Land use drives the distribution of free, physically protected, and chemically protected soil organic carbon storage at a global scale

Soil organic carbon (SOC) sequestration is increasingly emphasized as a climate mitigation solution, as scientists, policy makers, and land-managers prioritize enhancing belowground C storage. To identify key underlying drivers of total SOC distributions, we compiled a global dataset of soil C stocks held in three chemical forms, reflecting different mechanisms of organic C protection: free particulate organic C (fPOC), physically protected particulate organic C (oPOC), and mineral-protected soil organic C (mSOC). In our dataset, these three SOC pools were differentially sensitive to effects of climate, soil mineralogy, and ecosystem type, emphasizing the importance of distinguishing between physical and chemical C protection mechanisms. C stocks in all three pools varied among ecosystems: cropland soils stored the least amount in each pool, with forest and grassland soils both containing significantly more fPOC (40-60% greater in each ecosystem) than croplands. oPOC stocks did not significantly differ from zero in croplands, but were substantial in forest and grassland soils. Meanwhile, mSOC stocks were the greatest in grasslands and shrublands (90-100% greater than croplands). In cropland soils, there were no major effects of tillage on C storage in any of the three pools, while manure addition enhanced mSOC stocks, especially when added with inorganic N. Thus, the human land use intensity in croplands appears to reduce SOC storage in all major pools, depending upon management; retaining native vegetation should be emphasized to maintain current global SOC stocks.

Methods
We collected data from papers which utilized a density fractionation methodology to measure soil organic C (SOC) within the three fraction pools: free particulate organic fraction (fPOC), occluded particulate organic fraction (oPOC), and mineral soil organic carbon (mSOC). To filter through published studies reporting SOC stocks, we chose to extract data from papers which cited one of the following rigorous and influential fractionation methodologies: (Golchin et al., 1994; Shaymukhametov et al., 1984; Six et al., 1998; Sollins et al., 1984, 2006, 2009; Spycher et al., 1983; Steffens et al., 2009). On Google Scholar, there were 3915 papers citing one of the above papers through September 2020. We examined each publication and only extracted data from empirical studies, published in a peer-review journal in English, that measured SOC via density fractionation. When recording oPOC or mSOC data, we only collected data from papers which utilized sonication, sodium hexametaphosphate dispersion, or physical disruption (e.g., from glass beads) to separate oPOC from mSOC (Figure S1). Unless it was specified for a particular soil and reported in the respective study, we set a density cutoff of 1.85 g cm-3 for classifying mSOC. Since some studies isolated C pools along a density gradient, we summed SOC stored in fractions above or below the density cutoff.

To make meaningful global comparisons, we chose to only include papers which reported enough information to calculate the C stock in units of g C m2. In cases where C concentrations but not stocks were reported, we used soil depth and bulk density (BD, g cm-3) to calculate the C stock in a given pool (fPOC, oPOC, mSOC). Data were recorded separately for different locations, ecosystem types, treatments, soil depths, and fractions. Because there was a wide range in soil depth increments used within studies, we summed C stocks from specific depth intervals (e.g. 0-10 cm, 10-30 cm) into two categories: ‘topsoil’ (0-30cm) or subsoil (31+cm). Soil volume associated with each observation was calculated assuming a 1x1 m sampling area and the sampling depth increment reported by the study authors (e.g. 0-10 cm, 15-30 cm).