Increased Iron-Carbon Interactions Under Long-Term Acid Deposition Enhance Soil Organic Carbon Sequestration in A Tropical Forest in Southern China

Atmospheric acid deposition remains a widespread problem that may inuence the protection of carbon (C) in soil by altering organo-mineral interactions. However, the impacts of additional acidity on organo-mineral interactions and soil C sequestration in naturally acidic tropical soils with a high content of reactive iron (Fe) phases have not been well studied. Here we sampled a nearly 10-yr eld experiment with a gradient of acidity treatments (0, 9.6, 32, 96 mol H + ha − 1 yr − 1 as nitric acid + sulfuric acid) to examine how acidication alters organo-mineral interactions and soil organic carbon (SOC) pools in a tropical forest in southern China. As expected, soil acidication signicantly enhanced the leaching of base cations (e.g., Ca 2+ ), and it also altered the solubility and composition of Fe and Al phases. The acidity treatments converted more crystalline Fe (oxyhydr)oxides to short-range-ordered phases, resulting in a large increase in Fe-bound C vs. a relatively small decrease in Ca-bound C. Overall, the acidity treatments increased the mineral-associated C stock to 32.5–36.4 Mg C ha − 1 vs. 28.8 Mg C ha − 1 in the control, accounting for 71–83% of the observed increase in total SOC stock. These ndings highlight the importance of pH-sensitive geochemical changes and the key roles of Fe in regulating the response of SOC to further inputs of acid deposition even in highly weathered and naturally acidic soils. The magnitude of SOC changes observed here indicates the importance of including pH-sensitive geochemistry in Earth system models to predict ecosystem C budgets under future acid deposition scenarios.


Introduction
Since the industrial revolution, human activities such as fossil-fuel combustion and the use of chemical fertilizer have greatly increased nitrogen (N) and sulfur (S) inputs as acid rain into the terrestrial biosphere (Likens et al. 1996; Galloway et al. 2008). These acid inputs have signi cant effects on the function and structure of natural ecosystems (Vitousek et al. 1997). In particular, long-term acid deposition can aggravate soil acidi cation, which has multiple biological and geochemical impacts on soil biogeochemical cycles (Bowman et  Recent studies and conceptual frameworks underscored the importance of acid-sensitive organo-mineral interactions in regulating soil carbon (C) dynamics (Averill and Waring 2018). Soil pH plays a central role in understanding the multiple potential impacts of acidi cation on soil C. For example, in alkaline soils buffered by calcium carbonate (inorganic C), N-induced soil acidi cation accelerates calcium carbonate dissolution and releases a substantial CO 2 amount from the soil at the global scale (7.48 Tg C yr − 1 ; Zamanian et al. 2018). In Chinese croplands, Raza et al (2020) estimated a 7% (0.15 Pg C; 1.1 Mg C ha − 1 ) loss of carbonate C due to soil acidi cation from 1980 to 2020. In soils with alkaline to circumneutral pH, electrostatic bridging between soil organic matter and clay mineral surfaces by divalent base cations (Ca 2+ , Mg 2+ ) provides a primary mechanism of physico-chemical protection for SOC (Rasmussen et al. 2018; Yu et al. 2021). In these alkaline to circumneutral soils, inputs of acidity (H + ) are mostly buffered by replacement of base cations on ion exchange sites (Chadwick and Chorover 2001; Kirk et al. 2010), possibly releasing SOC bound to clay minerals via divalent cations. However, in acidic soils, exchangeable base cation concentrations are typically much lower, divalent cation bridging is less important for SOC protection (Rasmussen et al. 2018;Yu et al. 2021), and soil acidity is mainly buffered by the dissolution of silicates and aluminum (Al)-or iron (Fe)-bearing minerals (Johnson 2002;Ross et al. 2008; Li and Johnson 2016). The release of hydrolyzing cations (e.g., Fe 3+ and Al 3+ ) at low pH may favor subsequent formation of organo-mineral complexes through ligand exchange between carboxyl groups of organic matter and singly coordinated inorganic hydroxyl groups at mineral surfaces (Kaiser and Guggenberger 2000;Mueller et al. 2012). Consistent with these ideas, a recent study in a northern grassland with circumneutral pH found a large decrease in Ca-bound C vs. a slight increase in Fe-bound C following soil acidi cation (Ye et al. 2018). However, the impacts of additional soil acidi cation on organo-mineral interactions in highly weathered soils that are naturally acidic have not yet been addressed (Kuzyakov et al. 2021).
In China, about 40% of the total territory is affected by acid rain due to the continuing increase in anthropogenic activities (coal combustion, industrial emissions, and automobile exhaust) in recent decades. Particularly in southern China, acid deposition was currently estimated as 34.4 kg N ha − 1 yr − 1 and 32.6 kg S ha − 1 yr − 1 (Jiang et al. 2018), comparable with the highest levels of acid deposition in Europe and North America (Duan et al. 2016). High rates of acid deposition have increased public concern about its widespread impacts on the natural ecosystems in southern China . and play an important role in global C balance (Yu et al. 2014). Soils in these forest ecosystems are acidic (often with pH < 4.5, and some even with pH < 4) and are highly weathered with a high content of reactive Fe phases (Zhang et al. 2010;Hu et al. 2015). Thus, Fe-bound C is likely to comprise an important portion of the total SOC stock in these forests (Coward et al. 2017). However, whether the mineral phases and organo-mineral associations in these highly acidic soils still respond signi cantly to further inputs of acid deposition has not been tested, which may have important implications for forest soil C sequestration in southern China and in other global regions with naturally acidic soils.
We carried out a long-term (nearly 10 yrs) eld experiment with a gradient of acidity treatments (0, 9.6, 32, 96 mol H + ha − 1 yr − 1 as nitric acid + sulfuric acid) to examine acidi cation impacts on organo-mineral interactions and SOC pools (divided light and heavy fractions) in a tropical forest in southern China. Earlier work at this study site demonstrated that the acidity treatments signi cantly increased SOC stock and altered its biochemical composition (Wu et al. 2016). In this study we hypothesized that the acidity treatments enhanced SOC sequestration mainly arising from increased C in heavy (mineral-associated) fraction. We also expected that pH-sensitive organo-mineral interactions, especially changes in Fe phases, could explain the responses of SOC pools. Speci cally, under the acidity treatments we quanti ed 1) the shift of mineral-associated C pools (Fe-bound C vs. Ca-bound C) and its absolute contribution to the increase in total SOC stock, and 2) the extent of cation losses and changes in Fe phases.

Site description
This study was conducted in Dinghushan National Nature Reserve (1133 ha), located in the middle part of Guangdong Province in southern China (112°30′ -112°33′E, 23°09′ -23°11′N). This area is characterized by a typical south subtropical monsoon climate. The mean annual temperature is 21°C, with the maximum and minimum average monthly temperature of 28.0°C in July and 12.6°C in January, respectively. Annual average relative humidity is 82%, and the mean annual rainfall is 1927 mm. Acid rain is a threat in this area with high deposition rates of 34.4 kg N ha − 1 yr − 1 and 32.6 kg S ha − 1 yr − 1 in recent decades (Liu et al. 2007), which has signi cantly lowered the soil pH of some forests to < 4 (Jiang et al.

2018).
In the reserve, there are three types of forests of different successional stages: the broadleaf forest, mixed pine and broadleaf forest, and pine forest. In this study, the experimental site was set up in the broadleaf forest, the most mature forest with age > 400 year. The forest is located 250-300 m above sea level and occupies approximately 600 ha. The dominant species were Castanopsis chinensis, Cryptocarya concinna, Cryptocarya chinensis, Machilus chinensis, and Schima superba (Yan et al. 2006). The bedrock is sandstone and shale belonging to the Devonian Period. Soil type is lateritic red earth (Yan et al. 2015) and is classi ed as an Oxisol according to the Keys to Soil Taxonomy (Soil Survey Staff 2014).

Experimental treatments and design
The acidity experiment was initiated in June 2009. Twelve plots were established and divided into four acidity treatments with three replicates each. Each treatment plot was measured 10 × 10 m 2 and was surrounded by a 3 m wide buffer strip. All plots and treatments were randomly arranged. The acidity treatments were irrigated with water of different pH values: CK (control, pH 4.5), T1 (pH 4.0), T2 (pH 3.5), and T3 (pH 3.0). To re ect the real mole ratio of S:N in the region, acidic solutions were prepared by adding a mixture of H 2 SO 4 and HNO 3 in a 1:1 mole ratio to the local lake water based on the local acid rain records (Du et al. 2015). The simulated acid rain was applied to each plot below the canopy using a gasoline engine powered sprayer and sprayed twice a month. The amount applied to each plot was 40 L per application, equal to 4 mm rainfall. The H + added in each plot was estimated as 0, 9.6, 32, 96 mol H + ha − 1 yr − 1 in the T1, T2, T3 and T4 treatments, respectively, which was equal to about 0, 0.6, 2.0 and 6.0 times of that in the through-fall of the forest. More details on the experimental design and methods can be found in Wu et al (2016).

Sample collection and soil properties measurements
In September 2018, four cores (diameter = 5 cm) were randomly collected from the topsoil (0-10 cm) and subsoil (10-20 cm) in each plot, and combined to yield one composite sample per depth and plot. The soil sample was passed through a 2 mm sieve to remove rocks and plant roots, and then divided into two parts ( eld-moist soil sample and air-dried soil sample) for subsequent analysis. Fine roots (diameter ≤ 2mm) were picked and thoroughly rinsed in deionized water and dried at 60°C to constant biomass. Soil moisture was determined by drying the eld-moist soil samples at 105°C for 24 h. Soil pH values were measured using a glass electrode after shaking the samples for approximately 30 min in deionized water. The soil to water ratio was 1:2.5. Microbial biomass carbon (MBC) was measured using the chloroform fumigation extraction technique (Jenkinson 1987). The soil extractable dissolved organic carbon (DOC) was measured on the same samples used for the analysis of MBC and calculated as the K 2 SO 4 -extractable C concentration. Total SOC was measured with the air-dried soil samples by using the K 2 Cr 2 O 7 + H 2 SO 4 oxidation method (Schollenberger 1927

Density fractionations
Each eld-moist soil sample (30 g dry mass equivalent) was separated into two operationally de ned soil fractions, a light fraction (LF) and a heavy fraction (HF), following a modi ed density fractionation technique (Ye et al. 2018). Given that the soils used in this study were rich in clay, the LF was separated by otation after immersing soils in NaI solution at a density of 1.85 g cm − 3 combined with a ultrasonic treatment (a total energy input of 200 J mL − 1 ) in order to disrupt soil aggregates. The residual soil consisted of the remaining mineral-associated organic matter (HF). The separated soil fractions were then dried in an oven at 60°C and ground to a homogenized ne powder for SOC analysis (interpreted as LF-C and HF-C).

Determination of Ca-bound C and Fe-bound C
The HF soil samples were sequentially extracted with 0.5 M Na 2 SO 4 and the C released in the solutions was interpreted as organic C associated with Ca bridges (Xu and Yuan 1993). The difference between HF-C and the C concentrations of the residual soils after Na 2 SO 4 extraction was calculated as Ca-bound C.
Next, Fe-bound organic C was measured following a procedure adapted from Lalonde et al (2012). The residual soil after Na 2 SO 4 extraction was added to a solution containing sodium bicarbonate and trisodium citrate in a 50 mL polycarbonate centrifuge tube and heated to 80°C in a water bath, and then sodium dithionite was added to the tube and maintained at 80°C for 15 min. After centrifugation at 3000 g for 10 min, the supernatant was separated from the solid fraction. The procedure was repeated three times and the residual soil was rinsed three times with deionized water and then oven dried at 80°C for organic C analysis, respectively. The concentrations of organic C in soil after dithionite-citrate-bicarbonate (DCB) extraction were subtracted from the Na 2 SO 4 -extracted samples to obtain Fe-associated organic C.

Determination of Fe phases
During the determination of Fe-bound C, the extracted solution was also analyzed by ICP-OES to measure the total free Fe (oxyhydr)oxides (Fe d ). Dry HF subsamples were separately extracted with acid ammonium oxalate or sodium pyrophosphate to measure poorly crystalline (i.e., short-range-ordered,

Statistical analyses
Data analyses were conducted using SPSS 20.0 for Windows (IBM Corporation, Armonk, New York, USA) and R version 4.0.2. Analysis of Variance (ANOVA) was used to determine the statistical signi cance (α = 0.05) of the acidity treatment, soil layer and their interactive effects on bulk SOC and its fractions/pools as well as mineral elements (Ca, Al and Fe including Fe oxides). Tukey's multiple comparison test (HSD) was conducted if signi cant effects of acid addition or soil layer were found. Pairwise relationships between soil pH and biogeochemical variables, as well as biogeochemical variables and SOC fractions/pools were assessed using Pearson correlation coe cients.
A structural equation modelling (SEM) approach was also used to test a conceptual model for the acid deposition impacts on LF-C and HF-C. The SEM analysis was performed with the IBM SPSS Amos 22.0 using the maximum likelihood estimation method. Several tests were used to assess model t: the Chisquare (χ 2 )-test, comparative t index (CFI) and root square mean error of approximation (RMSEM).

Soil general characteristics
Soil pH values decreased signi cantly under the acidity treatments compared to the control (p < 0.05, Table 1), and pH was signi cantly lower in topsoil than in subsoil (p < 0.05, Table 1). There was no signi cant difference in soil moisture or bulk density among treatments (p > 0.05 for both, Table 1 Table 2). The exchangeable Ca 2+ in topsoil was signi cantly lower in T1 than in T3 treatment (p < 0.05, Table 2), but neither signi cantly differed from the control (p > 0.05, Table 2). In subsoil, exchangeable Ca 2+ in the control was signi cantly higher than in the T1 and T2 treatments (p < 0.05, Table 2), but not in the T3 treatment (p > 0.05, Table 2). Exchangeable Al 3+ and Fe 3+ in both topsoil and subsoil increased signi cantly with the acidity treatments compared to the control (p < 0.05 for both, Table 2).

Soil organic carbon fractions
The total SOC content increased signi cantly under the acidity treatments compared to the control, except for the T1 treatment in both topsoil and subsoil (p < 0.05, Fig. 1a). Both LF-C and HF-C concentration and stock also increased signi cantly under the acidity treatments compared to the control, except for the HF-C in the T1 treatment in subsoil (p < 0.05, Fig. 1b, c). Since the bulk density did not change among treatments (p > 0.05, Table 1), the stock of SOC and its fractions varied similarly with their content (Fig. 1). Overall, the acidity treatments increased LF-C to 3.1-4.1 Mg C ha − 1 vs. 1.6 Mg C ha − 1 in the control and increased HF-C to 32.5-36.4 Mg C ha − 1 vs. 28.8 Mg C ha − 1 in the control. The increase in LF-C and HF-C accounted for 22-32% and 78-83%, respectively, in the increase of the total SOC stock.
The content of Ca-bound C in the topsoil showed no signi cant change among treatments (p > 0.05, Fig. 1d), while in the subsoil it decreased signi cantly under acidity treatments compared to the control (p < 0.05, Fig. 1d). Fe-bound C in both topsoil and subsoil increased signi cantly with the acidity treatments compared to the control (p < 0.05, Fig. 1e).

Relationships among the SOC fractions and geochemical characteristics
Across all treatments, soil pH was negatively correlated with SOC, LF-C, HF-C and Fe-bound C, but not with Ca-bound C (Fig. 2a). SOC was positively correlated with both LF-C and HF-C (Fig. 2b,c). HF-C was positively correlated with Fe-bound C only, but not with Ca-bound C (Fig. 2d,e).
Soil pH was also negatively correlated with exchangeable Al 3+ , exchangeable Fe 3+ , Fe(II) HCl , Fe(III) HCl , and Fe ca , while total Ca, total Fe, total Al, and Fe p showed opposite relationships (Fig. 3a). The Ca-bound C was positively correlated with the loss of total Ca only, but not with exchangeable Ca 2+ (Fig. 3b). However, the Fe-bound C was positively correlated with exchangeable Fe 3+ and Al 3+ (Fig. 3c,d). The Fe-bound C was also positively correlated with the concentrations of Fe(II) HCl , Fe(III) HCl , and Fe ca (Fig. 3e,f,g).

Controls on SOC and fraction changes under the acidity treatments
The SEM model implied by our data showed that the decreased soil pH under the acidity treatments was the optimal predictor and directly explained 53%, 63% and 83% of the variance in ne root biomass, Febound C, and LF-C, respectively (Fig. 4). There was no signi cant relationship of soil pH with MBC or Cabound C. Fine root biomass had a direct negative relationship with MBC (R 2 = 0.46), whereas neither ne root biomass nor MBC was signi cantly related to LF-C. Change in LF-C directly explained 50% of the variance in DOC, which in turn together with Fe-bound C directly explained 81% of the variance in HF-C.
There was no signi cant relationship of HF-C with MBC and Ca-bound C, respectively.

Discussion
We found that the acidity treatments signi cantly increased total SOC content and stock including both light and heavy fractions (Fig. 1a, b, c). The increased SOC was unlikely to have come from greater plant C input, as soil acidi cation was previously shown to limit plant productivity and led to decreased ne root biomass as reported in this study ; Table 1). The acidity treatments also decreased MBC (Table 1) and could inhibit soil respiration (Wu et al. 2016), which might also have contributed to the observed SOC accumulation. However, decreased MBC might be expected to cause higher LF-C but lower HF-C, because microbial residues derived from catabolism of LF-C can potentially be important constituents contributing to mineral-associated C (Cotrufo et al. 2013;Liang et al. 2017). In contrast to this reasoning, we found that the HF-C increased much more in absolute amount than the LF-C under the acidity treatments. Consistent with our hypothesis, the acidity treatments increased the HF-C stock, which accounted for most of the total increase in SOC stock (Table 4).
With a SEM model implied by our data, we further found that pH-sensitive organo-mineral interactions particulatly changes in Fe phases, to a large extent, explained the increase of HF-C under the acidity treatments. The increase in Fe-bound C under the acidity treatments was much larger than the decrease in Ca-bound C (Fig. 1d, e). Consistent with pH buffering by dissolution of Al and Fe phases in acidic soils, we found that the acidity treatment generally decreased exchangeable Ca 2+ , while soil extractable Al 3+ and Fe 3+ increased signi cantly (Table 2). In an alpine grassland of Europe with soil pH < 3.5, acid deposition similarly led to buffering by Al 3+ and Fe 3+ (Bowman et al. 2008). Increases of extractable Al 3+ and Fe 3+ under the acidity treatments re ect the enhanced solubility of Al and Fe mineral phases (Gu et al. 1994;Mueller et al. 2012), causing losses of these metals via leaching as indicated by decreases in total Al and Fe in the subsoil (  (Qiu et al. 2013(Qiu et al. , 2015, and here we found positive relationships between soil exchangeable Al 3+ and Fe 3+ with mineral-associated C, consistent with DOC sorption or co-precipitation (Fig. 3). In addition, soil acidi cation could decrease the solubility of organic matter due to the protonation of organic acid functional groups at pH values approaching zero net charge (Chorover and Sposito 1995), as well as by increasing molecular aggregation (Louzao et al. 1990).
Changes in Fe phase compositions under the acidity treatments were also likely related to the large increase in Fe-bound C pool in this study (Fig. 4b). We found that the acidity treatments signi cantly decreased Fe d−ca while Fe(II) HCl , Fe(III) HCl and Fe ca increased (Table 3). These contrasting changes in Fe pools likely re ect the progressive transformation of crystalline Fe oxides into increasingly short-rangeordered Fe phases. Decreased Fe d−ca indicates that soil acidi cation increased the solubility of crystalline Fe oxides, as observed in other acid forest soils (Guo et al. 2007). Intriguingly, we also found that Fe p decreased under the acidity treatments, possibly indicating competition between H + and Fe 3+ for binding sites in soil organic matter under these highly acidic conditions, leading to release of Fe 3+ from monomeric organo-Fe complexes and its subsequent precipitation in Fe phases that were not extracted by Fe p . The formation of new short-range-ordered Fe phases was consistent with the observed increase in the Fe ca and Fe(III) HCl pools, which include highly reactive portions of soil Fe (Hall and Silver 2015). The fact that Fe(II) HCl was a small portion of the 0.5 M HCl-extractable Fe pool indicated that acid-mediated dissolution, rather than reductive dissolution, was likely the dominant process controlling the observed changes in soil Fe (Table 3) C under the acidity treatments (Fig. 3d,e,f).
Our results have also important implications for understanding the impacts on SOC sequestration of ongoing acidi cation from atmospheric N deposition in tropical forests. At the global scale, soil pH decreases linearly with N addition (Tian and Niu 2015), and at our study site, a parallel experiment with N addition alone also revealed signi cant soil acidi cation (Lu et al. 2014). N-induced increases in SOC have been detected from temperate to tropical forests, but the underlying mechanisms in tropical forests remain poorly understood (Lu et al. 2021). Because many tropical forests are more strongly limited by phosphorus (P) than N, increased plant productivity and C input of plant residue cannot likely explain increases in SOC (Lu et al. 2021). N-induced soil acidi cation is widely known to suppress certain aspects of microbial decomposition activity, such as production of oxidative enzymes , and these changes have previously been invoked to explain the increased SOC stock and the change of biochemical composition under N addition (Cusack et al. 2011). Rather, a recent study by Lu et al (2021) showed that in tropical forests N addition increased SOC sequestration, mainly arising from increased C in heavy (mineral-associated) fraction accompanied by acidi cation-induced the decreases in both soil CO 2 e ux and DOC leaching. Here our data suggest that pH-sensitive organo-mineral interactions, especially formation of new Fe-C associations that provide protection from SOC decomposition and DOC leaching, play a key role in regulating the response of SOC pools in tropical forests and explain the observed increases in mineral-associated C.

Conclusions
Using a long-term acidity manipulation eld experiment, we found that increased inputs of acid deposition in a highly weathered and naturally acidic tropical soil could enhance SOC sequestration, which mainly arising from increased C in mineral-associated fraction. Accordingly, soil organo-mineral interactions responded signi cantly to further inputs of acid deposition, with increased the leaching of base cations (e.g. Ca 2+ ) and the solubility of Al and Fe phases. These acidity treatments converted more crystalline Fe (oxyhydr)oxides to short-range-ordered phases, resulting in a large increase in Fe-bound C vs. a relatively small decrease in Ca-bound C. With a SEM model implied by our data, our results further highlight the importance of pH-sensitive geochemical changes and the key roles of Fe in regulating the response of SOC to acid deposition in highly weathered soils. Therefore, accounting for pH-sensitive geochemistry is critical to predict ecosystem C budgets under future acid deposition scenarios, and that these principles could be usefully incorporated in mechanistic biogeochemical models.
Declarations Figure 1 Responses of soil organic carbon fractions to the acidity treatments. The acidity treatments are: CK = control (0 mol H+ ha−1 yr−1), T1 = 9.6 mol H+ ha−1 yr−1, T2 = 32 mol H+ ha−1 yr−1, and T3 = 96 mol H+ ha−1 yr−1. SOC, soil organic carbon; Fe-bound C, iron-bound organic carbon; Ca-bound C, calcium-bound organic carbon; LF-C and HF-C indicate the organic carbon in light fraction and heavy fraction, respectively. Error bars represent ± standard errors. The columns and scatters represent soil organic carbon content and stock, respectively. Different lowercase letters within the columns in each soil layer denote signi cant difference (p<0.05) in the contents of soil organic carbon fractions among the treatments. Different capital letters above the scatters denote signi cant difference (p<0.05) in the stocks of soil organic carbon fractions among the treatments. As none of the interactions between treatment and depth on soil organic carbon fractions were signi cant, only main effects of treatment and soil depth are presented. *, ** and *** indicates that factors are signi cant at p<0.05, p<0.01 and p<0.001, respectively.

Figure 2
Pearson's correlation coe cients between soil pH and soil organic carbon fractions. SOC, soil organic carbon; Fe-bound C, iron-bound organic carbon; Ca-bound C, calcium-bound organic carbon; LF-C and HF-C indicate the organic carbon in light fraction and heavy fraction, respectively. The red and blue squares indicate positive or negative correlations between parameters at a signi cance level of α=0.05 respectively.

Figure 3
Pearson's correlation coe cients between soil pH, Ca-bound C, Fe-bound C and geochemical characteristics. Fe-bound C, iron-bound organic carbon; Ca-bound C, calcium-bound organic carbon; Feca, short-range-order Fe phases; Fed, Dithionite-bicarbonate-citrate extractable Fe phases; Fep, sodium pyrophosphate extractable Fe phases; Feo, ammonium oxalate extractable Fe phases. The red and blue squares indicate positive or negative correlations between parameters at a signi cance level of α = 0.05 respectively. In addition, R2 values associated with response variables indicate the proportion of variation explained by relationships with other variables. MBC, microbial biomass carbon; Fe-bound C, iron-bound organic carbon; Ca-bound C, calcium-bound organic carbon; LF-C and HF-C indicate the organic carbon in light fraction and heavy fraction, respectively.