Mineral states and sequestration processes involving soil biogenic components in various soils and desert sands of Inner Mongolia Authors: Xuemei Yang, Xin Gao, Khan M G Mostofa, Wang Zheng, Nicola Senesi, Giorgio S Senesi, Davide Vione, Jie Yuan, Si-Liang Li, Longlong Li, Cong-Qiang Liu Published: 2024 ABSTRACT AbstractSoil biogenic components are subject to continuous sequestration, and export from soils into the surrounding air and water environments. However, the processes involving the stability or lability of their mineral states remain still unclear. To assess these issues, we have measured various biogenic components in a number of agricultural, forest, grassland, and deep soils, as well as desert sands from Inner Mongolia, both in the solid state and liquid extracts. The contents of soil organic carbon (SOC) and soil total nitrogen (STN) were higher in soils than in sands, whilst those of soil total sulfur (STS) and inorganic carbon were higher in sands and deeper soils. The significant positive correlations found between STS and SOC, and STN, and their significant negative correlations with pH and δ13C-SOC in all soils suggest a pH-dependent sequestration of C, N, and S. The decreased stability of organo-mineral complexes at acidic pH, resulting from the acidification of humic substance (HS) functionalities, leads to a higher availability of nutrients that facilitates the sequestration of soil organic matter (SOM). Conversely, an increase in pH enhances the stability of organo-mineral complexes by promoting negatively charged HS functionalities, which reduces the availability of nutrients and the sequestration of SOM. The δ13C-SOC enrichment in desert sands (-17.63 to -7.10‰) and its depleted values in soils (-24.9 to -18.8‰) suggest the occurrence of C sequestration in desert, via uptake of enriched atmospheric CO2 (-8.4‰). The fluorescence spectra of humic substance components and their molecular weights in sands were typically different from those of soils. The predominant relatively low molecular weight (MW) (< 15–25 kDa) of alkali-extracted (complexed state: CS) components and the relatively high MW (> 25 − 15 kDa) of water-extracted (labile state: LS) components of all soils suggest, respectively, their involvement in organo-mineral complexes and for export into the surrounding environment. The quantities of LS and CS soil components differ significantly on dependence of soil characteristics, implying their corresponding lability or stability in soils. These findings will provide useful input for the management of the corresponding soil/sand ecosystems.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-024-80004-1.Keywords: Forest, agriculture, grassland and deep soils; Desert sands; Biogenic components (C, S, N and SiO32−); Humic substances; Nutrients; High performance size exclusion chromatographySubject terms: Carbon cycle, Element cycles, Biogeochemistry, Climate sciences, Environmental sciences Introduction Soil organic matter (SOM), and especially its humic substance (HS) fractions, play multiple fundamental functions in soils and sands, which include: formation of organo-mineral complexes1,2; soil structure, cation exchange capacity and buffer/water-holding capacity3,4; oxygenic/anoxygenic photosynthesis5,6 via uptake or sequestration of carbon7,8, nitrogen9,10, and sulfur11–13; microbial mineralization of HS or SOM2,14 with emission of CO215,16, N2, or N2O17,18, and COS and H2S19, as well as export of water-soluble components (e.g. DOM, DIC, nutrients and various elements) via rainwater discharge and/or groundwater infiltration20–23. These biogeochemical processes constantly fuel the microbial food chains in downstream river-lake-marine environments24. Continuous export and/or emissions of various components from soils into the surrounding environments are counterbalanced by their sequestration in soils/sands, particularly in the case of C, N, and S. These sequestration processes in soils and desert sands remain unclear. The mineralization of organic components in soil and sand primarily occurs at the mineral-microbes/organic interfaces2,25 via microbial heterotrophic respiration14. In this framework, labile state (LS) components, or lability, refer to the fresh, soluble organic molecules derived from decaying plant materials, as well as readily available byproducts produced from various soil minerals under specific environmental conditions26–28. These components are not bound in organo-mineral complexes and are thus susceptible to export into the surface water and groundwater of the surrounding environment. However, some key degraded byproducts can form organo-mineral complexes that remain in the complexed state (CS), or stability, in soil and sand28–31. However, the generation of LS and CS forms of soil or sand components under environmental conditions and their correlations among themselves, and with other components, remain unclear. Furthermore, HS fractions, i.e., humic acids (HA), fulvic acids (FA), and protein-like substances (PLS) primarily contribute to these processes by either exporting or stabilizing components as organo-minerals22,27,28,32. However, it is still unclear which molecular weight (MW) fractions of HA, FA, and PLS are involved in either the lability and export into surface water and groundwater, or in their stability. Based on the above considerations, we isolated a water extractable fraction (We, containing the LS fraction) and an alkali extractable fraction (Ae, CS fraction) from soils with different land uses (forest, agriculture, grassland) and at different depths (superficial and deep layers), and from desert sands. The key objectives of this study were to: (i) assess the occurrence of SOC and its stable isotopes (δ13C-SOC), and their relation with soil and sand total sulfur (STS), soil/sand inorganic carbon (SIC), and soil and sand total nitrogen (STN), in order to infer their roles in the sequestration of C, S, or N, and mineral neformation; (ii) measure the occurrence of both LS and CS components, including dissolved organic carbon (DOC), HS fractions, dissolved inorganic carbon (DIC), SO42−, dissolved N-nutrients, dissolved organic nitrogen (DON), PO43−, dissolved organic phosphorus (DOP) and dissolved SiO32−, to assess their simultaneous role in lability and export, or stability in organo-mineral complexes; (iii) investigate the lability and export, or stability role of HS fractions by high performance size-exclusion chromatography (HPSEC), and (iv) comprehensively assess the key factors that affect the LS and CS fractions in soils and sands. Results Biogeochemical occurrence, mineral states and sequestration of soil/sand components Organic and inorganic C plus δ13C-SOC The SOC content featured a large variation (from 0.21 to 16.7 g/kg) among soil types, with the highest average value in agricultural soils (12.8 ± 3.0 g/kg). It was, respectively, 7.5%, 137%, 1836%, and 3278% higher than that in forest, grassland, deep soils, and desert sands (Table 1, S1). The SOC contribution to STC showed a similar order: agricultural soils (70.0 ± 15.3%) > forest soils (64.0 ± 21.1%) > grassland soils (27.7 ± 9.3%) > desert sands (14.3 ± 3.0%) > deep soils (5.1 ± 3.8%). These findings suggest that agricultural and forest soils acted as primary storage/sequestration sites of SOC originating from plant materials, whereas the lowest content of SOC in deep soils would be due to its degradation. Table 1Average chemical and physical properties of the forest, agricultural, grassland and deep soils, and desert sands examined.Soil typepHECWater contentSTCSOCSICSTSSTNC/NSOC/Sδ13C-SOCδ15NSoil textureaµS cm-1(%)g kg-1g kg-1g kg-1g kg-1g kg-1‰‰%%%Forest (n = 4)8.17 ± 0.2968.2 ± 28.12.16 ± 1.3518.9 ± 2.111.9 ± 3.57.0 ± 4.30.204 ± 0.0071.51 ± 0.5917.3 ± 10.0155 ± 42‒23.8 ± 0.96.21 ± 0.8651.67 ± 33.9644.71 ± 30.663.6 ± 3.3Agriculture (n = 4)8.28 ± 0.14250 ± 3272.62 ± 1.2418.5 ± 2.512.8 ± 3.05.6 ± 2.90.221 ± 0.0121.34 ± 0.2616.6 ± 4.0155 ± 34‒23.5 ± 1.45.00 ± 1.6966.81 ± 30.431.69 ± 29.031.48 ± 1.38Grassland (n = 4)8.79 ± 0.3062.1 ± 12.51.32 ± 1.1319.1 ± 15.55.4 ± 4.213.7 ± 11.50.192 ± 0.0170.77 ± 0.7033.3 ± 10.072 ± 52‒20.1 ± 0.94.29 ± 0.1984.47 ± 1.1415.01 ± 0.980.53 ± 0.16Desert (n = 5)8.60 ± 0.0830.8 ± 9.10.22 ± 0.162.7 ± 1.20.38 ± 0.192.3 ± 1.00.177 ± 0.0060.10 ± 0.0235.8 ± 23.65.6 ± 2.6‒11.4 ± 5.1nd99.32 ± 1.410.654 ± 1.3840.10 (n = 1)Deep (n = 3)8.36 ± 0.2265.2 ± 17.21.70 ± 0.9614.2 ± 8.20.66 ± 0.4013.6 ± 8.10.198 ± 0.0060.79 ± 0.3020.8 ± 7.09.0 ± 5.6‒23.7 ± 1.03.50 ± 1.0275.4 ± 5.923.49 ± 5.561.14 ± 0.34Mean8.4798.641.5014.17.26.880.200.8822.679.2-19.25.2576.9721.701.33Range7.90–9.2123.8–7400.10–4.41.6–41.00.21–16.71.4–30.40.18–0.240.07–1.749.6–65.53.0-200-24.9 to -7.13.15–7.1318.52–99.990-74.670.0-6.77 Average chemical and physical properties of the forest, agricultural, grassland and deep soils, and desert sands examined. Differently, SIC showed the highest average content in grassland soils (13.7 ± 11.5 g/kg), which was very similar to that in deep soils (only ~ 1.2% higher), and approximately 94.0%, 141%, and 487% higher than, respectively, forest soils, agricultural soils, and desert sands. The SIC contribution to STC followed a similar order, on average: deep soils (95.0 ± 3.8%) > desert sands (85.7 ± 3.0%) > grassland soils (72.3 ± 9.3%) > forest soils (36.0 ± 21.1%) > agricultural soils (30.2 ± 15.3%). The δ13C-SOC varied from − 24.85‰ to -7.10‰, with very similar and low average values detected in forest, agricultural, and deep soils (respectively, -23.77 ± 0.88‰, -23.53 ± 1.41‰, and − 23.66 ± 1.04‰). Slightly higher values were found in grassland soils (-20.05 ± 0.9‰) and, especially, in desert sands (-11.37 ± 5.05‰). Similar δ13C-SOC values have been reported in top and deep forest and agricultural soils33–35. The average contents of DOCLS and DOCCS in agricultural soils (173 ± 64 and 571 ± 161 mg/kg, respectively) were higher than in forest soils (by 12.0% and 10.3%, respectively), grassland soils (by 19.0% and 166%, respectively), deep soils (by 108% and 78.0%, respectively), and desert sands (by 1215% and 580%, respectively; Fig. 1; Table 2). Noteworthy, grassland soils showed a relatively low content of DOCCS in comparison with agricultural and forest soils (respectively, 166% and 141% lower). Conversely, grassland soils featured a relatively high content of DOCLS, exceeding agricultural and forest soils by 19.0% and 6.0%, respectively, which was likely due to the presence of grassland vegetation. Grasslands might thus be minor contributors to organo-mineral stability. In particular, the average DOCLS and DOCCS contributions to SOC followed the order: deep soils (respectively, 15.0 ± 8.5% and 56.4 ± 25.6%) > desert sands (respectively, 3.3 ± 0.92% and 17.8 ± 12.6%) > grassland soils (respectively, 3.3 ± 2.1% and 4.7 ± 1.4%) > forest soils (respectively, 1.3 ± 0.67% and 5.3 ± 4.8%,) > agricultural soils (respectively, 1.4 ± 0.7% and 4.6 ± 1.6%). Various land uses thus show a diversity in soil DOM lability that, in turn, indicates different organo-mineral stability in the corresponding soil types. In particular, deep soils typically showed the highest levels of lability and organo-mineral stability, presumably due to intense SOC mineralization upon long-term burial. Differently, forest soils > agricultural ≈ grassland (in the order) exhibited a relatively higher CS organo-mineral stability, possibly due to HS or DOM primarily derived either from plant litters or from photosynthetic SOC in forest soil compared to those of agriculture or grassland29,31,36. Concentrations of labile state (LS) and complexed state (CS) dissolved organic carbon (DOCLS and DOCCS), dissolved inorganic carbon (DICLS and DICCS), dissolved organic nitrogen (DONLS and DONCS), nitrate (NO3‒LS and NO3‒CS), ammonium (NH4 + LS and NH4 + CS), and nitrite (NO2‒LS and NO2‒CS) in forest, agricultural, grassland and deep soils and desert sands. Different lower case and capital letters (A, B or a, b) indicate significant differences at p < 0.05 level. Table 2Average contents of labile state (LS: water extracts, or lability)*, and complexed state (CS: alkali extracts, or stability)** DOC, DIC, SO42−, and various nutrients in the forest, agricultural, grassland, and deep soils, and desert sands examined.Sample typeDOCDICSO42−TNDIPNO2−-NNH4+-NNO3−-NDONTPDOPSiO32−-Simg.kgmg.kgmg.kgmg.kgmg.kgmg.kgmg.kgmg.kgmg.kgmg.kgmg.kgmg.kgLabile state (LS: water extracts), or lability of soil componentsForest soil154.6 ± 96.699.6 ± 14.527 ± 29.730.8 ± 6.53.6 ± 3.91.1 ± 0.31.5 ± 0.917.5 ± 3.510.7 ± 2.25.9 ± 4.42.3 ± 0.644.3 ± 3.9Agricultural soil173.2 ± 64.3110.6 ± 18.8248.1 ± 445.533.7 ± 10.19.5 ± 91.2 ± 0.51.8 ± 1.422 ± 148.7 ± 3.411.8 ± 9.72.3 ± 0.733.4 ± 19.9Grassland soil145.7 ± 79.187 ± 44.211.6 ± 13.214.6 ± 9.71.7 ± 0.90.5 ± 0.21.2 ± 0.78.9 ± 64 ± 3.83.4 ± 1.61.6 ± 0.631.1 ± 5.8Desert sand13.2 ± 9.680.2 ± 31.86.6 ± 4.37.9 ± 21.2 ± 0.60.2 ± 0.10.8 ± 0.34.5 ± 0.42.4 ± 1.51.7 ± 10.5 ± 0.517.7 ± 2.1Deep soil83.2 ± 62.3143.3 ± 84.129.6 ± 5.620.1 ± 6.60.9 ± 00.5 ± 0.11 ± 0.612.5 ± 5.76.1 ± 3.12.2 ± 0.41.3 ± 0.429.4 ± 5.6Complexed state (CS: alkali extracts), or stability of soil components**Forest soil517.4 ± 260.5640.8 ± 191.818.5 ± 7.695.9 ± 36.12.6 ± 1.21 ± 0.46.2 ± 2.27 ± 1.281.7 ± 32.911.7 ± 4.49.1 ± 4.6352.3 ± 58.5Agricultural soil570.8 ± 161.5909.5 ± 620.110.1 ± 2.383 ± 16.95.1 ± 3.30.7 ± 0.36.7 ± 2.96.5 ± 1.869.1 ± 15.913.9 ± 5.48.7 ± 2.3278.6 ± 134.3Grassland soil214.4 ± 143.3667.5 ± 6924.5 ± 12.731.7 ± 21.71.6 ± 0.70.2 ± 0.12.5 ± 1.77.9 ± 5.621.1 ± 15.45.9 ± 3.14.3 ± 3.5232.8 ± 61.8Desert sand84 ± 94.4582.1 ± 13.620.7 ± 6.114.5 ± 8.63.2 ± 2.10.2 ± 0.11.4 ± 0.63.9 ± 0.39 ± 7.74.1 ± 2.50.9 ± 0.8195.5 ± 8.3Deep soil320.4 ± 187.3668.5 ± 89.412.6 ± 6.552.7 ± 26.12 ± 10.3 ± 0.14.5 ± 0.25.2 ± 1.042.6 ± 25.27.3 ± 4.15.3 ± 3.9263.1 ± 100.4*Definition: Labile state (LS), or lability, is operationally defined as the net amount of water-soluble soil components extracted by water from the soil. Lability, or LS, typically refers to the fresh, soluble organic molecules derived from decaying plant materials, as well as readily available byproducts produced from various soil minerals under specific environmental conditions at the mineral-microbes interfaces. These components are not bound in organo-mineral complexes and are therefore susceptible to export into the surrounding environment through water and rainwater discharges. A low amount of LS soil components indicates low lability, while a high amount indicates high lability. For instance, lability, or LS of DOC (DOCLS) varies from 0.31 to 13.4 mg/L for all soils and sands.**Definition: Complexed state (CS), or stability, is operationally defined as the net amount of water-insoluble and alkali-soluble soil components extracted by alkali (0.1 N NaOH solution) from the soil. This typically refers to the water-insoluble complexed-state fractions of soil organo-mineral complexes that are not water-soluble and, therefore, are not susceptible to export into the surrounding environment through water and rainwater discharges. A low amount of CS soil components indicates a low stability (e.g. organo-, carbonate, silicate or secondary minerals), while a high amount indicates high stability. For instance, the stability, or CS of DOC (DOCCS) varies from 0.53 to 45.2 mg/L for all soils and sands. Specifically, DOCCS values are typically significantly higher, ranging from approximately 1.26 to 14.0 times (with an average of 4.42 ± 3.82 times) than DOCLS values across all soil types, except for one grassland sample (site IM-12), where DOCCS is lower (18.8%) than DOCLS. This discrepancy is attributed to substantially higher DICCS values, which are 13.4 times lower than DICLS. Average contents of labile state (LS: water extracts, or lability)*, and complexed state (CS: alkali extracts, or stability)** DOC, DIC, SO42−, and various nutrients in the forest, agricultural, grassland, and deep soils, and desert sands examined. *Definition: Labile state (LS), or lability, is operationally defined as the net amount of water-soluble soil components extracted by water from the soil. Lability, or LS, typically refers to the fresh, soluble organic molecules derived from decaying plant materials, as well as readily available byproducts produced from various soil minerals under specific environmental conditions at the mineral-microbes interfaces. These components are not bound in organo-mineral complexes and are therefore susceptible to export into the surrounding environment through water and rainwater discharges. A low amount of LS soil components indicates low lability, while a high amount indicates high lability. For instance, lability, or LS of DOC (DOCLS) varies from 0.31 to 13.4 mg/L for all soils and sands. **Definition: Complexed state (CS), or stability, is operationally defined as the net amount of water-insoluble and alkali-soluble soil components extracted by alkali (0.1 N NaOH solution) from the soil. This typically refers to the water-insoluble complexed-state fractions of soil organo-mineral complexes that are not water-soluble and, therefore, are not susceptible to export into the surrounding environment through water and rainwater discharges. A low amount of CS soil components indicates a low stability (e.g. organo-, carbonate, silicate or secondary minerals), while a high amount indicates high stability. For instance, the stability, or CS of DOC (DOCCS) varies from 0.53 to 45.2 mg/L for all soils and sands. Specifically, DOCCS values are typically significantly higher, ranging from approximately 1.26 to 14.0 times (with an average of 4.42 ± 3.82 times) than DOCLS values across all soil types, except for one grassland sample (site IM-12), where DOCCS is lower (18.8%) than DOCLS. This discrepancy is attributed to substantially higher DICCS values, which are 13.4 times lower than DICLS. DICLS and DICCS (Fig. 1; Table 2) showed the highest average contents in agricultural soils (111 and 910 mg/kg), which were higher compared to forest soils (by, respectively, 11.0% and 42.0%) > grassland soils (by, respectively, 27.2% and 36.3%,) > desert sands (by, respectively, 38.0% and 56.0%). In particular, deep soils showed the highest DICLS content among all soils (30.0% higher than agricultural soils), but their DICCS content was 36.0% lower than for agricultural soils (Fig. 1). Furthermore, forest soils ≈ grasslands ≈ deeper soils > desert sands (in the order) exhibited relatively similar levels of DICCS-induced mineral neoformation. In particular, the DOCLS and DOCCS values in desert sands (lower by, respectively, 1216% and 580% compared to agricultural soils) and the DICLS and DICCS content (respectively, 38% and 56% lower than in agricultural soils) would primarily arise from atmospheric CO2 uptake by sand photosynthetic microorganisms8,37,38. These processes would subsequently lead to the generation of HS/DOM8,37,38 and their simultaneous mineralization via photoinduced production of reactive oxygen species39. It would follow production of DICLS and DICCS, lability of LS forms, and sequestration in CS forms via neoformation of carbonate minerals as discussed before. The highly enriched δ13C-SOC (-11.37 ± 5.05‰) would further support the uptake of atmospheric CO2 by photosynthetic microorganisms in desert sands. Fluorescence features of HS components Up to three HSLS and HSCS components (HALS and HACS, FALS and FACS, as well as PLSLS and PLSCS), together with their degradative fluorescent components as byproducts or residues were identified by the EEM-PARAFAC in HS from all soils, which are similar to earlier studies22,28,40. Both HSLS and HSCS isolated from forest soils featured three components with only the long-wavelength peaks (T and M) of PLSLS being degraded, whereas all the other components (HALS and HACS, FALS and FACS, as well as PLSCS) were non-degraded or unaffected by environmental conditions (Fig. 2; Table S2), as shown in an earlier study28. Noteworthily, the higher fluorescence intensity (Fint) of peak M compared to peaks C and T in both LS- and CS-forms of FA from forest soils, along with the fact that Fint of LS showed the highest contribution among all soils (Fig. 3). Fluorescence (excitation-emission matrix, EEM) spectra of water-extracted/labile state (LS) humic substances (HSLS) (a), and alkali-extracted/complexed state (CS) HS (HSCS) (b) from forest, agricultural, grassland, and deep soils and desert sands. The HS components include humic acid-like (HA-like), fulvic acid-like (FA-like) and protein-like substances-like (PLS-like) identified using EEM-PARAFAC modeling of detected fluorescence peaks. Fluorescence intensities of humic acid (HA)-like peak C, fulvic acid (FA)-like peak M and protein-like substances-like peak T of water-extracts (labile state, LS) (a), and alkali-extracts (complexed state, CS) (b) from forest, agricultural, grassland, and deep soils and desert sands. The corresponding peaks (C, M and T) are shown in Fig. 2. Agricultural soils exhibited two HSLS components, i.e., degraded HALS with two peaks and degraded PLSLS with two short-wavelength peaks (TUV and A), as well as three HSCS components that were mostly unaffected by environmental factors (Fig. 2; Table S2). This issue suggests that human activities influenced HSLS components, but HSCS remained unaffected. A similar highly degradative nature of HSLS components in agricultural soils was reported elsewhere28,41. Furthermore, the HA peak C in the LS- and CS-forms of agricultural soils showed, on average, the highest (LS) and the lowest (CS) Fint values among all soil types (Fig. 3). This might happen because human and environmental factors (e.g., ploughing, sunlight) would increase the HALS content with a corresponding decrease of CS forms from which LS might derive, and also increase the FACS contribution to organo-minerals (Fig. 3). The extended mineralization of HSLS and HACS components might be responsible for the high generation of DICLS and DICCS in agricultural soils (Fig. 1). Furthermore, PLSLS from agricultural soils exhibited the second highest Fint among all soil types, which might also be ascribed to human activities42. These processes would finally lead to leaching or export of HALS and PLSLS from agricultural soils into the surrounding ecosystems, e.g., surface water and groundwater. Grassland soils showed three HS components in LS- and CS-forms, which is identical to forest soils (Fig. 2). These results would suggest the occurrence of very similar processes in both soil types. In particular, HS components might be freshly originated from grassland vegetation and anoxygenic photosynthetically-derived SOC under wet conditions, and might remain mostly in a non-degraded state under low temperatures and intense precipitations or wet conditions. In grassland soils the Fint of HA peak C, FA peak M, and PLS peak T of the CS-form were, on average, higher than for the corresponding LS-forms. Furthermore, HACS exhibited the most intense peak C among all soil types. These results would suggest a high contribution of all three HS components to organo-mineral complexes (Fig. 3). Simultaneously the three HSLS components, freshly derived from grassland vegetation under environmental conditions (Fig. 2), would contribute to a high level of DOCLS that might subsequently be exported to the surrounding environments via rainwater discharge from the grassland ecosystem. Desert sands featured two HSLS components that could be attributed to degraded HALS and degraded PLSLS (Fig. 2; Table S2), as well as three HSCS components, i.e., HACS, FACS, and PLSCS, which could be assigned to relatively degraded states. In fact, the peaks of HACS and FACS were at shorter wavelengths than those of forest soils (Table S2), as also observed in a previous study28. The two peaks (270/312 nm and 220/312 nm) of PLSCS might be mainly attributed to tyrosine-like substances (TYLS) (Fig. 2; Table S2)18. The highly degradative nature of HSLS and partly of HSCS components in desert sands might be due to the action of photochemically produced ROS39. In particular, the HS components in desert sands would primarily originate from photo-microbial respiration of photosynthetically-derived sand microorganisms, due to the absence of plant materials8,37,38, in a condition similar to that occurring for aquatic planktonic microorganisms18,43,44. The detection of TYLS in desert sands18 is different from the soils examined in this and earlier studies22,28. Moreover, the photochemical degradation of HSLS and HSCS components in desert sands was further supported by the relatively low Fint of HALS peak C and PLSLS peak T, and would be caused by constant sunlight exposure in the absence of plants together with the lack of leaching or export via rainwater discharge/runoff from desert sands to the surrounding environments. Finally, deep soils exhibited only one degraded HALS component with two peaks (C and A), occurring at relatively shorter wavelengths than the corresponding peaks of forest and agricultural soils28, but showing the highest Fint among all soil types (Fig. 3). Furthermore, deep soils featured three HSCS components at very similar wavelengths as those of forest and agricultural soils, which would suggest that HSCS in organo-minerals remained in a non-degraded state via organic carbon stabilization, together with the complete mineralization of FALS and PLSLS components. Noteworthily, the highest Fint of FACS peak M and the lowest Fint of PLSCS, which exhibited a highly microbially degradative nature in deep soils, along with the relatively high Fint of HACS peak C, would suggest a high contribution of organo-mineral complexes to organic C stabilization1,2,45. Overall, the wavelengths of peak A of HALS and HACS varied widely among the soil types studied, e.g. 280/505,495 and 280/477,486,460 nm for forest soils, as well as 260/430,439,448 and 280/495,505,477 nm for deep soils (Fig. 2; Table S2), which suggests the involvement of different functional groups36,46,47. Similarly, the wavelength of peak C of HALS and HACS also varied between forest soils (e.g. 355,380/505,495 and 350,375/477 nm) and grassland soils (e.g. 340,355,385/486 and 310,350/442 nm), which implies the occurrence of diverse functional groups. In particular, these groups would be primarily responsible for complexation with diverse metals47,48, mostly via π–d electron bonding systems40, with formation of organo-mineral complexes that result in the sequestration/stabilization of organic carbon in the soil matrix40. Characteristics of molecular weights of soil DOM components The Mn and Mw values of HS components were determined by HPSEC and are listed in Table 3, S4. The DOM elution curves showed similar trends for all soils, exhibiting a total of 2 to 5 peaks, and the MW distribution of their components ranged from 1 kDa to 35 kDa. In particular, three MW fractions, i.e., large (MW > 25 kDa), medium (MW 25 kDa‒15 kDa), and small (MW < 15 kDa) are distinguished. The fraction with Mw > 25 kDa was particularly rich in LS-forms (1.9–66.0%) compared to CS-forms (1.3–3.9%), in which case the large Mw fraction was only detected in seven samples (Table 3, S4). On average, the percentage of DOM components with Mw > 25 kDa in LS-forms followed the order: desert sands (40 ± 25%) > grassland soils (5.0 ± 2.2%) > deep soils (3.0 ± 0.9%) > forest soils (3.0 ± 0.7%) > agricultural soils (3.0 ± 0.3), whereas the order in CS-forms was: deep soils (3.0 ± 1.3%) > grassland soils (2.0 ± 0.9%) > forest soils (2.2%) > desert sands (2.0%). In particular, the high Mw of LS-forms from desert sands might be ascribed to degraded HALS, which is also shown by EEM-PARAFAC analysis. Differently, the high Mw of CS-forms mostly exhibited by deep and grassland soils might be ascribed to the relatively higher contents of HACS and FACS in non-degraded state (Figs. 2 and 3). Table 3Average values of weight average molecular weight (MW), and number average molecular weight (Mn) of dissolved organic matter (DOM) in labile state (LS: water-extracted) and complexed state (CS: alkali-extracted) from the forest, agricultural, grassland and deep soils, and desert sands examined.Soil typeLability or LS Mn (Da)Lability or LS Mw (Da)ρ-MQ> 25 kDa (%)15–25 kDa (%)< 15 kDa (%)Stability or CS Mn (Da)Stability or CS Mw (Da)ρ-NaOH> 25 kDa (%)15–25 kDa (%)< 15 kDa (%)Forest (n = 4)23,896 ± 11324,715 ± 431.03 ± 0.0030.03 ± 0.0070.96 ± 0.0070.012 ± 0.00120,836 ± 14022,017 ± 1651.05 ± 0.0030.020.92 ± 0.0070.071 ± 0.006Agriculture (n = 4)23,098 ± 29224,134 ± 2581.04 ± 0.0020.03 ± 0.0030.95 ± 0.0040.018 ± 0.00220,888 ± 62622,186 ± 6951.06 ± 0-0.92 ± 0.0140.076 ± 0.014Grassland (n = 4)24,250 ± 103224,995 ± 6371.03 ± 0.010.05 ± 0.0220.93 ± 0.010.014 ± 0.00420,993 ± 43222,253 ± 5041.06 ± 0.0040.02 ± 0.010.92 ± 0.0060.062 ± 0.01Desert (n = 5)29,085 ± 218929,322 ± 20381 ± 00.40 ± 0.250.76 ± 0.28-20,447 ± 275821,460 ± 20971.05 ± 0.040.020.90 ± 0.100.121 ± 0.10Deep (n = 3)24,605 ± 76624,995 ± 4821.01 ± 0.0120.03 ± 0.0090.96 ± 0.0090.008 ± 0.00322,437 ± 95923,563 ± 9211.05 ± 0.010.03 ± 0.0130.94 ± 0.0090.039 ± 0.012 Average values of weight average molecular weight (MW), and number average molecular weight (Mn) of dissolved organic matter (DOM) in labile state (LS: water-extracted) and complexed state (CS: alkali-extracted) from the forest, agricultural, grassland and deep soils, and desert sands examined. The fraction exhibiting Mw between 15 and 25 kDa showed, on average, slight differences among the soils under study, except for desert sands, and followed the order: deep soils (96.4 ± 0.9%) > forest soils (96.1 ± 0.7%) > agricultural soils (95.4 ± 0.4%) > grassland soils (94.0 ± 1.6%) > desert sands (76.0 ± 28.0%) for DOMLS, whereas for DOMCS the order was: deep soils (94.2 ± 0.9%) > agricultural soils (92.4 ± 1.4%) > grassland soils (92.4 ± 0.6%) ≈ forest soils (92.3 ± 0.7%) > desert sands (90.4 ± 10.3%). Furthermore, DOMLS components exhibited, on average, higher Mn and Mw (25211 ± 2826 and 25490 ± 2509 Da, respectively) compared to DOMCS components (21051 ± 1551 and 22229 ± 1330 Da, respectively). The higher occurrence of DOMLS components exhibiting Mw and Mn between 15 and 25 kDa, compared to DOMCS, might be due to their high lability and/or hydrophilicity and to their possibly aliphatic nature8,37,38. In particular, the highest values for HSLS were shown by desert sands (29085 ± 2189 and 29322 ± 2038 Da, respectively) and the lowest ones (23098 ± 292 and 24134 Da, respectively) by agricultural soils (Table 3, S4). This phenomenon would be due to the difference in DOMLS sources, i.e., photosynthetically-derived SOC in desert sands8,37,38 and plant -derived material in agricultural soils29–31. In the case of DOMCS, the Mn and Mw values in deep soils (22437 ± 959 and 23563 ± 921 Da, respectively) were higher than those in desert sands (20447 ± 2758 and 21460 ± 2097 Da, respectively; Table 3, S4), probably because the three DOMCS components remained in a non-degraded state due to their stabilization in organo-mineral complexes in deep soils (Fig. 2)5,6,49,50. In particular, the relatively low abundances of DOMCS components in desert sands might be due to intense sunlight-induced photochemical degradation by ROS39, which make reduce the stability of organo-mineral complexes and was further supported by the degraded states of the three DOMCS, which exhibited peaks at EEM wavelengths that were shorter than in the other soil types (Fig. 2; Table S2). Higher amounts of components with Mw < 15 kDa was measured in DOMCS (3.1–23.7%, average: 7.3 ± 5.0%) compared to DOMLS (0.6-2.0%, average: 1.4 ± 0.5%). These results are presumably due to the strong complexation properties of relatively low Mw components with various metals in soils, in comparison with high Mw components39. The content of the fraction with Mw < 15 kDa in DOMLS followed, on average, the order: agricultural soils > grassland soils > forest soils > deep soils, and was completely absent in desert sands. The corresponding order for DOMCS was: desert sands > agricultural soils > grassland soils > forest soils > deep soils. Specifically, the complete absence of the fraction with Mw < 15 kDa in DOMLS of desert sands and its highest amount in DOMCS would suggest that intense photochemical degradation by ROS39 would be responsible for the complete degradation of DOMLS, whereas the organo-mineral complexes of the corresponding DOMCS components would be relatively less labile46,51. Differently, the highest content of the small Mw fraction in DOMLS from agricultural soils might be ascribed to the degradation of organo-mineral complexes by the impact of human activities (e.g. ploughing, sunlight) on the uppermost soil, as discussed before. STS, SO42−LS+ CS, and SOS + Sn2− The STS contents ranged from 0.168 to 0.236 g.kg−1 and, on average, showed a slight variation among the studied soil types, i.e., the highest value was measured in agricultural soils (0.221 ± 0.013) which was higher (by 8.0%) than that in forest soils > deep soils (12.0%) > grassland soils (15.0%) > desert sands (24.0%). Similarly, SO42−LS varied from 1.86 to 916 mg.kg−1 and showed, on average, a large variation with the highest content in agricultural soils (248 ± 446 mg.kg−1), which was higher (by 738%) than in deep soils > forest soils (819%) > grassland soils (2040%) > desert sands (3650%) (Fig. 4). The highest STS and SO42−LS contents in agricultural soils might be primarily responsible for the relatively high DOCLS, DOCCS, and DICLS+CS production and export15,16,34 from the dissolution or mineralization of DOM-bound clay, carbonate and/or secondary minerals2,52–54 via microbial sulfate reduction55,56. Differently, SO42−CS ranged from 7.3 to 34.0 mg.kg−1 among all soils and showed the highest average content in grassland soils (24.0 mg.kg−1), which was higher (by 19.0%) than desert sands > forest soils (32%) > deep soils (94%) > agricultural soils (142%) (Fig. 4). The highest occurrence of SO42−CS in grassland soils might be primarily responsible for the high DOCCS mineralization and subsequent production of relatively high DICCS (668 mg.kg−1), which would end up in neoformation of carbonate minerals in soil55,56. Concentrations of labile state (LS) and complexed state (CS) sulfate (SO42‒LS and SO42‒CS), dissolved organic phosphorus (DOPLS and DOPCS), dissolved inorganic phosphate (DIP or PO43‒LS and DIP or PO43‒CS), and silicate ion (SiO32‒LS and SiO32‒CS) in forest, agricultural, grassland and deep soils and desert sands. Different lower case and capital letters (A, B or a, b) indicate significant differences at p < 0.05 level. Due to the small amounts of SO42−LS + CS, soil organic sulfur (SOS) plus sulfides (Sn2−) contents were very similar to the STS content, showing the highest levels in agricultural soils (0.204 ± 0.012 g.kg−1), which was slightly higher (by 7.8%) than that in forest soils > deep soils (11.0%) ≈ grassland soils (11.2%) > desert sands (18.4%). These results suggest that SOS + Sn2− were the predominant forms of S in soils, and contributed approximately 93.0–96.0% to STS. The significant positive correlations of STS with SOC, DOCLS, or DOCCS (Fig. 5; Table S3) suggest that the major STS fractions were partly derived from DOM and SOC-bound microorganisms. This result was further supported by the positive correlations of STS with STN and STH, which are primarily related to SOC plus DOM or HS. Furthermore, STS showed a significant positive correlation with SO42−LS and a negative correlation with SO42−CS (Table S3), which suggests that increasing STS via COS sequestration by lichen-associated cyanobacterial-fungal symbioses53,54,57 would increase the SO42−LS content that might be exported during DOM-bound mineral dissolution or mineralization via microbial SO42−CS reduction, associated with simultaneous sequestration (S2−-S°) and redox processes (S°-SO42−-S2−/S22−-SO42−)55,56. These results were further supported by the significant negative correlation of SO42−CS with DOCLS, which implies that DOMLS might arise from the dissolution or mineralization of DOM-bound clay and/or carbonate secondary minerals via microbial sulfate reduction. Therefore, the involvement of SO42−CS in the dissolution or mineralization of HS-bound minerals would affect the export of SO42−LS by leaching into the ambient surface waters and groundwater, whereas SOS + Sn2− would remain in the soil matrix in either mineral forms or as structural constituents of microorganisms. Pearson correlations of STS, SO42‒, SOS + Sn2‒ and SOC/STS ratios with some relevant parameters. The complete correlations between all parameters studied are presented in Table S3. STN, δ15N-STN, DON, and other nutrients The STN content varied from 0.070 to 2.16 g.kg−1 and showed, on average, the highest value in forest soils (1.51 ± 0.59 g.kg−1), which was higher (by 13.0%) compared to agricultural soils > deep soils (91.6%) > grassland soils (96.4%) > desert sands (1480%). These results suggest the occurrence of biogeochemical steady-state/sequestered N in the various soil types examined9,58, as well as the small14N loss via N2O or N2 emissions by HS-bound mineral dissolution or mineralization9,17,18,34. These processes are supported by the δ15N-PON values that, on average, were the highest in forest soils (6.21 ± 0.86‰) followed by those of agricultural soils (5.00 ± 1.69‰), grassland soils (4.29 ± 0.19‰), and deep soils (3.50 ± 1.02‰), whereas δ15N-PON could not be detected in desert sands, possibly due to the very low content of STN9,17,18,34. Lower δ15N values were measured in deep soils compared to top soils, as found in previous studies59,60, which would presumably arise from biological N fixation by specific N-fixing plant taxa59 via various bacteria61,62. The highest δ15N enrichment in forest soils, also reported in other studies63, and the wide variation of δ15N-PON values among the top soils might be ascribed to the rapid N-bound HS mineralization or dissimilation, accompanied by the preferential loss of the light14N isotope64 through small N2O or N2 emissions via denitrification60. These processes would produce NO3− with variable isotopic δ15N-NO3− 63. The remaining parts of this subsection are presented in the SI (subsection 3.1.5.). Discussion Sequestration of atmospheric CO2 in desert sands and HS-derived DIC in soils The δ13C-SOC values in desert sands (-17.63 to -7.10‰, average − 11.37 ± 5.05‰) were comparable with the atmospheric CO2 content (-8.4‰ in 2015)65, which suggests that atmospheric CO2 can be sequestered by sand microorganisms for their growth and metabolic respiration processes63. Such sequestration could for instance occur by COS uptake11,13,55 via reduction-oxidation reactions (S2−-S°-SO42−-S2−/S22−-SO42−)61,62 by lichen-associated cyanobacterial-fungal symbioses33–35,65,5,6,57,70. This interpretation is supported by the significant positive correlations of STS and SOS + Sn2− with SOC, and by their negative correlations with δ13C-SOC (Fig. 5; Table S3). This issue implies a simultaneous sequestration of S and C via COS uptake during photosynthesis50,66. Additional evidence is provided by the average SOC/S ratios in desert sands (5.7) compared to forest, agricultural, grassland, and deep soils (155–156, 75, and 9, respectively). These findings would suggest that desert sands mostly uptake atmospheric COS during photosynthesis50,66, whereas forest, agricultural, and grassland soils would use the abundant DIC or CO2 released from HS and DOM degradation30,31,33. Furthermore, the highly depleted δ13C-SOC values (-24.9 to -18.8‰: average − 22.7 ± 1.9‰) in the four soil types compared to desert sands (Table 1) are a key signature of sequestration/uptake of lighter CO2 arising from plant material-derived HS and DOM in soils33,35,67. Notably, due to the absence of plant material-derived DIC and CO2, the uptake of atmospheric COS would be the only pathway for microbial metabolic respiration processes and microorganisms growth in desert sands that also accounts for the observed S levels8,37,38. Due to the intense precipitation (445 mm in 2016) and relatively low temperature (highest temperature, 26.3–29.1○C in 2016) in Inner Mongolia (Table S2, S3), many HS components might be diagenetically produced from either terrestrial plant materials31,32,36 or anoxygenic photosynthetically-derived SOC, by soil or sand microorganisms under wet conditions (Table S1)31,32,36. They would then mostly remain in a non-degraded state in soil and sand. These conditions would lead to subsequent C sequestration via organo-mineral complexes29–31,37. Sources of SOC and molecular composition of exported DOM and HS The SOC basically includes soil microorganisms (e.g., bacteria, archaea, fungi, viruses, and microeukaryotes) and plant-derived materials, both of which simultaneously export DOM, HS and nutrients8,37,38. Specifically, SOC in desert sands mostly includes microorganisms that uptake atmospheric CO2 for their metabolic growth/respiration, as discussed before, and no plant-derived materials. Desert autotrophic microorganisms produce HS, which EEM images and corresponding peak positions differ from those of HS isolated from other soils (Fig. 2; Table S2)22,28,32 and rather resemble aquatic autochthonous humic-like and protein-like substances18,68. Desert HS would be produced rapidly and be subject to complete photochemical and microbial degradation within 24 h18,69,70. These effects might be the key factors for the high lability of DOCLS and DOCCS, as well as HSLS and HSCS components in desert sands. Differently, the higher fluorescence of peak M compared to peaks (C and T) in both LS- and CS-forms of FA from forest soils would suggest that individual HS components would contribute to the formation of organo-minerals as well as non-mineral states, while the highest contribution of fluorescence of LS FA peak M among all soils (Fig. 3) would suggest that there is higher export/leaching of FALS from forest soils into ambient environments. The higher Mw and Mn values of their LS-forms compared to the CS-forms (Table 3; Figs. 1 and 3) might presumably arise from the degradation of HALS, as also indicated by the absence of fluorescence EEM peaks (Fig. 2). The significant correlation found between DOCLS and DOCCS (Fig. 6) would suggest that both quantities are primarily originated during the mineralization of CS-forms, as reported elsewhere71. These results are also supported by the predominant occurrence of HSLS components with MW > 25 kDa and 15–25 kDa. Furthermore, the difference between HSCS components in desert sands and other mostly plant-derived HSCS was confirmed by the short-wavelength of peaks (C and A) of HACS, the degradation peak M of FACS, and the TLS peak, which resembled those of autochthonous DOM. These HSCS components mostly exhibited relatively low MW (15–25 kDa and < 15 kDa, Table 3, S4) and would originate from autotrophic microorganisms, which are primarily responsible for the relatively high lability of HSCS-bound organo-minerals in desert sands. The remaining parts of this subsection are presented in the SI (subsection 4.2). Pearson correlations between some most important parameters. The significance of the p-values are presented at the level p < 0.01 and p < 0.001. The complete correlations between all parameters studied are presented in Table S3. The role of natural and anthropogenic processes The key natural and human processes and factors occurring in soils primarily include sunlight-induced photochemical reactions, temperature, precipitation, microbial reactions, types of plant materials, ploughing, and cultivation (Table S5;26,35,39. Specifically, high contents of DOCLS and DOCCS in agricultural soils compared to other soil types and sands are primarily attributed to human activities (e.g., tilling, ploughing) which can subsequently influence the lability and export of DOM in LS-form, as well as the stability of organo-minerals. Due to the absence of plants in desert sands, sunlight plays a dual role, i.e., photosynthetic uptake of atmospheric CO2 by sand autotrophic microorganisms37,38,72 and simultaneous photodegradation of DOM and HS (produced by autotrophic microorganisms upon photosynthesis and further metabolic processing) via photoinduced generations of ROS39, as well as microbial neoformation of DOM and HS-bound organo-minerals2,73. In the case of deep soils, only microbial degradation would occur due to the absence of sunlight, while in forest, agricultural, and grassland soils the effect of sunlight would be operational but only in the top layers, where photosynthetically- and microbially-derived SOC might be produced under high humidity6,37,74,75. In particular, the DOM or HS originated from plant materials in these soil types would cause both organic C stabilization by neoformation of organo-minerals29–31 as well as simultaneous photochemical and microbial degradation, with final generation of DICLS and DICCS, DINLS and DINCS, DONLS and DONCS, PO43−LS and PO43−CS, and SiO32−LS and SiO32−CS (Figs. 1 and 4). In particular, the LS-forms would be partly exported into the surrounding environments (groundwater, surface waters, and air in gaseous forms, e.g. CO2, NOx), whereas a portion of LS-forms and CS-forms would contribute to the neoformation of organo- and/or carbonate minerals29–31. Sources of SIC and DICLS/DICCS, and mineral neoformation The highest average content of SIC in grassland and deep soils acted as the main storage of SIC. In particular, the high SIC content in deep soils might be considered a potential signature of inorganic C sequestration/storage as carbonate minerals33–35. The low SOC vs. high SIC content in deep soils, desert sands, and grasslands might be due to SOC heterotrophic microbial mineralization and subsequent generation of CO2 or SIC in deep soils14,15,45,76–78, where CO2 is sequestered in the form of clay and carbonate minerals14,45,76,77. In particular, an intense microbial mineralization of SOC in grassland soils is further supported by the relatively enriched δ13C-SOC values (-20.05 ± 0.9‰) compared to those in forest and agricultural soils (23.77 ± 0.88‰ and − 23.53 ± 1.41‰, respectively). The observed, high occurrence of DOCLS and DOCCS in agricultural soils (Fig. 1) and their subsequent mineralization via heterotrophic respiration and microbial sulfate reduction14,79 would lead to the generation of DIC or CO215,16,34 and to highly labile DICLS, which would undergo sequestration in CS forms via neoformation of carbonate minerals53,54,57. However, the long-term mineralization of DOCLS and DOCCS in deep soils would lead to the generation of DICLS or DICCS, which implies, respectively, the highest DICLS lability and the second highest organo-mineral neformation via DICCS in deep soils. The δ13C-SOC values measured in the soils and sands analyzed in this study (Table 1, S1), as well as in other studies, i.e., -27.7 ± 0.1‰ to -23.6 ± 0.3‰ during 3-180 days incubation and − 20.0 to -16.4 at 0–10 cm in forest and croplands33,34, in comparison with atmospheric CO2 (-8.4‰;65, suggest that photosynthetic microorganisms would mostly uptake the lighter/depleted13C-DIC or CO2 from soil. In contrast, desert sands collected from three sites where plants were completely absent (δ13C-SOC, -7.1‰ to -8.8‰), could uptake CO2 exclusively from the atmosphere, whereas the sandy soils (sand fraction, 96.79–99.96%; Table 1, S1) collected from two newly-planted nearby sites, following the desert afforestation program80 and relatively depleted in δ13C-SOC (-17.6 to -16.0‰) would uptake CO2 from both atmosphere and (depleted) plant-derived DIC or CO2. Therefore, desert sands could act as an effective sink of atmospheric carbon via carbonate formation81. Differently, all other soil types featuring13C-SOC values from − 24.9‰ to -18.8‰ (Table 1) would mostly uptake highly available DIC and CO2 derived from plant materials. Especially, grassland soils, which were relatively rich in sand (82.93–85.70%) and thus relatively rich in13C-SOC (-18.8 to -20.9‰), might uptake both atmospheric CO2 and DIC and CO2 from plant-derived DOM degradation15,16,31. Actually, enriched DIC and CO2 might originate from either emission of lighter or depleted CO2 into the atmosphere15,16, or carbonate mineral neoformation38,53,57. These findings are further supported by the significant positive correlation of δ13C-SOC values with the soil sand fraction and DICCS (Table S3), which also implies that the uptake of DIC and CO2 would involve CS-forms at the mineral-microbes interface during oxygenic/anoxygenic photosynthesis38,53,57. The occurrence of three processes (emission of CO2 to the atmosphere, carbonate mineral neoformation, and DIC uptake during photosynthesis) would primarily affect SIC storage, so that DICLS and DICCS did not show any significant correlation with SIC (Table S3). The remaining parts of this subsection are presented in the SI (subsection 4.4.). pH-dependent lability and stability of minerals The pH values showed significant negative correlations with SOC, STN, STS, and SOS + Sn2− for all soil types (Fig. 6; Table S3), which suggests the occurrence of pH-dependent C, N, and S sequestration, and of S in organo-mineral and sulfide forms, i.e., the lower the pH, the higher the sequestration and vice versa. In particular, the pH-dependent sequestration was primarily due to the increasing stability of mineral forms with increasing pH, whereas a decreasing pH would produce minerals destabilization40,82,83. The significant negative correlations of pH with SiO32−LS, SiO32−CS, NH4 + CS, and NO3−LS (Table S3) suggests high availability of these nutrients at low pH and vice versa. N- and P-nutrients would be extensively generated at low pH conditions, which is also supported by the significant negative correlations of pH with DONLS, DONCS, and DOPCS (Fig. 6; Table S3), in agreement with a previous study84. Furthermore, the significant positive correlation of pH with δ13C-SOC (Fig. 6; Table S3) would suggest that sequestration of lighter δ12C-SOC is favored at low pH conditions, while sequestration of enriched δ13C-SOC would be enhanced at high pH. These results provide a novel understanding of the export and sequestration dynamics of water-extractable DOCLS-normalized DICLS (and CO2) in soils. Sequestration of STS and SOS + Sn2−, and redox processses involving SO42−LS and SO42−CS The significant positive correlations of STS with both SOC and STN for all soil types studied (Fig. 5; Table S3) suggest that S sequestration could occur simultaneously with C and N sequestration in soils and sands. Similar results, i.e., the continuous STS increase with simultaneous C and N sequestration, were obtained under sunlight and microbial dark/control conditions for 150 consecutive days, whereas other authors have reported specific sequestration of S11–13, C8,37, and N9,10 under either oxygenic or anoxygenic photosynthesis5,6,85. These processes usually occur by lichen-associated cyanobacterial-fungal symbioses in soils5,6,49,85. The sequestration process is further supported by another experimental study conducted on forest (haplic luvisol) soil where a net increase in STS ranging from 1.2 to 41.0% and STN from 1.7 to 7.0% was measured under sunlight, dark, and control conditions86. Similarly, a net increase in SIC ranging from 3.1 to 10.1% was measured during the 0–75 day period under sunlight, and a 2.5% increase occurred during the 0–30 day period under dark conditions. Simultaneously, a net increase in SOC was measured only under sunlight (0.80%) and control (0.40%) conditions during the 75–150 day period following mineralization, with decreases of 1.8–2.4% and 2.3–6.0% over the initial 0–75 days. Notably, SOC sequestration under both sunlight and dark conditions simultaneously offsets microbial respiration, which contributes to the low carbon sequestration. Note that three experimental protocols—sunlight, dark, and control—were implemented using three aliquots of approximately 50 g each of homogeneously mixed, 2-mm sieved soil contained in 500 mL quartz bottles during the incubation periods of 0, 30, 75, and 150 days. The sunlight and dark samples were placed on the roof of a building at Tianjin University in Tianjin, China, while the control samples were maintained in a thermostat at a constant temperature of 25 °C86. Sulfur sequestration was further supported by the significant negative correlations of STS and SOS + Sn2− with δ13C-SOC (Fig. 5; Table S3), implying that lower sequestration of organic and mineral S might be associated with sequestration of enriched CO2, and the reverse. Furthermore, STS correlated significantly and positively with DOCLS, DOCCS, DONLS, DONCS, DOPLS, and DOPCS, thereby suggesting that increasing soil S would increase both LS- and CS-forms of organic N- and P-bound DOM, and vice versa. Soil S could thus control the occurrence of DOM and the photochemical and microbial degradative generation of its N- and P-nutrients, based on mutual positive correlations (Table S3). Most importantly, SO42−CS showed significant negative correlations with DOCLS and DOCCS (Fig. 5), which suggest that SO42−-mediated microbial reduction would mostly be responsible for the microbial mineralization of DOM, and in particular for the mineralization of SOC-derived HS-bound clay, Fe, and LDHs-like minerals via redox reactions5,6,49,85. The substantially lower SOC/S ratios in desert sands (5.6 ± 3.0) and deep soils (9.0 ± 6.0) compared to other soil types (72 ± 52–155 ± 42) are a key indicator of the substantially higher C and S sequestration in desert sands and deep soils compared to the other soil types. Furthermore, the significant positive correlations of SOC/S ratios with SOC, STN, STS, SOS + Sn2−, DOCLS, DOCCS, SiO32−LS, and SiO32−CS (Fig. 5) suggest the occurrence of C, S, and N sequestration, as well as their lability and/or stability via concurrent mineralization of mineral-associated components by STS-induced redox processes. Finally, organic and mineral S (SOS + Sn2−) contents showed significant positive correlations with SOC and STN for all soil types studied (Fig. 5), suggesting that organic and mineral S sequestration would also occur during the uptake of C and N. The persistent influence of STS or SO42−-mediated redox processes13,55,56,87 in the various soils, and even in desert sands, would imply that soil biogenic components always remain in a steady-state during the overall degradation processes. Therefore, S could play an important role in controlling biogenic components in soils and sands. Materials and methods Sampling sites and soil/land types The sampling sites are located in the southeastern edge of the Gobi desert, Inner Mongolia Autonomous region, China (40°14′-41°08′N, 109°56′-111°42′E). This region features a typical semi-arid and temperate continental climate, strongly affected by the East Asian monsoon and characterized by relatively intense precipitations, relatively short summers with relatively moderate temperatures and a long, extremely windy and cold winter season (Table S5). The temperature data of the tast 10 years (2012–2022) (Inner Mongolia Statistical Yearbook, 2022, http://tj.nmg.gov.cn) show an increasing trend of the summer peak temperatures since 2019, which reached the highest values of 25.6 in 2019 to 33.1○C in 2022 (Table S5). At the same time, winter temperatures are becoming progressively less rigid and ranged from the lowest value of -29.1○C in 2016 to -22.0○C in 2022. Very reasonably, these temperature variations might be affected by climate change. At the same time, annual precipitation in the 2012–2022 period showed relatively low variations, ranging from 500.3 mm in 2012 to 308.5 mm in 2017 with an average of 412.3 ± 53.5 mm. The main soil types in Inner Mongolia are chestnut soils, as classified by Chinese standards, which were developed from Aeolian sediments. This classification is further aligned with the World Reference Base for soil resources (WRB; Table S1). The soil grounds are characterized by a high content of sand, ranging from 60 to 75% 88. Notably, half of the region typically suffers from desertification that affects negatively about 90% of the local population, due to its secondary effects89. In particular, climate change and human activities (e.g., afforestation, deforestation, cropland and settlement expansion, overgrazing by livestock) and their interactions are considered the key responsible for desertification and land degradation in Inner Mongolia90. Other land types of the region include saline-alkali lands (0.59%), wetlands (5.08%), and unused lands (11.51%). Soil samples A total of 20 soil samples were collected in late October 2016 in the Inner Mongolia region, i.e., four soil samples from each of forested land, agricultural land, and grassland, three samples from deeper soil layers, and five samples from the Gobi desert sandy areas (Fig. S1). In particular, deep soils include a forest site with a depth of 1 m and two desert sand sites with depths of 1 m and 5 m, respectively. The sampling sites were selected by adopting the principle of sampling far away from the highways, to minimize human influence. Approximately 1 kg of top soil/sand (0–15 cm) was randomly collected from each sampling site using a 10-cm diameter stainless steel auger, after removing manually plant litters, roots, grasses, and debris, as described previously28. Three subsamples for each sample was then mixed homogeneously to obtain a spatially representative sample at the field scale. Soil samples were then oven dried at 40 °C, passed through a 2-mm sieve, ground in a mortar to yield particle sizes lower than 0.20 mm, and stored at -20 °C until further use. Forest soils, primarily planted with coniferous and deciduous broad-leaved trees, cover approximately 22% of total land (Inner Mongolia Statistical Yearbook, 2022, http://tj.nmg.gov.cn). Agricultural lands cover approximately 4.7% of total land, with the principal crops grown being beetroots, potatoes, tomato, cereal, wheat, rice, soybean, and tubers. Grasslands cover about 36.9% of the total, with the predominance of two grassland types, i.e., shrub 5–30% and herb 5–25% 91. The sandy desert covers 13% of total land, from which three samples were collected, one from bare sand (0–15 cm) and two (ID-4 and ID-7) from a nearby two-three years planted site. Desert afforestation via plantation in Inner Mongolia is a top priority of the Chinese Government, because afforestation in this area has potential advantages for dust mitigation and carbon sequestration80. Water and alkaline extracts of soil samples A novel method28, based on soil extraction with water first, and then with a 0.1 M NaOH solution was applied to separate the labile state (LS) from the complexed state (CS) soil components. The two extracts were obtained by several subsequent steps (Fig. S2). In the first step, ultrapure water (18.2 MΩ·cm, Mill-Q, Millipore) was added to each ground and sieved soil sample at a 1:10 soil: water ratio, the mixture was vortexed for 1 min in closed 500-mL brown bottles, shaken for 6 h at 25 °C, and then centrifuged for 20 min at 4000 rpm using a Thermo Fisher Scientific SORVALL ST 16 centrifuge, to remove suspended solids. The supernatant solution was then filtered through a 0.45-µm membrane filter (GF/F type, Shanghai Xin Ya Purification Equipment Co. Ltd, China), whereas the remaining solid residue was extracted again (repeating all the above procedure) with fresh ultrapure water for 1 h. We thus obtained a supernatant solution that was mixed with the previous one and stored in a freezer at − 20 °C until further processing. This solution represented the soil water extract (We). In the second step, the soil residue after water extraction was subjected to alkaline extraction under N2, with a 0.1 M NaOH solution at 1:10 soil residue: alkaline solution ratio, by shaking for 3 h at 25 °C. The mixture was then centrifuged as described above, and the supernatant solution was filtered through a 0.45-µm membrane filter (polytetrafluoroethylene membrane, PTFE, Shanghai Xin Ya Purification Equipment Co. Ltd, China). The remaining solid residue was extracted again with a fresh alkaline solution for 3 h, and the procedure described above was applied again to obtain another supernatant solution. The latter was mixed with the previous one to obtain the alkaline extract (Ae), and stored in a freezer at − 20 °C until further processing. All glass bottles used for extraction purposes were soaked with 10% HNO3 and HCl for 24 h before use, rinsed with Mill-Q (MQ) water, and then placed in a muffle furnace for 4 h at 550 °C, in order to remove any residual organic impurity. Similarly, each glass fiber filter was individually wrapped in an aluminum foil and then placed in a muffle furnace for 4 h at 550 °C, in order to remove any organic and inorganic impurities. The We and Ae extracts thus contained, respectively, the water-extracted LS soil components (i.e., DOCLS, HSLS, NO3−LS, NH4 + LS, NO2−LS, PO43−LS, SiO32−LS, and SO42−LS) and the alkali-extracted CS components (i.e., the corresponding DOCCS, HSCS, NO3−CS, NH4 + CS, NO2−CS, PO43−CS, SiO32−CS, and SO42−CS). A small portion of We and Ae was kept at 4 °C and then analyzed for the HS fractions by fluorescence, absorbance, and DOC within 48 h, whereas the residual main portions of We and Ae were kept at − 20 °C for further analysis of other parameters. In particular, because both We and Ae contained high DOC, these solutions were diluted to reasonable absorbance values92 before measuring fluorescence and other parameters. Analytical methods Physical and chemical analyses of soils Soil water content was measured gravimetrically in triplicate, in an oven dryer at 40 °C. The pH and EC values were measured in 1:2.5 slurries of soil and ultrapure water, using a pH meter (Multi 3630 IDS, Germany) equipped with a combined glass electrode93. Soil total carbon (STC) and soil total nitrogen (STN) were determined on approximately 20 mg of dried, ground, and homogenized sample contained in a clean, carbon-free pre-combusted tin boat, placed on the autosampler rack assembly of an Elemental Analyzer (Elemental Vario EL cube, Germany). Soil organic carbon (SOC) was analyzed by an Elemental Analyzer on dried, ground and homogenized samples pretreated with 1 M HCl placed in a vial on an autosampler tray reported elsewhere4. We used sulfanilamide as a standard after every ten measurements. Soil texture/particle size distribution was determined using the hydrometer method (Master sizer 3000, Malvern). The classification of soils was referred to the WRB Texture Classes, i.e., sand (63–2000 μm), silt (2–63 μm), and clay (< 2 μm). Stable carbon (δ13C) and nitrogen (δ15N) isotopes were measured, respectively, on approximately 2.0 mg and 30.0 mg of < 0.2-mm-ground soil samples, contained in a tin boat placed on the autosampler rack assembly of an Elemental Analyzer (Flash 2000 HT), interfaced with a Stable Carbon Isotope Ratio Mass Spectrophotometer (MAT 253 Plus, Thermo Fisher, Germany). For the δ13C analysis of SOC, carbonates were removed with an excess of HCl (2 M). The IAEA-600 Caffeine was used as δ13C standard, and both IAEA-600 Caffeine and USGS 40 & 41 L-Glutamic Acid were used as δ15N standard. All results were expressed in delta notation (Eq. 1): where Rsample and Rstd are the absolute ratios of δ13C/δ12C or δ15N/δ14N for sample and standard, respectively. and values are expressed in parts per thousand (‰) relative to the Vienna Pee Dee Belemnite (VPDB) international reference94. Analyses of We and Ae Dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) The concentrations of DOC and DIC were measured by an Aurora combustion total organic carbon (TOC) auto-sampler analyzer (O.I. Analytical Aurora, Model 1030, USA). An aliquot of sample was transferred to the reaction chamber and added with a predetermined volume of H3PO4. The acidified sample (pH ≤ 2) was heated at a temperature that is able to decompose carbonates and bicarbonates to CO2. After the reaction was completed, TIC was detected. After removing TIC, the strong oxidizing agent Na2S2O8 was added, which reacts quickly with organic C at 100 °C to generate CO2. At least three injections were performed for each DOC and DIC analysis. Sodium carbonate was used as the DIC standard solution and potassium hydrogen phthalate as the DOC standard solution. Before measurement, the instrument was calibrated with the standard at concentrations of 1, 2, 5, 10, and 20 ppm. Quality control and quality assurance were estimated by determining the contents in blank or standard reference materials. Nutrients Total nitrogen (TN), nitrate (NO3−), nitrite (NO2−), ammonia (NH4+), total phosphorus (TP), dissolved inorganic phosphate (DIP: PO43−) and silicate (SiO32−) concentrations were measured colorimetrically within 72 h from extraction, using an automated continuous flow analyzer (Skalar San + + System). The concentrations of dissolved inorganic nitrogen (DIN), dissolved organic nitrogen (DON), and dissolved organic phosphorus (DOP) were calculated by using Eqs. 2 and 3. Fluorescence EEM measurement and PARAFAC modeling The excitation emission matrix (EEM) fluorescence spectra were acquired by a fluorescence spectrophotometer (F-7000, Hitachi, Japan), equipped with a 700-W xenon lamp as the excitation source. The response time was set at 0.5 s and the slit widths of excitation and emission were set at 5 nm. The excitation wavelength was scanned from 220 to 450 nm at a scan speed of 1200 nm min− 1 and increments of 5 nm, whereas the emission wavelength was recorded from 280 to 550 nm at 1-nm intervals93. Ultrapure water (18.2 Mill-Q) used as a blank was scanned before sample analysis and after every ten samples, to check the instrument performance and ensure data quality. The EEM spectrum of ultrapure water was subtracted from the EEM spectrum of each sample. A quinine sulfate (QS) solution of 4 µg L− 1 in 0.01 N H2SO4 was used to normalize fluorescence, and the fluorescence intensities were calibrated using the intensity of the QS peak at Ex/Em = 350/450 nm (1 µg/L = 1 QS unit, QSU)95. The well-known scattering effect that appears when a high concentration of DOM (> 15–30 mg/L) is present in the sample requires the fluorescence spectra to be measured at low DOM concentrations (~ 10 mg/L)96–99. Furthermore, the inner-filter correction was obtained by diluting the analyte solution at a fixed absorbance value (0.1 cm−1) of the Ex wavelength. In particular, a satisfactory correction of primary and secondary inner filter effects can be achieved when the absorbance at the Ex wavelength of 254 nm is below approximately 0.3 cm−1 100. In this work, to avoid inner-filter effects and fluorescence quenching, fluorescence was measured after diluting the analyte solutions with reference to the initially measured DOC concentration32. Furthermore, each fluorescence peak intensity was rechecked and corrected using the common absorbance-based approach92. Parallel factor (PARAFAC) analysis was conducted on preprocessed EEM data using the N-way Toolbox for MATLAB101, described elsewhere102. First, Rayleigh and Raman peaks as well as the spectrum of an ultrapure water blank were subtracted from the experimental EEM spectra, using a homemade Excel program28. Afterwards, to avoid mixing of fluorescent components of different samples that might produce artifacts, PARAFAC analysis was performed on each soil sample by the appropriate procedures28,103. Finally, non-negative constraints were applied to the PARAFAC model. The detailed procedure used for PARAFAC modeling of EEM spectra is reported elsewhere28,103. High-performance size-exclusion chromatography (HPSEC) The Mw profiles were measured by a Shimadzu High Performance Liquid Chromatograph (HPLC) using an Acclaim Mixed-Mode HILIC-1 chromatographic column that was equilibrated at 25 °C. An aliquot of 100 µL of each We and Ae sample was injected, after previous filtration through a 0.45-µm membrane filter. The mobile phase used was a mixture of Na2HPO4·12H2O and NaH2PO4·2H2O at a concentration of 0.02 mol/L, adjusted to an ionic strength of 0.1 mol/L with sodium chloride. The mobile phase flow rate and the sample volume were set, respectively, at 0.5 mL/min and 100 µL. The wavelength of the UV detector was set at 254 nm. An important issue in the determination of the relative MW by HPSEC is to select a calibration compound with shape and structure similar to that of the molecule to be measured. In studying the relationship between the MWs of standards and samples, Chin et al.104 found that the molecular configuration of polystyrene sulfonate standards (PSS) was the most similar to that of HS, a significant constituent of DOM in natural environments. Based on previous research results, a number of polystyrenes of appropriate MWs (1100, 3610, 5580 and 33400 Da) were selected in this experiment, to construct a calibration curve and determine the relationship between retention time and MW (Fig S3). Compared to the ultrafiltration method to measure the MW of DOM, HPSEC has the advantages of simple sample pretreatment process, lower dosage (< 100 µL), rapidity (~ 15 min), good reproducibility, and high sensitivity. In addition, HPSEC could simultaneously provide the weight-averaged molecular weight (Mw), the number-averaged molecular weight (Mn), and the polydispersity coefficient ρ that represents the degree of dispersion of the MW distribution of DOM. The Mw, Mn, and ρ values were calculated from the HPSEC-UV chromatograms according to Eqs. 4–6 105. where hi is the height of the chromatogram at the elution time i, MWi is the apparent MW in Da corresponding to the elution time i, and n is the number of data points. The apparent Mw values were estimated from the calibration curve in the elution time range 3.0–10.0 min, for which an excellent fitting equation was achieved. Thus, the calculated Mw, Mn, and ρ did not consider the fraction below an elution time of 3 min (biopolymers), whereas a minor fraction appeared at elution times between 3 and 8 min (neutrals) (Fig. 3S). GIS and statistical analysis The ArcGIS (ArcMap 10.5) method was used to locate the spatial distribution of sampling sites. All data were subjected to the analysis of variance (ANOVA) and Pearson correlation analysis (PCA), using the SPSS21.0 software. The significance levels were reported as significant at p < 0.05 and highly significant at p < 0.01. Data were analyzed by multivariate analysis and descriptive statistics, using the Origin 2018 software. Figures were plotted using Origin 2018 and GraphPad Prism 9. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements This study was supported by the National Natural Science Foundation of China (Grant No. U1612441), the Ministry of Science and Technology of China (Grant No. 2019YFC1804400) and also by the Key Construction Program of the National “985” Project, Tianjin University, China. Abbreaviations Labile state, i.e. water extracted fractions (We) Complexed state, i.e. alkali extracted fractions (Ae) Dissolved inorganic carbon Dissolved inorganic nitrogen Dissolved inorganic phosphorous Dissolved organic carbon LS DOC CS DOC LS and CS DOC Dissolved organic matter Dissolved organic nitrogen LS dissolved total nitrogen CS dissolved total nitrogen Fluorescence intensity Fulvic acids LS and CS FA Humic acids LS and CS HA High performance size exclusion chromatography Humic substances LS and CS HS Number average molecular weight Weight average molecular weight Molecular weight LS and CS PLS Soil/Sand organic carbon Soil/Sand total hydrogen Soil/Sand total nitrogen Soil/Sand total sulfur LS and CS SO42− Author contributions Conceptualization and Supervision: K.M.G.M., C.Q.L. Writing—original draft: K.M.G.M., N.S., X.Y. Investigation and formal analysis: X.Y., X.G. Funding acquisition: K.M.G.M., W.Z., S.L.L., C.Q.L. Validation: X.Y., X.G., K.M.G.M., W.Z., N.S., G.S.S., D.V., J.Y., S.L.L, L.L., C.Q.L. Writing-review & editing: W.Z., G.S.S., D.V., S.L.L. PARAFAC analysis: J.Y. Data availability All data is provided within the manuscript or supplementary information files. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Khan M. G. Mostofa, Email: mostofa@tju.edu.cn Cong-Qiang Liu, Email: liucongqiang@tju.eu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All data is provided within the manuscript or supplementary information files.