Organic Fertilization Enhances Microbial-Mediated Dissolved Organic Matter Composition and Transformation in Paddy Soil

Abstract

Dissolved organic matter (DOM) is a crucial carbon source for soil microorganisms and plays a vital role in nutrient cycling and carbon (C) sequestration in soils. However, the extent to which soil microbes mediate DOM transformation at the molecular level, and whether this is regulated by different organic fertilization, remains unclear. Here, we designed a field experiment to investigate the transformations of DOM under three types of organic fertilization (straw, biochar, and manure) using Fourier transform ion cyclotron resonance mass spectrometry and metagenomic analysis. Compared to the control, manure fertilization increased the molecular chemodiversity of DOM by 33.2%, with recalcitrant compounds (e.g., highly unsaturated phenolic compounds and lignins) increasing by 47.2%. In contrast, labile compounds (e.g., aliphatics) decreased by 73.5%. Compared to straw treatment, manure application significantly increased the average conversion rate of dissolved organic matter (DOM). This process was accompanied by a significant increase in the Shannon index of the soil microbial community (p < 0.05) and upregulation of ABC transporter-encoding genes (e.g., livK, livM). DOM composition directly governed transformation potential (p < 0.01), whereas functional genes enhanced transformation indirectly by modulating DOM composition. This study elucidates microbial-mediated DOM transformation mechanisms under varying organic fertilization practices, providing a scientific basis for optimizing soil organic matter management in paddy ecosystems.

1. Introduction

Globally, paddy soils store approximately 18 Pg of carbon, accounting for 14.2% of the total carbon present in agricultural soil ecosystems [1]. The carbon cycle in paddy soils plays a crucial role in maintaining soil fertility and crop productivity. As a pivotal active constituent of the soil carbon cycle, the chemical composition and bioavailability of dissolved organic matter (DOM) mediate microbial metabolic pathways, thereby influencing carbon cycling processes [2,3]. DOM represents a highly dynamic pool of organic compounds, an essential carbon source for soil microorganisms [4]. Previous studies have emphasized the role of organic fertilizers in enhancing DOM concentration and microbial utilization [5,6]. However, the highly active and diverse DOM molecules complicate a comprehensive understanding of their transformations and microbial interactions in soils [7]. Transformation potential refers to the capacity of DOM molecules to undergo biochemical transformations, thereby influencing their composition and stability [8]. Existing research indicates that small-molecule DOM compounds of plant origin undergo preferential degradation, gradually yielding more persistent DOM molecules. Furthermore, extant research suggests that among the explicable factors, microbial community characteristics explain the most significant proportion (approximately 25%) of DOM molecular variation, surpassing soil physicochemical factors (approximately 19%) and root biomass (approximately 10%). These patterns suggest that DOM persistence stems not from inherent resistance to degradation, but from ongoing microbial consumption, transformation, and synthesis [9]. However, questions remain: How does microbial community structure dynamically adjust in response to shifts in DOM molecular composition? Do ‘key species’ and functional genes dominate DOM transformation processes?

During the initial stages of straw decomposition, DOM entering the soil predominantly consists of hydrophilic components (such as carbohydrates, amino acids, and proteins), exhibiting high biodegradability. In contrast, during later decomposition phases, the input DOM is dominated by aromatic compounds, which are difficult for microorganisms to degrade due to their biochemical inertness, leading to their persistent accumulation in the soil [10]. DOM molecules derived from manure are rich in lignin-like molecules, lipids, and protein-like compounds [11]. Studies indicate that long-term manure application significantly enhances DOM recalcitrance compared to inorganic fertilizer or unfertilized treatments, manifested as increased humification and enhanced aromaticity [12]. Biochar derived from plant feedstocks typically yields DOM molecules exhibiting stronger aromatic and aliphatic characteristics [13]. Given these differences in molecular composition and degradability, the environmental fate of these organic materials when applied to soil requires investigation through long-term laboratory or field trials to comprehensively understand the interactions between DOM molecules and microorganisms.

Organic fertilizer applications have been shown to modify the molecular composition of soil DOM, thereby enhancing its transformation potential [14]. Microbial communities significantly influence the transformation and composition of DOM in soil, such as converting C/H dissolved organic molecules to reduce their C/H [15,16]. Keystone taxa are particularly influential within these microbial communities, shaping network structures and ecosystem functions [17]. Keystone taxa are often central to critical carbon cycling processes, such as decomposing complex organic compounds and synthesizing new organic matter [18]. Microbial DOM utilization patterns, particularly by keystone taxa, depend on substrate availability and environmental conditions, reflecting the intricate interplay between microbial ecology and carbon dynamics [19].

Transformation potential refers to the capacity of DOM molecules to undergo biochemical transformations, influencing their environmental persistence, degradation rates, and contribution to soil organic matter cycling. It is commonly expressed as the maximum number of transformations a DOM molecule can undergo. Unlike bioavailability, which pertains solely to the fraction directly utilized by microorganisms, transformation potential encompasses all components and reactions ranging from readily degradable to recalcitrant [8]. Fertilizer management significantly impacts DOM transformation potential. Long-term application of organic fertilizers increases the maximum number of transformations for DOM, yet simultaneously enhances carbon stability by promoting the conversion of lignin into tannins and aromatic compounds, thereby reducing its biodegradability [20]. Unraveling the mechanisms governing DOM molecular transformation provides a scientific foundation for optimizing agricultural management (e.g., fertilization) and addressing climate change (e.g., enhancing carbon sink functionality).

Understanding the interactions between microbial functional genes and DOM biochemical transformations is essential for accurately forecasting soil carbon dynamics. Functional genes regulate enzymatic processes that control organic matter decomposition and synthesis, thus influencing DOM composition and stabilization. Previous researchers have shown that quantitatively demonstrated this mechanistic connection using structural equation modeling, achieving high predictive accuracy for carbon-degrading enzyme activities based on functional gene abundance [21]. Earlier studies have indicated significant associations between DOM-related functional genes (e.g., those involved in glycolysis, lignocellulose degradation, and anaerobic carbon fixation pathways) and carbon cycling parameters [22]. Previous studies have demonstrated that the abundance of relevant functional genes within soil microbial communities can effectively predict changes in enzyme activity involved in carbon degradation [21]. Extracellular enzymes initially hydrolyse DOM into small molecules capable of crossing membranes, thereby initiating microbial transformation of DOM [23]. Among these, phenol oxidases utilize oxygen as the terminal electron acceptor, catalyzing the oxidation of recalcitrant aromatic compounds into more readily utilized substrates. Concurrently, by degrading phenolic compounds, they indirectly relieve the inhibition exerted on extracellular hydrolases, thereby promoting the release and enhanced activity of these enzymes. β-glucosidase (β-GLU) catalyzes the final step in cellulose degradation, mediating the release of glucose monomers to microorganisms [24]. Research has confirmed the existence of multiple anaerobic methane oxidation (AOM) pathways in paddy field anaerobic environments, driven by various electron acceptors, including nitrate, nitrite, and Fe³⁺. Among these, nitrate-dependent AOM constitutes the predominant pathway [25]. Regarding resource allocation and succession, different bacteria exhibit specific utilization of DOM: during the initial phase, Idiomarina and Alteromonas primarily participate in fatty acid degradation and tricarboxylic acid (TCA) cycle-related processes; by the late stage, Methylophaga predominantly utilizes one-carbon compounds [26]. Overall, heterotrophic metabolism directly links microbial metabolic activity to DOM biodegradation; under varying DOM environmental conditions, both community taxonomic composition and metabolic functions show significant differentiation, with different taxa potentially undertaking the degradation of distinct DOM types. Notably, Sphingomonas from the Proteobacteria phylum frequently serves as an indicator microorganism closely associated with DOM, playing a crucial role in DOM utilization [27]. Further research indicates that key species within the community primarily consume readily degradable DOM, partially converting it into refractory compounds. At the same time, other microbial groups may subsequently utilize these relatively inert substrates [28].

Whilst traditional holistic analyses (such as DOC concentration, UV-visible and fluorescence spectroscopy) provide valuable parameters for the average chemical properties of DOM, these methods inherently obscure the vast molecular diversity underpinning these composite metrics. This “black-box” limitation constrains our ability to establish mechanistic links between DOM composition and its environmental reactivity. Advancements in analytical techniques, notably the emergence of Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), have significantly contributed to our enhanced comprehension of the molecular composition of dissolved organic matter (DOM) in soil. As a foundational analytical platform, FT-ICR MS provides mass accuracy below one ppm and can resolve over 10,000 molecular signatures within individual DOM samples [29,30]. This high-resolution molecular mapping elucidates structure-function relationships—from heteroatom stoichiometries to aromaticity indices—that govern DOM persistence mechanisms and biogeochemical interfaces in soil [31]. Such detailed molecular mapping is indispensable for elucidating specific biogeochemical mechanisms—for instance, identifying which molecular subclasses are stabilized, metabolized, or transformed under different conditioning treatments.

To establish a close link between soil microbial communities, functional genes, and DOM molecular components, we have developed a conceptual framework. Extracellular enzymes secreted by soil microbes can target polysaccharides and lignin-like polymers, hydrolyzing them into monomers such as sugars, phenols, and amino acids that are released into the DOM pool. Subsequently, these DOM components are taken up by microbes and enter corresponding metabolic pathways. Functional genes act as ‘mechanistic bridges’ within this process, encoding the enzymes and metabolic pathways required to mediate microbe-DOM interactions. For instance, genes associated with central carbon metabolism (such as glycolysis) and methanogenesis often exhibit positive correlations with specific DOM components. In contrast, genes linked to lignin degradation show inverse trends with these DOM fractions [32]. This microbially driven DOM transformation reshapes the readily available and recalcitrant components of DOM molecules, ultimately influencing carbon cycling.

Despite these advances, the microbial mediation of DOM molecular restructuring through carbon cycling pathways remains poorly understood, particularly regarding feedback mechanisms induced by different organic fertilizations. This study introduces a multi-omics framework integrating FT-ICR MS based molecular characterization with metagenomic profiling to explore: (i) Analyze the effects of different organic materials on the molecular chemical diversity and compositional characteristics of soil DOM, elucidating their regulatory role in DOM transformation potential; (ii) Establish coupling relationships between dynamic changes in microbial functional groups and DOM molecular composition and transformation processes. Based on the distinct properties of organic materials, the following scientific hypotheses are proposed: (i) Compared to crop straw, the application of biochar and manure will enhance DOM’s molecular chemical diversity and transformation potential. (ii) The underlying mechanism may involve the increased nutrient availability promoting microbial activity, which in turn drives the transformation of DOM toward a molecular composition with higher energy value (e.g., lower ΔG).

2. Materials and Methods

2.1. Study Site Description

The experimental site (27.114° N, 114.86° E) within the Zhongke-Ji’an Eco-Environmental Sciences Innovation Base exhibits a monsoon-driven subtropical climate, characterized by a mean annual temperature range of 17.1–18.6 °C and precipitation of 1500–1700 mm with pronounced spring-summer seasonality [33]. The Haplic Anthrosols (FAO) feature a loamy texture (sand 32.7%, silt 33.7%, clay 33.5%) and acidic topsoil (pH 5.7) containing 14.5 g kg⁻¹ TC, 1.5 g kg⁻¹ TN, and 0.4 g kg⁻¹ TP. Situated at 51–135 m a.s.l. with 10–20% slopes, this terraced paddy system supports intensive double-cropping rotations (rice-rice or rice-rapeseed), reflecting typical red soil hill agroecosystem management practices.

2.2. Field Experiment Design

The organic inputs comprised rice straw, rice-straw biochar, and locally sourced livestock/poultry manure. Their total carbon contents (dry weight) were 39.5%, 41.7%, and 40.6%, respectively; the biochar BET specific surface area was 1.3 m²/g⁻¹. All organic materials were applied once at rice sowing as a basal input, with no additional applications during the growing season. Treatments received 0.3 kg C m⁻² (except CK) through organic fertilization. 0.4 m buffer ridges separated experimental plots (13.5 × 7.2 m). All plots received standard fertilization (180 kg N ha⁻¹, 75 kg P₂O₅ ha⁻¹, 150 kg K₂O ha⁻¹) split into three applications: basal, tillering, and panicle initiation stages. Post-application tillage to 20 cm depth ensured homogeneous incorporation of amendments within the plow layer.

2.3. Soil Sampling and Determination of Physicochemical Properties

Soil samples were collected during the rice maturation period (September 2023). Five sampling points were first randomly distributed within each experimental plot. Topsoil from the 0–20 cm layer was collected, and equal quantities of soil samples were thoroughly mixed to form a composite sample. Composite samples from each plot were homogenized and partitioned into triplicate aliquots following debris removal (crop residues, roots, gravels). Aliquot processing protocols included: (i) Air-dried <2 mm fraction for physicochemical characterization. (ii) 4 °C preservation for DOC quantification. (iii) −80 °C cryopreservation for metagenomic profiling. The determination of soil’s physicochemical properties primarily encompasses the following parameters and methodologies: Dissolved organic carbon (DOC) concentration was measured using the ultrapure water extraction method. Fresh soil samples were mixed with water at a mass-to-volume ratio of 1:5 and subjected to shaking, followed by quantitative analysis using a total organic carbon analyser (TOC, Shimadzu Corporation, Kyoto, Japan) [34]. Total carbon (TC) and total nitrogen (TN) contents were determined using an elemental analyser (Vario Macro Cube elemental analyser, Elementar, Frankfurt, Germany) [35]. Total phosphorus (TP) and available phosphorus (AP) content were determined using the molybdenum-antimony colourimetric method [36]. Soil pH was measured using a pH meter (FiveEasy Plus, Mettler-Toledo, Zurich, Switzerland), with a soil-to-water mass-to-volume ratio of 1:2.5 during measurement [37].

2.4. DOM Extraction and FT-ICR-MS Determination

Soil-water extracts (1:5 w/v) were filtered (0.45 μm) and processed through PPL solid-phase extraction cartridges (Agilent, Santa Clara, CA, USA). Methanol-eluted DOM fractions were cryopreserved at −20 °C before ultrahigh-resolution MS analysis. Molecular characterization was performed on a 15.0 T FT-ICR MS (Bruker SolariX, Bremen, Germany) with ESI(−) ionization under optimized parameters: continuous infusion: 120 μL h⁻¹, capillary voltage: −4.0 kV, ion accumulation: 200 ms, mass window: 100–1600 Da, and signal averaging: 300 scans.

Before FT-ICR MS analysis, the ESI source and transfer lines were rinsed with HPLC-grade methanol, followed by injection of a methanol solvent blank to verify the absence of carryover or contaminant peaks in the working m/z range. When necessary, spectra were post-acquisitionally fine-tuned using an abundant homologous series within DOM to ensure mass accuracy within specification. Due to the high analytical cost, each DOM extract was injected only once for FT-ICR MS data acquisition. The injection order was randomized across treatments. Solvent blanks were interspersed throughout the analytical sequence to monitor carryover effects and ensure measurement accuracy during the run.

Data processing was performed in Bruker Compass DataAnalysis (Bruker Daltonics, Bremen, Germany, v4.2). The mass spectrometry data were calibrated using DOM internal standards for known CHO compounds. After calibration, molecular formula matching was performed under the following parameters: a mass range of 100–800, a signal-to-noise ratio (S/N) > 4, and elemental compositions constrained to ¹²C_0–60,¹³C_0–1,¹H_0–120,¹⁶O_0–50,¹⁴N_0–5,³²S_0–2, and ³¹P_0–2. Potential molecular formulas were then determined based on these criteria. If an observed m/z value corresponded to multiple possible molecular formulas, the correct molecular formula was selected using the homolog and minimum heteroatom count rules. The final screening ensured that more than 90% of the molecular formulas were accurate.

2.5. Assessment of Thermodynamic Stability and Transformation Potential of DOM

The aromaticity index (AI), double bond equivalent (DBE), nominal oxidation state of C (NOSC), and Gibbs free energy (∆G) of each DOM molecule were calculated using the established Equations (1)–(4).

$$\mathrm{AI}=(1+\mathrm{C}-\mathrm{O}\text{-}\mathrm{S}\text{-}0.5\mathrm{H})/(\mathrm{C}\text{-}\mathrm{O}\text{-}\mathrm{S}\text{-}\mathrm{N}\text{-}\mathrm{P})$$

(1)

$$\mathrm{DBE}=(1+2\mathrm{C}\text{-}\mathrm{H}+\mathrm{N})/2$$

(2)

$$\mathrm{NOSC}=(4+\mathrm{Z}\text{-}4\mathrm{C}\text{-}\mathrm{H}+3\mathrm{N}+2\mathrm{O}\text{-}5\mathrm{P}+2\mathrm{S})/\mathrm{C}$$

(3)

$$∆\mathrm{G}=60.3 - 28.5 \times \mathrm{NOSC}$$

(4)

where C, H, O, N, P, and S represent the number of C, hydrogen (H), oxygen (O), N, P, and sulfur (S) atoms in the molecular formula, respectively, and the variable Z represents the net charge of the molecule, which is assumed to be neutral in this study [14]. The higher the DBE value, the more stable the compound molecule is. The NOSC value is used to evaluate the oxidation state of the C atoms in the compound. The higher the NOSC value, the stronger the oxidation state and the higher the reactivity of the compound. Also, the higher the ∆G value, the more thermodynamic stability of DOM molecules.

Based on the proportions of C, H, and O elements and the AI values, DOM molecules were classified into ten groups as follows: polycyclic aromatics from combustion processes (PAHs) (AI > 0.66), polyphenols from plant sources (0.50 < AI ≤ 0.66), highly unsaturated and phenolic compounds (HUPs, AI ≤ 0.50, H/C < 1.5), aliphatic (Ali, 1.5 ≤ H/C ≤ 2.0), lipids (0 ≤ O/C ≤ 0.3, 1.5 ≤ H/C ≤ 2.0), carbohydrates (CA, 0.67 ≤ O/C ≤ 1.2, 1.5 ≤ H/C ≤ 2.0), unsaturated hydrocarbons (0 ≤ O/C ≤ 0.1, 0.7 ≤ H/C ≤ 1.0), lignin (0.1 ≤ O/C ≤ 0.67, 0.7 ≤ H/C ≤ 1.5, AI < 0.67), condensed aromatics (O/C ≤ 0.67, 0.2 ≤ H/C ≤ 0.7, AI ≥ 0.67), and tannins (0.67 ≤ O/C ≤ 1.2, 0.5 ≤ H/C ≤ 1.5, AI < 0.67) [32]. Relative abundances of molecular formulas were determined by normalizing the signal intensity of each assigned peak to the sample’s total ion intensity. In addition, using molecular paired mass differences, we constructed the transformation network of DOM molecules and thereby calculated the number of DOM transformations [14].

2.6. DNA Extraction and Metagenomic Sequencing

Genomic DNA from four biological replicates per treatment (4 treatments; 16 samples in total) was used to construct one library per sample. Microbial DNA was extracted from all soil samples using the ALFA-SEQ DNA Library Prep Kit (Magigene, Shenzhen, China) following the manufacturer’s protocol. DNA purity and concentration were assessed using Qubit 4.0 and NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA) on a 1% agarose gel electrophoresis. Metagenomic sequencing was performed on the Illumina platform to generate 150 bp paired-end reads. The sequencing depth of approximately 10 Gb per library. First, paired-end raw reads were subjected to sliding-window quality trimming with fastp (v0.14.1; parameters: -W 4 -M 20). Primers at both read ends were removed using cutadapt, yielding quality-controlled paired-end clean reads. Next, reads were merged based on their overlap using USEARCH (v10; -fastq_mergepairs) with a minimum overlap of 16 bp and up to 5 mismatches permitted within the overlap; sequences failing these criteria were discarded, and the remaining merged sequences were designated as Raw Tags. Finally, fastp (v0.14.1; -W 4 -M 20) was applied again to the Raw Tags for sliding-window quality trimming to obtain the final Clean Tags.

The clean data were then assembled using MEGAHIT (v1.2.9), and open reading frames (ORFs) were predicted for both individual samples and co-assembled scaftigs (≥500 bp) using Prodigal (v2.6.3). Gene abundance in each sample was quantified by mapping the clean data back to the gene catalog using BBMap. Taxonomic and functional annotations were performed using the NCBI non-redundant (NR) database and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.

For microbial network topology, we implemented a dual-threshold framework: (1) Filtering genera below 50% abundance percentile; (2) Constructing interaction networks (Gephi, v0.10.1) using robust correlations (Spearman’s r > 0.6, p < 0.05). Keystone taxa were identified through Z-normalized centrality metrics (degree/closeness), with the top 20 ranked species validated against established ecological network criteria [38].

2.7. Statistical Analysis

One-way analysis of variance (ANOVA) and Duncan’s multiple tests were used to assess the differences in means between experimental treatments. The common and unique aggregation patterns of DOM molecules in different treatments were analyzed using the online tool ImageGP (v2.0) [39]. Redundancy analysis (RDA) was performed to investigate the relationship between soil microbial community composition and environmental variables using the “vegan” package in R (version 4.4.1). We employed the “psych” package to analyze correlations between functional genes, keystone taxa, and DOM composition. Using “plspm” and “rfPermute”, we constructed path models and conducted random forest analyses to determine how soil properties influence microbial, functional genes, DOM composition, and transformations. This two-step approach revealed inter-variable associations and identified critical drivers governing these pathways.

3. Results

3.1. Soil Physicochemical Properties

Organic fertilization differentially modulated soil biogeochemistry: Straw and manure enhanced phosphorus pools (TP: +12.7–19.8%; AP: +56.2–109.6%), while biochar preferentially elevated carbon (TC: +47.3%) and nitrogen (TN: +26.7%) reservoirs (Figure 1a–f). Different from biochar’s neutral pH effect, acidification mitigation was observed under straw and manure treatments. Notably, exogenous carbon inputs universally maintained DOC concentrations across treatments.

3.2. Chemodiversity and Composition of DOM Molecules

Ultrahigh-resolution mass spectrometry delineated treatment-specific DOM profiles, with molecular inventories expanding from 6035 (CK) to 7785–8037 under organic fertilization (Figure 2a). These fertilizations enhanced DOM chemodiversity by 29.0–33.2% while generating 266.1–286.8% more unique molecular signatures compared to control. Compositional analysis revealed three dominant compositions—‘HUPs’, ‘lignins’, and ‘Ali’—exhibiting differential responses: ‘HUPs’ (32.3–47.2%) and ‘lignins’ (29.2–45.1%) enriched, whereas Ali decreased by 52.6–73.5% across amended treatments (Figure 2b).

3.3. DOM Molecular Potential Transformations

Our findings demonstrate that the application of all three organic fertilizers significantly enhanced the transformation potential of DOM, with manure and biochar treatments exhibiting particularly pronounced effects (Figure 3a). Thermodynamic parameter analysis further corroborated this conclusion: compared to the CK, the Gibbs free energy change (ΔG) of DOM in biochar- and manure-amended soils decreased significantly by 3.3% and 6.4% (p < 0.05), respectively, indicating a reduced energy threshold required for DOM transformation (Figure 3b). Additionally, density distribution analysis of the nominal oxidation state of carbon (NOSC) values revealed that the NOSC peaks of DOM molecules in all organic fertilizer treatments were distributed within negative ranges (CK < Straw < Biochar < Manure), with broader negative distributions observed in manure and biochar treatments (Figure 3c,d). This suggests that DOM molecules in these treatments were dominated by reduced-state compounds, which may more readily participate in microbially mediated redox reactions.

3.4. Soil Microbial Diversity and Functional Gene Abundance

The application of organic fertilizers significantly enhanced soil microbial diversity, as indicated by the Shannon index (Figure 4a). Both straw- and manure-amended treatments showed notably higher Shannon index values compared to the control (CK) (p < 0.05), suggesting that organic fertilization promoted a more even and complex microbial community structure. This increase in α-diversity reflects that organic amendments created a more favorable environment for diverse microbial taxa, likely due to improved nutrient availability and habitat heterogeneity.

Cluster analysis of the top 20% most abundant functional genes revealed distinct grouping patterns: biochar- and manure-amended soils clustered together, whereas straw-amended and CK soils formed a separate group (Figure 4b). Specifically, functional genes associated with branched-chain amino acid transport systems (e.g., livK, livM, livF, livH, livG) and fatty acid degradation (ACAT2, ACADM) exhibited higher relative abundances in organic fertilizer treatments than in the control. Furthermore, genes linked to ABC transport systems (ABC.CD.A, ABC.PE.P) and fatty acid synthesis (fabG) were significantly enriched in manure-amended soils compared to other treatments (Figure 4b). In particular, genes associated with branched-chain amino acid transport systems (livK, livM, livF, livH, livG) showed distinct responses among organic amendments. Biochar application resulted in the highest average increase in relative abundance (8.2%), followed by manure (6.7%), whereas straw amendment led to a slight decrease (–1.3%) compared with the control (CK).

For functional genes involved in fatty acid metabolism, the fabG gene, which participates in fatty acid synthesis, showed the greatest enhancement under manure treatment, with its relative abundance increasing by up to 5.6% compared with CK. Similarly, ABC.CD.A and ABC.PE.P, encoding components of ABC transport systems, increased by 5.8% and 14.1%, respectively, under manure amendment. Moreover, the fatty acid degradation–related genes ACAT2 and ACADM were also elevated, showing relative abundance increases of 6.7% and 5.6% under organic fertilizer treatments relative to CK.

Microbial co-occurrence network analysis partitioned the community into three modules, with Module I dominated by Bradyrhizobium and Pseudolabrys (Figure 4c, Supplementary Figure S1a). Redundancy analysis (RDA) further demonstrated clear β-diversity differentiation among treatments (Figure 4d). The first RDA axis (RDA1) explained 78.4% of the total variance in microbial community structure, underscoring strong treatment-driven separation. Available phosphorus (AP), total phosphorus (TP), total carbon (TC), and soil pH were identified as the major environmental drivers shaping community distribution patterns. Together, these results indicate that organic fertilization not only enhanced microbial α-diversity but also amplified β-diversity differences, leading to distinct community compositions under different organic amendment treatments.

3.5. Relationships Among DOM Molecules, Soil Properties and Microbial Communities

Cluster analysis integrating functional gene abundance and microbial taxa correlations revealed that DOM molecular composition were partitioned into two distinct groups along the y-axis (Figure 5a,b): (1) labile composition (e.g., “Ali”, “Lipids”, and “CA”) and (2) recalcitrant composition (e.g., “HUPs”, “Tannins”, and “Lignins”). Functional gene association analysis demonstrated that genes ABC.CD.A, ybbP, prkC, ABC.PE.P, and yadG exhibited negative correlations with readily utilizable composition (r < −0.6, p < 0.05) but strong positive correlations with recalcitrant composition (r > 0.5, p < 0.05). Conversely, the transposase gene and branched-chain amino acid transport system genes (livM, livG, livH, livK) showed positive correlations with readily utilizable composition (r > 0.5, p < 0.05) and negative correlations with recalcitrant composition (r < −0.5, p < 0.05) (Figure 5a). Furthermore, microbial taxa (Afipia, Defluviicoccus, Phenylobacterium, and Bradyrhizobium) were significantly negatively correlated with recalcitrant composition (r < −0.5, p < 0.05) and positively correlated with readily utilizable composition (r > 0.5, p < 0.05) (Figure 5b), suggesting their potential role in steering DOM metabolic preferences via gene expression regulation.

3.6. Drivers of DOM Molecular Transformation

This study integrated random forest modeling and partial least squares path modeling (PLS-PM) for multi-dimensional analysis to elucidate the driving mechanisms underlying the transformation potential of soil DOM molecules. The random forest model revealed significant differences in the relative importance of various factors influencing DOM transformation potential (Figure 6). DOM molecular composition exhibited the most pronounced contributions: both recalcitrant composition (HUPs, Lignins) and labile composition (Ali) showed highly significant effects (p < 0.01). Functional genes, including the ABC transport system gene ABC.CD.A, the branched-chain amino acid transport gene livK, and the gene transposase, also demonstrated significant impacts (p < 0.01). Soil physicochemical properties—TC and DOC—exerted moderate yet significant regulatory effects on DOM transformation (p < 0.05).

PLS-PM analysis (goodness-of-fit, GOF = 0.7) further delineated the pathways: DOM transformation potential was directly driven by its molecular composition (standardized path coefficient, r = 0.91, p < 0.01) and soil physicochemical properties (r = −0.93, p < 0.01). Notably, DOM compositions were significantly regulated by functional genes (r = 0.63, p < 0.01), which indirectly influenced transformation potential through DOM compositional shifts, unveiling a “gene-component-function” cascading regulatory pathway. These findings align with the factor importance rankings from the random forest model, underscoring the pivotal roles of molecular characteristics and functional genes in DOM transformation dynamics (Figure 6).

4. Discussion

4.1. Soil Physicochemical Modulation and Its Impact on DOM Composition

Organic fertilization induced distinct shifts in soil nutrient pools and pH dynamics. Straw and manure amendments significantly increased total phosphorus (TP) by 12.7–19.8% and available phosphorus (AP) by 56.2–109.6%, whereas biochar preferentially enhanced total carbon (TC) by 47.3% and total nitrogen (TN) by 26.7% (Figure 1). These biogeochemical changes provided a foundation for increased DOM chemodiversity, reflected by up to a 33.2% rise in molecular complexity and a 266–287% increase in unique molecular formulas relative to the control. DOM composition is closely associated with carbon (C) and nitrogen (N) utilization, playing a critical role in their mineralization within paddy soils [40,41]. Multivariate analyses demonstrated that soil nutrient and pH changes positively influenced microbial community restructuring (r = 0.90, p < 0.001; Figure 6), consistent with findings that microbial metabolism regulates DOM formation through extracellular enzymatic catalysis and redox shuttle systems [42].

Straw application modulated microbial decomposition pathways, promoting the accumulation of aromatic-rich DOM [43]. Enhanced microbial diversity-especially elevated Shannon indices in straw and manure treatments-and the upregulation of functional genes (e.g., ABC transporters and fatty acid degradation pathways) jointly strengthened microbial-DOM interactions. These metabolic shifts likely facilitated the buildup of recalcitrant DOM, including highly unsaturated phenolics and lignin-like compounds, which increased by 32–47% and correlated with specific microbial taxa and gene markers (Figure 2 and Figure 5). These findings align with the “microbial carbon pump” paradigm, wherein microbial substrate preferences and energy allocation drive the stabilization of complex organic matter [21,44,45].

When applied at equivalent rates, the utilization of biochar and manure has been shown to generally outperform straw incorporation. From a chemical perspective, biochar adsorbs soluble organic matter and neutralizes soil acidity through its porous structure and surface functional groups. The impact of this phenomenon on the enhancement of soil pH and the increase in soluble organic matter concentration is particularly evident in acidic paddy fields [46]. A meta-analysis of paddy fields (1799 datasets) further corroborates the hypothesis that manure application increases available phosphorus (AP) content by 134%, while significantly enhancing phosphatase, urease, and dehydrogenase activities. This finding is directly associated with its superior capacity for chemically–biologically coupled phosphorus supply [47]. Conversely, straw frequently demonstrates delayed nutrient release, a consequence of its elevated lignin content and significant carbon-to-nitrogen ratio. It has been demonstrated that, whilst increasing organic carbon inputs in flooded paddy fields, the subject in question readily stimulates methanogenic microbial communities, thereby elevating CH4 fluxes. Furthermore, the slow rate of substrate degradation and the early nitrogen fixation reduce the utilization efficiency of exogenous nutrients [48]. The results of research into microbial mechanisms indicate that biochar promotes functional microbial proliferation by establishing stable microenvironments [49]. The application of manure has been demonstrated to have a significant effect on soil respiration rates and enzyme activities associated with nitrogen and phosphorus mineralisation. This effect has been observed to be independent of the type of manure applied, whether it be conventional or modified [50]. This divergence can be attributed to the synergistic interaction between biochar and manure in nutrient release kinetics and microbial regulatory pathways.

4.2. Thermodynamic Shifts and Enhanced Transformation Potential of DOM

Shifts in soil physicochemical properties drove microbial succession, upregulating functional gene expression and altering DOM composition. Compared with the CK, the proportion of reductive DOM molecules (characterized by a negative nominal oxidation state of carbon, NOSC) in organically amended soils increased slightly. At the same time, their Gibbs free energy change (ΔG) decreased by 3.3–6.4% (Figure 3). This change suggests that the DOM pool is thermodynamically more favorable for microbial oxidation, providing microbial communities with a higher potential energy output and thereby enhancing the thermodynamic driving force for the microbial transformation of DOM molecules [51]. Notably, Defluviicoccus communities positively correlated with labile DOM but negatively correlated with recalcitrant DOM. The abundance of Defluviicoccus decreased following organic fertilization compared to control conditions (Figure 5, Supplementary Figure S1b), aligning with previous studies, which showed eutrophic microbes preferentially metabolize labile DOM [52,53]. Coupling DOM chemical properties with microbial metabolic strategies potentially explains the reduced ΔG and increased carbon transformation potential observed. Previous studies indicate higher microbial carbon utilization in paddy soils than dry fields, further amplified by organic fertilization through enrichment of reduced DOM and its conversion into refractory compounds, optimizing microbial DOM utilization [54,55].

The maintenance of reduced DOM characteristics appears consistent with adaptive microbial metabolic strategies. Genes encoding ABC transporters (e.g., ABC.CD.A, livK) exhibit a significant positive correlation with refractory dissolved organic matter (DOM) components (e.g., highly unsaturated phenols (HUPs), lignin). The potential mechanism may be as follows: refractory DOM is degraded by other extracellular enzymes to produce small-molecule products, and ABC transporters are responsible for transporting such products into cells; therefore, in environments enriched with refractory DOM, the expression levels of these transporter genes are higher. In contrast, transposase genes were positively associated with labile DOM (e.g., Ali), indicating niche adaptation via the selective metabolism of readily degradable substrates by mobile genetic elements (Figure 5). In agreement with earlier arguments that identified ABC transporters as selective transmembrane hubs responsive to DOM availability, our results collectively support a model in which organic amendments may facilitate the formation of hierarchical regulatory networks, enhancing the expression of functional genes linked to refractory DOM transformation [56,57,58].

DOM is commonly regarded as the most readily decomposable fraction of soil organic matter; however, review studies indicate that its lignin-derived aromatic compounds may constitute the most stable components. Carbohydrates and nitrogen-rich compounds synthesized by microorganisms during DOM transformation also exhibit notable stability and are abundant in mineral subsoil DOM. These stable components achieve significant stabilization effects through adsorption onto soil mineral surfaces or coprecipitation with aluminum minerals (particularly characteristic of aromatic components). Concurrently, this study estimated the potential contribution of DOM to the total carbon pool in mineral soils. Results indicate that in mineral soils, DOM-derived soil carbon stocks range from 20 to 55 Mg ha⁻¹, accounting for 19% to 50% of the total soil carbon. This figure underscores DOM’s pivotal role in stabilizing organic matter accumulation [59]. Similarly to our findings, organic fertilization increased the total DOM molecular transformations by 10.4–14.1%, with polyphenols and condensed aromatic compounds exhibiting net positive transformations. This suggests they may be converted from lignin-like and aliphatic precursors, components of these recalcitrant compounds potentially stored within soil particulate organic carbon, to achieve soil carbon sequestration [60,61].

Microbial residues may serve as a source of soil-resistant soluble organic matter (DOM), with their formation and accumulation during DOM transformation contributing to enhanced soil carbon sequestration. Research indicates that the pathway of microbial death determines the chemical composition of the residue and its subsequent fate. During the transformation of microorganisms into residues, their internal nutrient content and readily degradable components significantly decrease. This highly efficient internal material recycling mechanism provides a crucial pathway for soil organic carbon sequestration, enabling stable carbon fixation while minimizing nitrogen loss [62]. In studies of plant rhizosphere or soil microbial networks, Bradyrhizobium has been identified as a key member of the microbial community. For instance, research on oilseed rape microbial communities has indicated its pivotal role within the rhizosphere soil microbial network. Furthermore, the presence of Bradyrhizobium may enhance nitrogen influx and transformation within the rhizosphere ecosystem, thereby influencing plant growth, soil nitrogen availability, and microbial community structure. Through network analysis, we identified Bradyrhizobium as a hub microorganism. This, combined with its known ecological role from literature [63], leads us to hypothesize that it may enhance nitrogen input via fixation and rhizosphere interactions, thereby potentially contributing to the stability of the microbial community structure and function.

The incorporation of crop residues, biochar, and manure into paddy soils not only enhances carbon sequestration by boosting the transformation potential of soil DOM, but also significantly influences methane (CH₄) and carbon dioxide (CO₂) emission processes. Studies indicate that, compared to control treatments, straw incorporation increases CO₂ and CH₄ emissions while reducing nitrous oxide (N₂O) emissions. Based on changes in the cumulative CO₂ to CH₄ emission flux ratio, straw incorporation promotes greater carbon release in the form of CH₄ [64]. In contrast, biochar application reduced CH₄ emissions by 51.1% and decreased CO₂ emissions from flooded paddy soils [65]. Manure application typically enhances CH₄ and CO₂ emissions, but when composted, it effectively suppresses soil greenhouse gas emissions [66].

4.3. Organic Fertilization Increase the DOM Transformation Potential via Microbial Functional Gene

Numerous studies have shown that organic fertilization enhances DOM transformation potential by lowering the Gibbs free energy (ΔG) required for microbial oxidation [14,20]. Building on these findings, our results indicate that manure and biochar were substantially more effective than straw in promoting DOM transformation and reducing ΔG (Figure 3). Furthermore, manure and biochar treatments significantly increased the abundance of functional genes encoding ABC transporters, branched-chain amino acid transporters, and fatty acid synthesis enzymes, relative to straw and control treatments. These results suggest that manure and biochar selectively enrich microbial metabolic pathways, modifying soil DOM composition.

This study further revealed that, compared with the control group and straw treatment, both biochar and manure treatments significantly enriched functional genes encoding ABC transporters (such as ABC.CD.A and livM). ABC transporters comprise a broad superfamily of integral membrane proteins that facilitate the transmembrane transport of diverse substrates using energy derived from ATP hydrolysis. This enrichment suggests that microorganisms may enhance the uptake of substrates, such as nitrogen-containing organic matter, to promote carbon utilization and metabolism [56,67]. Moreover, manure treatment specifically enriched key genes involved in fatty acid degradation and synthesis pathways (such as ACAT2 and fabG). Fatty acid metabolism by soil microorganisms constitutes a pivotal component of the carbon cycle, directly influencing the transformation and stability of organic carbon. These findings suggest that manure further promotes the sequestration and transformation of soil organic carbon by regulating the metabolic potential of microbial fatty acid metabolism [68].

Correlation analyses revealed significant associations between specific functional genes and DOM composition. Manure and biochar treatments increased the relative abundance of refractory DOM components—such as highly unsaturated phenolics (HUPs), lignins, and tannins—through elevated gene expression, while simultaneously reducing labile fractions, including aliphatics and lipids (Figure 2 and Figure 4). Random forest analysis underscored the critical role of functional genes in shaping DOM transformation potential (p < 0.05), primarily via positive regulation of DOM composition (p < 0.01; Figure 6). Pathway-level analyses further indicated that microbial functional genes directly mediate DOM transformations. For example, straw application stimulated the production of free radical site redox enzymes in paddy soils, thereby enhancing DOM transformation processes [69,70]. In addition, previous studies demonstrated that microbial metabolism of short-chain fatty acids leads to the generation of refractory organic compounds, with transformation efficiency declining as carbon chain length increases, reinforcing that DOM composition fundamentally governs its transformation potential [71,72].

Our findings demonstrate that manure and biochar applications enhance microbial metabolic functions—such as ABC transporter activity and branched-chain amino acid transport—and improve microbial substrate utilization efficiency. We therefore propose that the observed increase in DOM transformation potential under these treatments could be driven by shifts in microbial community structure and upregulation of functional gene expression [16,73]. Soil conditioners (straw, biochar, and manure) promote the formation of stable carbon through continuous biogeochemical processes. This process commences with the input of exogenous carbon, wherein the dissolved organic matter (DOM) serves as a readily accessible carbon source for microorganisms. Differences in DOM composition drive the differentiation of microbial community structure and function, enriching key functional genes, such as ABC transporters and those involved in fatty acid metabolism. This, in turn, regulates the molecular transformation of DOM and the direction of carbon cycling. Ultimately, the addition of soil amendments enhances DOM transformation potential, promoting the formation and accumulation of recalcitrant components. Due to differences in chemical composition and availability, straw, biochar, and manure exhibit distinct efficiencies and stability in carbon sequestration.

Future research should aim to clarify the mechanistic links between microbial community composition and gene function to deepen understanding of DOM transformation processes. Such insights will inform more targeted organic fertilization strategies and contribute to improved soil fertility management. This study has certain limitations. We primarily analyzed the functional gene profiles of microbial communities, but did not concurrently measure associated enzyme activities, soil greenhouse gas fluxes (such as CH₄ and CO₂), or gene expression (mRNA) levels. These data are crucial for directly linking microbial metabolic potential to their actual functional responses in soil, providing more causal evidence for carbon transformation mechanisms in this process. Future studies integrating metatranscriptomics, proteomics, and in situ gas monitoring could provide a more comprehensive understanding of the complete pathway from genetic potential to ecosystem function, thereby deepening our understanding of DOM transformation and carbon cycle dynamics.

5. Conclusions

Organic fertilization alters soil physicochemical properties and microbial metabolic activity, exerting a significant influence on DOM composition and its transformation potential. Straw and manure amendments primarily enhanced soil phosphorus availability and microbial diversity, facilitating the accumulation of chemically complex and recalcitrant DOM components such as highly unsaturated phenolics (HUPs) and lignins. In contrast, biochar and manure treatments preferentially increased total carbon and nitrogen pools, promoting a shift in DOM toward more thermodynamically favorable states by lowering the Gibbs free energy required for microbial processing. Functional gene analyses revealed microbial metabolic strategies underpinning these changes, highlighting the involvement of ABC transporters and fatty acid metabolism genes in DOM stabilization and utilization. Further research is needed to unravel the molecular linkages between microbial gene expression and DOM transformation pathways to better understand the microbial mechanisms driving long-term soil carbon sequestration. From a practical perspective, manure and biochar applications represent promising strategies for enhancing DOM stability, optimizing microbial function, and advancing sustainable agricultural practices through improved nutrient cycling and carbon sequestration efficiency.