The robust concept of mineral-associated organic matter saturation: A letter to Begill et al., 2023.

Sequestration of persistent soil organic carbon (SOC) is seen as a cost-effective and scalable CO2 removal strategy, so it is critical to understand whether soils have a finite capacity to store mineral-associated organic C (MAOC)—on average more persistent than particulate organic C (POC). Since the conceptualization of “soil C saturation” around the turn of the century, many have embraced it and studied its upper limits, found to vary up to 78–86 g C kg−1 fine fraction (Feng et al., 2014; Georgiou et al., 2022), its controls, and effects on SOC dynamics (e.g., Cotrufo et al., 2019; Six et al., 2002; Sokol et al., 2022). The conceptual definitions of fractions do not fully correspond to their procedural definitions (Leuthold et al., 2022). Begill et al. dispersed aggregates using sonication, which is known to disintegrate POM and disperse OM in solution even at energies below 100 kJ (Leuthold et al., 2022). They also used flocculant to precipitate the OM in solution (<50 μm) as mineral-associated OM (MAOM). To discount POM contamination, they noted the consistent C-to-nitrogen (N) ratio (10.7) of their fine fractions (Figure 1a), assuming POM has a higher C:N than that. However, the latter has been contradicted by observations of C:N ratios of 10–12 for fine POM (Six et al., 1998). Furthermore, since they selected soils with C:N < 13, several may have POM C:N ≤ 10.7 and, for soils with C:N ratios of 10.7–13, a minor shift from the linear relation would indicate POM contamination. That is what their data show; all fine fractions with high C contents have C:N > 10.7 (Figure 1a), suggesting POM contamination. Begill et al. (2023) questioned our observation of MAOC saturation in European grasslands and forests (Cotrufo et al., 2019) because there were no soils with >47 g MAOC kg−1soil. The samples were selected to represent the distribution of total SOC and soil texture observed across over ∼20,000 European sampling locations (Lugato et al., 2021). Thus, that all the soils had <47 g MAOC kg−1 soil was a finding rather than the result of a biased selection. While we recognize that our data were sparser for higher SOC soils, there is adequate evidence for saturation behavior (Figure 1b) for all land uses: polynomial fits of the MAOC-to-SOC relationship were statistically better and different from linear fits at p < .001 (Figure 1b). Rather than challenging the C saturation concept, the growing evidence for a patchy “skyscraper” model for what we define as MAOC and the range in 14C ages for MAOC (Six et al., 2002), should serve as a reminder that separating a MAOC fraction by size and/or density is not necessarily sufficient to identify the persistent SOC pool. It is well known that MAOM can exchange with dissolved OM (Kleber et al., 2021), and some process-based models already represent two MAOM pools: a dynamic MAOM and a more stable MAOM. Yet, data are needed to calibrate and verify these models. Identifying the most effective and scalable method to separate and characterize the dynamic exchangeable (possibly OM bound to OM) from the stable MAOM (possibly OM bound to minerals) is critical to better inform the broad and active field of soil C sequestration. MFC discloses to be a cofounder of Cquester Analytics LLC. The authors have no conflict of interests. The data that support the findings of this study are available from Zenodo at https://doi.org/10.5281/zenodo.8227702.