campaigns. We simulated changes in topsoil SOCS, assuming SOC content constant over time; we found negative estimates that were the largest in soils with low CF content (−0.2 ± 0.06 % yr −1 an -1 ) as well as differences between using site-specific average CF values over both campaigns and using site-specific values of each campaign [-0.17 ± 0.04 % yr −1 vs −0.05 ± 0.06 % yr −1 , respectively]. These differences were caused by rocky soils, and were not significant when using only sites with low CF content. Our work highlights some limitations in assessing BD changes using broad-scale soil monitoring networks. These findings have important consequences on the methods used to assess changes in SOCS and their uncertainties at broad-scale. They are timely and relevant given the current proposals to implement soil health monitoring at national, continental and global scale.