Chemical Composition of Labile Carbon Fractions in Forest Soils as Affected by Soil Parameters

Understanding how the chemical composition of dissolved and particulate organic matter (DOM and POM) is affected by environment factors is critical because these labile pools of carbon are involved in an array of biological, chemical and physical processes. In this study, the chemical composition of DOM and POM was measured in 13 forest soils using UV-Vis spectroscopy, uorescence spectroscopy with PARAFAC modelling and FT-IR spectroscopy. There were signicant differences between the soils for the SUVA indexes, PARAFAC components and relative intensities of different IR bands. Redundancy analysis (RDA) revealed that soil parameters had a great inuence on the chemical composition of DOM and POM with high constrained variability (77.9 and 77.1 %, respectively). The pH of the soils proved to be an important controlling factor for both DOM and POM, regulating the concentration of the C3 PARAFAC component (low-molecular-weight compounds associated with biological activity) and the aromatic compounds of POM (aromaticity, rA 1630 and rA 1515 ). The silt content was the other main regulating factor controlling the chemical characteristics of the labile pool, having a strong negative correlation with the SUVA values of DOM due to the preferential adsorption of hydrophobic moieties. RDA analysis also revealed that, despite their different origins, there is a strong correlation between the chemical composition of POM and DOM. for SUVA 280 in the individual soils. The highest values were obtained for the CEG, NYIR1 and JOS2 samples and the lowest values for BAT, SOP3 and KIS.


Introduction
Soil organic matter (SOM) has a number of functions in the environment, in connection with soil quality functions such as fertility (Tiessen et al. 1994), buffering capacity (Ritchie and Dolling 1985) and structural stability (Six et al. 2000). Furthermore, SOM plays a potential role in the release and sequestration of CO 2 , through the decomposition of organic matter or the accumulation of carbon by capturing plant residuals (Lal et al. 2015). However, soil organic matter is not a homogeneous material but consists of molecules ranging from the simple to the more complex in very different stages of decomposition from recognizable plant residues to low-molecular-weight carbohydrates and proteins (Lehmann and Kleber 2015). The labile pools of soil carbon can be considered as the active carbon in soils and have central role in short-to medium-term nutrient availability and soil structural stability. Also, these pools are sensitive indicators of minor changes in both the climate or local environment (Haynes 2005;Fang et al. 2005) and soil quality (Filep et al. 2015), so there is increasing interest in how it responds to such changes.
Dissolved organic matter (DOM) is a complex mixture of low-and high-molecular-weight organic molecules originating from litter, soil leachates, plant root exudates and microbial by-products (Thurman 1985; Guggenberger et al. 1994). The dissolved organic matter fraction is often de ned operationally as the organic carbon fraction that can pass through a lter of about 0.45 µm pore size (Kalbitz et al. 2000).
DOM cycling in terrestrial ecosystems is of particular interest in the light of a changing climate, because DOM is one of the largest sources of available organic carbon for microbes (De Troyer et al. 2011;Xenopoulos et al. 2021). Particulate organic matter (POM) is also considered as a rapidly changing pool of organic matter, which could respond rapidly to environmental changes (Six et al. 2002). This fraction contains organic matter with a size of > 63 µm and a density of > 1.6 g/cm3 (Christensen 1992). POM makes a substantial contribution to soil organic matter, typically comprising > 50% of SOM in mineral soils and > 70% in sediments (Hayes et al. 2017). Particulate organic matter can be considered to be in a kind of metastable phase (Huang et al. 2019), meaning that although it is insoluble, it is chemically and biologically active, e.g. it can adsorb metals via several mechanisms (Guo et al. 2006;Zhao et al. 2021).
Radiocarbon dating suggests that the DOM and POM pools are not connected, as the average age of DOM is 5-40 years, while POM may be hundreds or thousands of years old (Lu et al. 2014). Furthermore, research has provided additional evidence for the lack of connection between DOM and POM, showing a clear compositional difference between them, which could be explained by their different sources and different degradation pathways (Feng et al. 2016). Matiasek and Hernes (2019) reported that fractionation may occur during the solubilisation of particulate-bound organic matter to dissolved organic forms, as indicated by the distinct amounts of amino acids and lignin in POM and DOM.
However, Li et al. (2018) pointed out that despite the compositional differences in DOM and POM, they are nevertheless coupled to some extent, exhibiting a similar shift in diagenetic status. They may be linked via processes such as dissolution, sorption/desorption, aggregation and disaggregation. More direct evidence is thus needed to elucidate whether DOM and POM are connected or not.
In this study, the spectroscopic characteristics of the dissolved organic matter and particulate organic matter fractions of thirteen forest soils from Hungary were evaluated as affected by soil parameters. The chemical composition of the pools was analysed by combining UV-Vis, uorescence and FT-IR spectroscopy, using PARAFAC modelling. Redundancy analysis was used to reveal which soil factors control the chemical compositions of these labile pools and what connection there is between them. The speci c objectives were: (i) to gain an insight into the relationship between soil parameters and the chemical composition of the labile fraction of the soils, (ii) to evaluate whether the chemistry of POM and DOM is connected.

Site description and sampling
Thirteen forest topsoil samples were collected from Hungary. More details about the sites are available in Zacháry et al. (2018). Samples were taken from the upper 0 − 20 cm horizon and the soils were air-dried, homogenized, passed through a 2-mm sieve and stored at room temperature.

Soil fraction preparation
The dissolved organic matter was extracted with ultrapure water at a 1:10 soil:solution ratio for 2 h and ltrated through a 0.45 µm membrane nylon lter. Particulate organic matter was fractionated according to Zimmermann et al. (2007). Brie y, soil samples were added to distilled water and dispersed. After dispersion the suspension was wet-sieved over 63-µm sieve. The > 63 µm fraction was dried at 40°C and separated using NaI at a density of 1.6 g cm − 3 . The POM was then rinsed with distilled water to remove NaI and subjected to further analysis.

Soil analysis
The soil pH was measured in 1:2.5 soil:water and soil:1M KCl supernatants 12 h after mixing. The total organic carbon (TOC) content was analysed using an elemental analyser (Apollo 9000, Tekmar Dohrmann). The particle size distribution was determined by the pipette method (Gee and Or 2018). The total N content was measured by the Kjeldahl method (Bremner 2019). The dissolved organic carbon (DOC) and total soluble nitrogen (TSN) were analysed in a TOC/TN analyser (TOC-L, Shimadzu). Cation exchange capacity (CEC) was determined according to the method of Gillman (1979). The basic parameters of the soils are presented in Table 1. Spectroscopic analysis and PARAFAC modelling UV-Vis spectrophotometric analysis of the DOM samples was performed with a UV-3600 dual-beam scanning spectrophotometer (Shimadzu Corp.) using a 1 cm quartz cuvette with deionized water as the reference. The speci c UV absorption (SUVA 254 and SUVA 280 , L mg − 1 m − 1 ) was calculated by dividing the absorption at 254 and 280 nm by the DOC concentration.
The uorescence excitation-emission matrices (EEM) of DOM were measured with a RF-6000 instrument (Shimadzu Corp.). The EEMs were obtained by measuring uorescence intensity excitation wavelengths ranging from 230-450 nm and emission wavelengths ranging from 260-600 nm with 2 nm increments. After blank subtraction and the correction of scattering the EEMs were Raman normalized using the area under the water Raman peak at an excitation wavelength of 275 nm. The PARAFAC modelling was conducted with MATLAB (Mathworks, Natick, MA) using the drEEM toolbox (Murphy et al. 2013). A nonnegativity constraint was applied to allow only chemically relevant results. The correct number of components was determined using the core consistency diagnostic score, which should be close to 100% for appropriate models (Bro and Kiers 2003). The validity of a PARAFAC model is often evaluated using split-half analysis. However, split-half analysis is problematic for validating a model if the dataset is from geographically diverse locations, which leads to highly variable uorescence (Dubnick et al. 2010). This is probably why none of the models explored in this study gave satisfactory results in the split-half analysis, thus preventing it from being used as a validation tool. Table 2 shows the uorescence characteristics of the six components obtained by PARAFAC analysis. range of 4000-400 cm − 1 , a resolution of 4 cm − 1 , 128 scans, and three replicates were recorded. The spectra were corrected (atmospheric water and CO 2 correction) and subjected to 17-point Savitzky-Golay smoothing, SNV normalization and baseline correction.
The relative absorbance (rA) and aromaticity index were calculated to evaluate relative changes in the spectra. Relative absorbance was calculated by dividing the given peak height (2920, 1710, 1630, 1515, 1270 and 1160 cm − 1 ) by the sum of the absorbance of all peaks measured and multiplying it by 100: The aromaticity index (Inbar et al. 1989) was calculated by dividing the intensity of absorption at 1630 cm − 1 (aromatic C = C) by the intensity of absorption at 2920 cm − 1 (aliphatic C-H).

Statistical evaluation
One-way analysis of variance (ANOVA) with a post hoc Duncan test was used to evaluate differences between the spectroscopic parameters of DOM and POM (SUVA, PARAFAC components and relative intensities of IR bands).
The relationship between soil parameters and the chemical composition of DOM and POM was analysed using redundancy analysis (RDA). The spectroscopic parameters of DOM and POM (SUVA, PARAFAC components and relative intensities of IR bands) were selected as response variables, while the parameters of the forest soils (pH, SOC, clay, silt and sand content, carbon to nitrogen ratio) were applied as explanatory variables. All the response and exploratory variables were standardised as suggested by Braak and Smilauer (2012). The collinearity of the variables was checked using the variance in ation factor (VIF), and variables having VIF < 20 were used for further analysis.

Chemical composition of DOM and POM fractions
The molar absorptivity of the DOM fraction of the samples was measured at 254 and 280 nm (Table 3). Besides the traditional 254 nm, where a group of much weaker bands appear for benzene, 280 nm was chosen because π-π* electron transitions may occur for phenolic substances, aromatic structures and polycyclic aromatic hydrocarbons (Weishaar et al. 2003   For the band at 1160 cm − 1 , which can be assigned to vibrations in the glucosidic C-O-C bond and in the whole glucose ring (Pandey 1999; Olsson and Salmén 2004), great variation was found between the soils, ranging from 9.9 to 31.8.
In most of the forest soils, a low aromaticity index was calculated by dividing the intensities of the peaks of 2920 and 1630 cm − 1 . Although the band in the 1630 cm − 1 region was very mixed, representing coupled vibrations, for example the C-C stretching of aromatic rings or the C = O vibrations of amides, a strong relationship was found between the intensity ratio of 1620 and 2920 cm − 1 and the ratio of aryl and alkyl C obtained from NMR measurements (Dick et al. 2006).

Effect of soil parameters on the composition of labile fractions
To evaluate the effect of soil parameters on the chemical composition of the labile fractions in the soils redundancy analysis was performed for both DOM and POM (Figs. 2 and 3). Based on VIF values, the pH H2O , TOC, carbon to nitrogen ratio, clay and silt content, and CEC were selected as explanatory variables. The full RDA model explained more than 75 % of the total variance for DOM composition (77.9 %). A strong correlation was found between soil pH and the C3 component, a negative correlation between the pH and the C1 and C4 components. There was a strong negative correlation between soil texture, expressed as clay and silt content, and both the SUVA values and the C2 component of PARAFAC.
The cation capacity of the soils (CEC) was found to be closely correlated with the C5 component of the DOM fraction, while the C/N ratio was correlated with the C4 and C1 components.
For POM, redundancy analysis showed that the constrained variability was 77.1 %. As in the case of DOM, the pH appeared to be the main controlling factor for chemical composition in POM: several relative absorbance values, rA 1630 , rA 1515 , rA 1270 and the aromaticity index, were governed by pH, though the cation exchange capacity (CEC) was also found to be a controlling factor for these absorbance values and for aromaticity. A strong correlation was found between the carbon to nitrogen ratio of the soils and the relative absorbance of the band at 1710 cm − 1 . The silt content was in close negative correlation with rA 1630 , rA 1515 , rA 1270 and the aromaticity index. The clay content also played an important role in regulating POM quality, as indicated by the correlation between clay content and rA 2920 .

Connection between the compositions of POM and DOM
Redundancy analysis on the spectroscopic parameters of the DOM and POM fractions (Fig. 4)  The connection between higher microbial activity and the C3 component was con rmed by the fact that the C1 component was negatively correlated with pH, suggesting that the high-molecular-weight compounds in DOM could be the primary source for biological degradation. In addition to pH, low N availability was con rmed as a limiting factor for microbial degradation, because the C/N ratio of the soils was found to be in strong correlation with the C1 component, indicating a connection between N status and microbial activity. The biological activity, controlled by pH, also affected the concentration of the C4 component, which exhibited a negative correlation with pH.
Protein-like ourescences have been reported to be a useful proxy for measuring the biodegradibility of DOM (Fellman et al. 2008(Fellman et al. , 2009; Chen and Jaffé 2016), but a more complete picture emerged from the present analysis, the data of which not only revealed a possible connction between the PARAFAC components, but also how organic matter is degraded, with high-molecular-weight compounds (represented here as C1) being degraded into lower molecular-weight compounds (C4), after which the increased microbial activity results in a high level of microbial by-products such as peptides and amino acids (C6 component) (Figs. 1 and 2).
In this study the soil texture was found to be a signi cant controlling factor for the composition of the DOM fraction, as con rmed by the strong negative correlation between the silt and clay contents and the SUVA values of DOM and the C2 component (Fig. 2) due to the preferential sorption of large, hydrophobic organic compounds (Jardine et al. 1989;Kaiser and Zech 2000). Avneri-Katz et al. (2017) found a signi cant reduction in SUVA 254 values, con rming the accelerated adsorption of hydrophobic compounds suggested by the present evaluation.
In the case of POM, the organic matter was found to be selectively decomposed, which could lead to an increase in the relative abundance of compounds with high resistance to microbial degradation, such as the aromatic and phenolic compounds represented by rA 1630  structures. In addition to pH, the cation exchange capacity of the soils proved to be a further regulator of the relative amounts of aromatic and phenolic compounds (Fig. 3). CEC is well known as a dominant factor that stimulates bacterial respiration by maintaining the pH, replacing the H + ions produced during metabolism with basic cations (Stotzky 1966). Due to this mechanism the less degradable aromatic moieties were abundant in POM in soils with high CEC, e.g. in CEG and NYIR2.
Several studies demonstrated a strong relationship between microbial respiration and either the C/N ratio of the litter layer (Gödde et al. 1996;Michel and Matzner 2002;Spohn 2015) or the available N (Craine et al. 2007). These nding are in accordance with the present results: the strong, negative correlation between the soil C/N ratio and the relative intensity of the band at 1160 cm − 1 (representing polysachharides) clearly demonstrated the enhanced degradation of easily decomposable materials, such as sugars, in soils with higher N availability. This was con rmed by the study of Gallo et al. (2005), who reported a common microbial response to higher N: the activity of cellulase and other glycosidases tended to increase. In parallel with this, the activity of oxidative enzymes tended to decline (Saiya-Cork et al. 2002). This was clearly demostrated in the present study by the positive correlation of the C/N ratio with the relative intensity of the band at 1710 cm − 1 , which represents oxidative materials; in other words, in soils with high N availability the activity of oxidative enzymes is restricted, so the amount of carboxyl groups in POM is low.
Is the chemical composition of POM linked with that of DOM?
Particulate organic matter is considered to be one of the major sources of dissolved organic carbon (Zsolnay 2003). It has long been known that several bacteria are able to decompose the native lignin in the soil (Brauns 1952;Sørensen 1962) and that the decomposition of POM in soils can result in several aromatic compounds such as lignin-derived materials, tannins and phenols (Kaiser et al. 2001; Kalbitz et al. 2006). It is thus not surprising that the relative amount of aromatic compounds (rA 1515 , rA 1630 for aromatic rings and rA 1270 for phenolic compounds) in the POM fraction of forest soils was closely correlated with the SUVA values of the DOM fraction (Fig. 4). Based on the thermochemolysis data of forest soils Klotzbücher et al. (2013) stated that at least half the aromatic carbon comes from litter. Although Matiasek and Hernes (2019) reported different solubilisation patterns for lignin phenols, indicated by the C:V ratios (ratio of cinnamyl phenols to vanillyl phenols), which were six-fold lower in DOM, the optical characteristics of these compounds were probably identical.
The degradation of lignin-carbohydrate complexes is a prerequisite for the solubilisation of lignin, a process is linked with the degradation of cellulose, resulting in a relatively labile C source partly shielded by lignin (Jeffries 1991). Furthermore, lignin oxidation is a metabolic process that requires easily degradable carbon sources (Kirk and Farrell 1987). Klotzbücher et al. (2013) also suggested that lignin degradation and solubilisation could be related to the production of soluble carbohydrates that provide energy for microbes. In the present study, these enzyme-mediated processes were con rmed by the The study also revealed that, in addition to factors directly or indirectly in uencing the biodegradation of organic matter, purely physico-chemical properties also in uenced the composition of the DOM pool through adsorption. A complementary pattern was found between the C1 and C2, and the C3 and C4 PARAFAC components of DOM, suggesting biological linkage between them. This was con rmed by RDA, which provided a clear picture of the transformation and coupling of various biodegradable DOM components, high-molecular-weight compounds (C1) being degraded into lower molecular-weight compounds (C4), and increased microbial activity resulting in a high level of microbial by-products such as peptides and amino acids.
Despite their completely different origins and diagenetic status, a strong relationship was observed between the chemical characteristics of POM and DOM, suggesting that although they represent different levels of biodegradation, they are biogeochemically coupled.

Con ict of interest
The authors declare no con ict of interest. .