Formation of mineral-associated organic matter via rock weathering: an experimental test for the organo-metallic glue hypothesis
Abstract. Mineral-associated organic matter (MAOM), representing the dominant form of relatively stable C in soil, contains high physicochemical heterogeneity. The co-localization of organic matter (OM) with reactive aluminum (Al) and iron (Fe) phases in various MAOM fractions—across a range of natural and cultivated soils from five soil orders—has led to the “organo-metallic glue” hypothesis. The hypothesis proposes that coprecipitates formed between mineral-derived metals and microbially processed OM act as a binding agent, promoting the formation of stable microaggregates and thereby enhancing soil OM persistence. However, the formation mechanism remains unclear as the observed associations reflect multiple soil processes. We thus designed a simple laboratory experiment to test if the supply of metals and metalloids through rock weathering controls MAOM formation and if the OM-to-metal ratio of the material formed is consistent with complexation, sorptive association, or their mixture (i.e., coprecipitates). Two end-member igneous rocks (granite and basalt) crushed to have 38–75 µm size and, additionally, 20–38 µm size for basalt, as well as river sand (100–300 µm) as control were mixed with leaf compost (powdered to 100–250 µm) as single OM source. The mineral-OM mixtures were incubated aerobically at 30 oC with the natural soil microbial community and subjected to 8 wet-and-dry cycles using artificial rainwater (pH 4.73) over a 55-day experiment. The mixtures were then fractionated by density to examine the formation of meso-density, organo-mineral aggregates (1.8–2.4 g cm–3: MF) by distinguishing it from the compost-dominant low-density fraction (< 1.8 g cm–3: LF) and high-density fraction (>2.4 g cm–3: HF) consisting of the crushed rock. The MF formation assessed as C content was 1.49 ± 0.06 mg C g–1 rock (fine basalt), 1.04 ± 0.08 (coarse basalt), and 0.62 ± 0.06 (granite) over the 55 days, while the net MF mass increase was detected only in fine basalt due to the presence of meso-density materials in the crushed rock (< 7 % by mass). Faster chemical weathering of the fine basalt was indicated by a significant increase in extractable Fe and Al phases, largely in MF, and the highest leaching of Fe and base cations (esp. Na and Ca). The organo-mineral aggregates formed in the fine basalt treatment had the C-to-metal (Fe+Al) ratio of 0.36 ± 0.01 (molar basis), consistent with organo-metal coprecipitation. Further analysis focusing on the two basalt treatments revealed a significant decline in C:N ratios by 23–25 units and enrichment of δ13C and δ15N by 0.9–1.2 ‰ and 0.6 ‰, respectively, in MFs compared to LFs, indicating a strong contribution of microbial N-containing compounds to the MAOM formation. While microbial community composition differed among the treatments, no significant difference was found in qPCR-based bacterial number or species richness. Microscopic analyses using SEM and STXM confirmed the presence of shaking-resistant microaggregates and co-localization of C, Fe, and Al in MF from selected MF samples. Together, our results strongly supported the organo-metallic glue hypothesis and provided laboratory evidence of basalt-induced MAOM formation as well as some insights into early pedogenesis and organo-mineral interactions when applying crushed rock to soils.
General comments
This manuscript reports results of a well-designed incubation experiment aimed at quantifying the formation of organo-mineral aggregates from leaf compost and different rock powders. The manuscript is generally well-written and the amount of analyses performed is impressive, ranging from classical soil fractionation and extractions, through microbial community analyses, to nanoscale spectroscopy.
The introduction is effective, and the methods are well-described and largely appropriate (see comments below). The results section is very rich, but I did not find it overly clear. In particular, the figures’ legend need to be more explicit (see comments below). The discussion is generally well-written and interesting. I thought that the largest shortcoming was the lack of an explicit conceptual framework for the organo-mineral associations. Sorptive association, complexation, cation bridging, co-precipitation are mentioned, but the relation of these mechanisms among one another and with the « organo-metallic glue » hypothesis is not made clear. Perhaps a diagram would help? Co-precipitation usually has a different definition than the one implied here - a mixture of adsorption and complexation, if I understood correctly (see for instance https://doi.org/10.1038/s41467-025-61273-4). Furthermore, it is not clear how the results can discriminate between these mechanisms. In fact, I am not sure that they can.
I would recommend that the authors streamline the results section to bring the most important ones to the forefront, and re-focus the discussion on substantiated trends rather than speculative ones. There are a lot of exciting findings here – rapid mineral weathering, formation of reactive secondary phases, association with microbially processed organics, etc.!
Specific comments
L 102+: Your wet sieving procedure probably induced some mineralogical fractionation compared to the initial rock. For instance, for granite, I assume that the micas were preferentially lost in the discarded < 38 um particle size class and quartz was preferentially retained. I don’t think that it is necessarily a problem, but it should probably be discussed somewhere.
L 111: A C content of 0.3 mg C / g rock is not a lot, but still very significant compared to your reported C content in the middle density fraction (0.6 – 1.5 mg C / g rock). For basalt, this is likely to consist of secondary carbonates. Did you consider the possible effect on your 13C results?
For granite, I am not sure where this C could be coming from.
L 187: I don’t think that it is appropriate to analyse Ca in these extracts, as Ca-pyrophosphate, Ca-oxalate and Ca-citrate salts are only very sparingly soluble.
L 213: « The subsets of MF from selected treatments (granite and coarse basalt) ... »: I might have missed it, but why was the fine basalt omitted?
This sentence is also missing a conjugated verb.
L 244: « The effect of mineral type on the measured variables was tested by one-way ANOVA » followed by a Tukey test, if I understand the legend of your figures correctly – to clarify.
This is a minor point, but I would argue that it is not coherent to use a family-wise correction (Tukey) for the effect of rock, whereas you used simple t-tests (4 times, once for each rock) for the effect of incubation. The inflation in type I error is close to the same in both cases. I would recommend t-tests everywhere. If you are concerned about type I error, you can simply decrease your alpha level (e.g., to 0.01).
L 246-249: For me this approach is ok, but the explanation is not entirely convincing, since the portion used for density fractionation was sub-sampled again prior to C and N analysis, right? Or did you not like the results you got from the bulk samples?
Your recovery rates do look very nice.
L 346-349: See previous comment – your extractions are inappropriate for Ca; I don’t think that the results are interpretable. The salts of Ca with oxalate, citrate and pyrophosphate have a very low solubility. In addition, Ca pyrophosphate solubility decreases with increasing free Ca ions. I would recommend removing this part.
L 396, Table 5: It is noteworthy that the amount of base cations leached from the fine basalt was about the same as from the coarse basalt, except for Na, which was 3 x greater. This suggests preferential weathering of Na-phases?
L 501: In the absence of soil respiration data, I find hypotheses about relative heterotrophic activity highly speculative.
L 520+: This section seems quite nebulous to me. Co-precipitation first « cannot be excluded » (even if sorptive associations alone could account for the observations). I don’t understand why the associations are « best characterized as … coprecipitates » in the next sentence. Based on which evidence?
L 525+: I agree that the increase in Fe phases suggest a predominance of sorptive associations. I don’t quite understand how this relates to the « organo-metallic glue hypothesis » (4.1). Overall, I think that the discussion is missing an explicit conceptual framework.
L 564+: This discussion of the relevance of the study for enhanced rock weathering is potentially interesting but it is not sufficiently grounded in the authors’ results, in my opinion. The nearly one-unit pH increase in the fine basalt treatment, together with Na leaching, points to significant mineral dissolution; this contradicts the idea that the higher abundance of aggregates « likely slows down the rate of basalt weathering ».
Similarly, I don’t think that this study supports the « increase in soil OM upon the mixing of basic rock powders ». It does not contradict it either, but you did not see changes in bulk C. It is possible that you had more heterotrophic microbial activity, thus enhancing both C mineralisation and formation of organo-mineral associations between microbial compounds and reactive phases.
Technical corrections
L 48-49: SRO aluminosilicates are not oxides, strictly speaking. Please rephrase.
L 87: … linkages BETWEEN
L 170: centrifuged
L 215: Other subsets (plural) or Another subset (singular)
L 243: Replace ‘mineral type’ with ‘rock type’ (or equivalent). You did not look at individual minerals. Also, L 260, 289
L 285: Clarify the Fig. 2 legend. « The same letters … are not significantly different » does not mean much. What did you compare?
L 305: Same comment for Fig. 3. Yes, the same letters are not different, but what is the comparison? The differences represented by letters and stars are not clear.
Also L 308, for Fig. 4. It looks like the letters are for different things in part a and b (part a, for rock type, part b, for fractions?)
L 366: « C was MORE enriched for Fe than Al… »
L 566: Monovalent base cations (Na and K) are released too.