Reactive Oxygen Species Formation in Paddy Soils under Alternating Redox Conditions: An Iron-Microbe Synergy Driving Mechanism and Environmental Impacts Authors: Zhiwei Sun, Xinxin Wang, Xinyue Wu, Xinran Zhang, Xinya Li, Ru Wang, Xiaomeng Wang, Guanyu Zheng Published: 2026 ABSTRACT Alternating wetting and drying cycles in paddy soils drive active biogeochemical processes and profoundly influence pollutant migration and transformation. This study systematically investigates the formation and mechanisms of reactive oxygen species (ROS) under anaerobic–aerobic alternation in five typical paddy soils, as well as their implications for the attenuation of tetracycline (TC) and associated antibiotic resistance genes. The results demonstrate that ROS generation primarily arises from microbially mediated Fe­(II) accumulation during flooding, followed by Fenton-like oxidation upon reoxygenation. The cumulative hydroxyl radical (•OH) yields during the 12 h aerobic phase following 30 days of flooding followed the order: MaAnShan (1175.7 μmol/kg) > QiongHai (794.8 μmol/kg) > SuiHua (711.8 μmol/kg) > ChenZhou (391.6 μmol/kg) > GuiLin (297.0 μmol/kg). Iron-reducing bacteria, including Anaeromyxobacter, Desulfosporosinus, and Geomonas, were identified as the primary drivers of Fe­(III) reduction. The variation in ROS production potential among soils was primarily governed by poorly crystalline iron content and indigenous iron-reducing bacteria, with alkaline pH and high organic matter content serving as key inhibitory factors. The ROS generated under fluctuating redox conditions contributed substantially to TC degradation (24.8–66.9%) and tetA gene attenuation (60.4–79.2%). Collectively, this study elucidates the biogeochemical mechanisms underlying ROS formation in paddy soils during drying–wetting cycles and their role in pollutant attenuation, offering a theoretical framework for advancing our understanding of contaminant dynamics in agricultural wetlands and for guiding the development of targeted pollution control strategies. Introduction Paddy field ecosystems, as one of the most widespread agricultural wetlands worldwide, frequently undergo alternating cycles of flooding and drainage. These wetting-drying cycles serve as primary driving force for soil biogeochemical processes. They also exert profound influences on the environmental migration and transformation of contaminants. Emerging pollutants, particularly antibiotics and antibiotic resistance genes (ARGs), have recently received increasing attention due to their persistence and mobility in soil environments. Numerous studies have demonstrated that the long-term application of livestock manure containing antibiotics can significantly elevate the concentrations of tetracycline antibiotics in paddy soils, with reported levels ranging from 1.1 to 24.66 μg/g. , These antibiotics can accumulate in rice grains, with an average concentration of 0.21 μg/g detected, thereby introducing potential health risks through food chain transfer. , Residual antibiotics in soils, even at sub-inhibitory concentrations, can impose continuous selective pressure on microbial communities. This selection promotes the proliferation and dissemination of ARGs and associated mobile genetic elements (MGEs), further amplifying environmental and public health risks. Paddy soils act as a major environmental sink for tetracycline antibiotics. However, the attenuation behaviors of antibiotics and ARGs under wetting-drying cycles, as well as their underlying mechanisms, remain poorly understood. Previous studies have predominantly emphasized microbially mediated degradation pathways, , whereas recent evidence has indicated that wetting-drying alternation can induce the formation of highly reactive and strongly oxidizing species. , These findings provide new perspectives for elucidating the abiotic degradation mechanisms of antibiotics under fluctuating redox conditions. Soil serves as both a critical source and a primary reservoir for various highly active reactive oxygen species (ROS). The abundance of redox-active substances, coupled with frequent fluctuations in oxygen and moisture, drives complex electron transfer and redox reactions at biotic–abiotic interfaces, leading to substantial ROS generation. Recent studies have increasingly recognized the widespread presence of ROS in paddy soils, which is closely linked to periodic shifts in soil redox potential induced by frequent flooding-drainage regimes during rice cultivation. , Chen and colleagues systematically revealed the ubiquitous generation pattern of hydroxyl radical (•OH) during alternating anaerobic–aerobic conditions in paddy soils, identifying the soil–water interface as a major hotspot for hydrogen peroxide (H2O2) and •OH production. Subsequent research by Huang et al. demonstrated that long-term flooding (1 year) could enhance the •OH production potential in soil from 227–271 μmol/kg to 499–902 μmol/kg, with notably higher ROS activity observed in the surface layer. Furthermore, Chen et al. elucidated that under oxidizing conditions, paddy soils can generate •OH (yielding 271.5–581.2 μmol/kg) through transformation of reactive iron species, highlighting a significant abiotic formation pathway. Due to their high reactivity, ROS play critical roles in various soil processes, including carbon turnover, elemental cycling, microbial succession, and pollutant degradation. ,, Wang et al. reported that paddy soils under oxic conditions could generate •OH radicals at concentrations ranging from 40.3 to 1061.4 μmol/kg, which contributed to approximately 25% degradation of pesticides such as imidacloprid. Chen et al. further demonstrated that •OH radicals act as the dominant reactive species responsible for the degradation of polycyclic aromatic hydrocarbons (PAHs). In anaerobic paddy soils subjected to 12 h of aerobic oxidation, the contribution of •OH to the degradation of naphthalene (NAP), phenanthrene (PHE), and pyrene (PYR) reached 36.1, 13.2, and 3.7%, respectively. In another study, the application of Fe–Mn-modified biochar was shown to enhance •OH generation by activating a Fenton-like reaction, thereby driving the efficient degradation of the antibiotic sulfamonomethoxine (SMM), with a notable contribution of 79.5–83.3% to the degradation process. The substantial generation of ROS during alternating wetting-drying cycle in paddy soils involves multiple chemical and biological pathways, including photochemical reactions in surface layers, microbially mediated redox processes, humification-related transformation of organic matter, and redox reactions of active mineral components such as Fe and Mn oxides. ,, Among these, the Fe­(III)/Fe­(II) cycling driven by frequent fluctuations in soil redox conditions represents the most important abiotic mechanism for ROS production. , Under flooded conditions, a rapid decline in soil Eh promoted the accumulation of Fe­(II) to concentration of μM-mM levels. , Subsequent brief oxygenation events, triggered by oxygen intrusion, facilitate the reaction of Fe­(II) with O2 or superoxide radicals (O2 •–) via the Haber-Weiss mechanism, generating H2O2, which is further converted to •OH through homogeneous or heterogeneous Fenton reactions. In addition to chemical pathways, biological processes play a crucial role in amplifying ROS generation. On one hand, Fe-reducing microorganisms can dominate the production of Fe­(II) (particularly the poorly crystalline Fe (oxyhydr)­oxide-associated fraction, hereafter referred to as reactive Fe­(II)) during the flooding phase, thereby building a “precursor pool” for subsequent ROS formation. On the other hand, microbially derived metabolites (e.g., water-soluble phenols compounds) are recognized for their redox reactivity, acting as reductants during reoxygenation to promote Fe­(III)-to-Fe­(II) transformation and sustain Fenton-like reactions. The full nature of their involvement in electron transfer processes, however, is yet to be clarified. , Furthermore, microbial metabolism and cell lysis can directly release H2O2 and secrete related redox enzymes, that jointly contribute to the biogenic sources of ROS. Importantly, these biological and chemical processes do not operate independently; instead, they are mutually reinforced through cycling of Fe, phenolics, and other intermediates, forming a “bio-chemical coupled” ROS amplification network. Notably, the intensity and efficiency of this coupled network are highly dependent on the intrinsic properties of soils, such as the speciation and abundance of initial iron oxides, the composition and quantity of organic matter, pH conditions, and the native microbial community structure. Accordingly, this study selects five representative paddy soils and conducts a controlled flooding-drainage simulation experiment to systematically elucidate the biotic–abiotic coupled mechanisms driving ROS production and its associated environmental effects. The specific objectives of this work were (1) to quantify the ROS generation potential across different paddy soils and identify their dominant driving factors; (2) to unravel the microbially and chemically coupled mechanisms governing ROS generation; and (3) to elucidate the attenuation behaviors of antibiotics and ARGs during the flooding-drainage cycle in paddy soils. The findings will not only advance the mechanistic understanding of biogeochemical processes in paddy soils but, more importantly, provide critical theoretical guidance for assessing and predicting the natural attenuation efficiency of chemical and biological pollutants under wetting-drying alternation. Materials and Methods Soil Collection and Preparation Surface paddy soils (0–20 cm) were collected from five representative rice-growing regions across China, including Chenzhou, Hunan Province (CZ); Suihua, Heilongjiang Province (SH); Qionghai, Hainan Province (QH); Guilin, Guangxi Zhuang Autonomous Region (GL); and Ma’anshan, Anhui Province (MA). After sampling, all soils were air-dried indoors, and visible plant residues and gravel were removed manually. The soils were then ground and passed through a 0.15 mm sieve and stored in the dark until further use. The basic physicochemical properties of the five paddy soils are summarized in Table . 1Basic Physicochemical Properties of the Tested Paddy Soils Chenzhou (CZ)Suihua (SH)Qionghai (QH)Guilin (GL)Ma’anshan (MA) latitude25°27′36″N46°44′24″N19°3′N24°27′36″N31°33′Nsampling sitelongitude113°1′12″E127°45′36″E110°12′E111°17′24″E118°34′48″EpH 7.87.15.37.85.6SOM (g/kg) 37.168.028.342.733.4SOC (g/kg) 21.539.416.424.819.4CEC (cmol+/kg) 19.9281415.719.9total Fe (g/kg) 34.134.419.834.421.4soil texturesand (%)42.840.862.148.442.6 silt (%)29.13117.926.632.3 clay (%)28.128.2202525.1 Basic Physicochemical Properties of the Tested Paddy Soils Assessment of ROS Production during Alternating Redox Condition A separate experiment was conducted to investigate ROS generation patterns of five paddy soils under different flooding durations (0–30 days). For each treatment, 20 g of air-dried soil was mixed with 50 mL of deionized water in 100 mL serum bottles. After purge with N2 for 30 min, the bottles were sealed and incubated at 28°C for continuous flooding. Destructive sampling was performed on Day 0, 5, 10, 15, 20, and 30. At each sampling, 5 mL of homogenized slurry was transferred to a 50 mL Erlenmeyer flask within an anaerobic chamber. The samples were then shaken at 180 rpm for 12 h aerobic oxidation at room temperature. The cumulative production of •OH, H2O2, and O2 •– was monitored using specific chemical probe methods. Specifically, •OH was quantified using benzoic acid (BA) as a trapping agent, O2 •– was determined by the XTT reduction method, and H2O2 was measured using the potassium titanium oxalate colorimetric assay (for details, see Text S1). Experiment on TC and ARGs Attenuation under the Anaerobic–Aerobic Cycle The five paddy soils were spiked with a TC solution to an initial concentration of 10 mg/kg. The flooding-drainage experiment was conducted in 100 mL amber serum bottles, each containing 20 g of contaminated soil and 50 mL of deionized water (2.5:1 water-to-soil ratio) to form a homogeneous slurry. After purging with N2 gas for 30 min to establish anaerobic conditions, the bottles were sealed and incubated at 28°C for 15 days. Following this anaerobic period, the systems were exposed to aerobic conditions for 12 h. Sampling was conducted at six time points: start (ay 0) and end (ay 15) of the anaerobic phase, and 1, 2, 6, and 12 h during the aerobic phase, for analyzing TC concentration and tetA gene abundance. The relative contribution of each ROS to the attenuation of TC and tetA was determined through scavenger quenching experiments, in which isopropanol, catalase, and nitroblue tetrazolium were added after the 15 day anaerobic phase to quench •OH, H2O2, and O2 •–, respectively. Detailed protocols and calculation are described in Text S2. Chemical Analyses The TC concentration was determined using high-performance liquid chromatography (HPLC). Dissolved TC was separated by centrifugation, and the sorbed TC in soils was extracted with a mixed acetonitrile-EDTA-and Mcllvaine buffer solution. Soil DNA was extracted with a soil genomic DNA extraction kit (Solarbio, China), and the tetA and 16S rRNA genes were quantified by quantitative PCR (qPCR). The specific experimental procedures and instrumental methods are provided in Text S3. The basic physical and chemical properties of the soils, including pH, Eh, dissolved organic carbon (DOC), dissolved Fe­(II) and Fe­(III), water-soluble phenolic substances, soil urease, dehydrogenase, and catalase activities, were analyzed according to the methods provided in Text S4. The contents of different Fe forms in the solid phase were determined by sequential extraction of 1 mL of suspension samples with 0.5 M HCl and 5.0 M HCl, followed by colorimetric analysis. In this study, the 0.5 M HCl-extractable Fe fraction (primarily targeting poorly crystalline Fe (oxyhydr)­oxides and surface-adsorbed Fe species) is operationally defined as the reactive Fe pool that serves as the precursor for ROS generation, consistent with previous studies. , Microbial Community Analysis Microbial community analysis was performed on soil samples collected after 0, 5, 15, and 30 days of flooding. Total DNA was extracted using the PowerSoil DNA isolation kit (MP Biomedicals, USA). Bacterial abundance was quantified by qPCR targeting the 16S rRNA gene, and community composition was analyzed throughout high-throughput sequencing of the 16S rRNA gene V3–V4 region on an Illumina MiSeq platform (Miseq PE300, Shanghai Meiji Biomedicine Technology Co., Ltd.). Additionally, metagenomic sequencing was conducted specifically on Day 30 samples. Detailed methodological procedures are provided in Text S5. Statistical Analysis All experimental data are expressed as mean ± standard deviation. Pearson correlation analysis and linear regression were used to examine the statistical significance of variable relationships, and piecewise structural equation modeling was applied to identify the key determinants and potential underlying causal pathways responsible for ROS generation. Additional details are provided in Text S6. Results Divergent ROS Production Capacity in Five Paddy Soils under Redox Fluctuation Electron paramagnetic resonance (EPR) analysis of the representative CZ soil confirmed the generation of •OH and O2 •– during anaerobic–aerobic transition (Figure S1). The ROS production characteristics of different paddy soils under alternating redox conditions were evaluated by measuring the cumulative generation of three ROS species during a 12 h aerobic phase after 30 days of flooding (Figure ). The term “cumulative production” refers to the net amount of a specific ROS species detected during the 12 h aerobic phase, representing the total accumulated concentration that had reached an equilibrium plateau under the given experimental conditions. Kinetic monitoring confirmed that ROS reached a steady–state equilibrium within 12 h oxygenation period (Figure S2), with the most rapid generation occurring within the first few hours followed by stabilization. The •OH concentration generally exhibited a rapid initial increase followed by stabilization with prolonged flooding, reaching its maximum after 15 days of flooding (Figure a). MA soil showed the highest •OH production, with a peak concentration of 1244.1 μmol/kg, whereas CZ and GL soils exhibited the lowest levels. Overall, the •OH production capacity of the five paddy soils followed the order MA > QH > SH > CZ > GL, with corresponding cumulative yields (measured after the 12 h aerobic phase following 30 days of flooding) of 1175.7, 794.8, 711.8, 391.6, and 297.0 μmol/kg, respectively. Cumulative production of •OH (a), H2O2 (b), and O2 •– (c) during the 12h aerobic phase in five paddy soils (CZ, SH, QH, GL, and MA) after different flooding incubations. In contrast to •OH, H2O2 accumulation in the five paddy soils exhibited distinct behaviors under alternating flooding-drainage conditions (Figure b). Detectable H2O2 was observed after 12 h of aerobic exposure even at the initial flooding stage (0 day), with SH soil showing the highest level (38.7 μmol/kg). MA and QH soils exhibited similar concentrations (21.7 μmol/kg), while CZ and GL soils showed minimal accumulation (0.9–2.7 μmol/kg). After 5 days of flooding, H2O2 concentrations increased by approximately 8.6–11.2 μmol/kg across all soils. However, further flooding led to a slight decline in H2O2 levels, which then stabilized after 10 days of flooding. Under alternating flooding-drainage condition, O2 •– production increased rapidly during early flooding period (5 days) (Figure c). Cumulative O2 •– concentrations increased from an initial level of approximately 0.8–8.6 μmol/kg (Day 0) to 52.7 μmol/kg (MA), 44.4 μmol/kg (SH), 38.1 μmol/kg (CZ), 32.8 μmol/kg (QH), and 16.9 μmol/kg (GL), respectively. Prolonged flooding beyond this period did not lead to further significant accumulation. Overall, MA soil consistently generated the highest O2 •– levels, GL soil the lowest, and SH, CZ, and QH soils maintained comparable production capacities. Responses of TC and tetA to Alternating Anaerobic–Aerobic Condition Attenuation of TC and tetA Gene Under anaerobic–aerobic alternating conditions, the degradation behaviors of TC and the tetA gene varied significantly among the five paddy soils (Figures a and S3). After 15 days of anaerobic incubation, followed by 12 h of aerobic exposure, notable changes in TC concentrations were observed across all soils. At the end of the anaerobic stage, TC concentrations decreased to varying extents, with the SH soil, characterized by a higher organic matter content, showing the most pronounced decline (50%), while the other four soils exhibited comparable reductions (27–32%). Upon transition to aerobic conditions, TC degradation accelerated rapidly within the first 2 h. MA soil exhibited the highest decrease, with the TC concentration dropping from 7.1 to 2.3 mg/kg (66.9% degradation), followed by QH soil (from 7.0 to 3.0 mg/kg, 56.8% reduction) and SH soil (from 5.2 to 3.3 mg/kg, 37.1% degradation). In contrast, CZ and GL soils exhibited only modest declines, decreasing from 6.9 to 4.9 mg/kg (28.2%) and from 7.2 to 5.5 mg/kg (24.8%), respectively. (a) Dynamics of TC and tetA gene under alternating redox condition (15 day anaerobic/12 h aerobic) in five paddy soils. (b) Relative contributions of specific ROS species (•OH, O2 •–, and H2O2) to the oxidative attenuation of TC and tetA. Regarding antibiotic resistance genes, the tetA gene was significantly enriched in all five soils after 15 days of flooding (Figures S3b and S4). Although its initial background abundance was low, addition of 10 mg/kg TC followed by 15 day anaerobic incubation triggered substantial proliferation. The SH soil displayed the highest tetA/16S rRNA ratio (3.4 × 10–5) (Figure S3b), approximately twice that of other soils, implying that its higher organic matter content may facilitate ARG amplification under anaerobic conditions. Upon aerobic transition, tetA abundance declined rapidly, decreasing by ∼50% within the first 2 h (Figure a). By the end of the aerobic phase, tetA levels were reduced by 60.4–79.2% relative to anaerobic peaks, with the greatest decreases in MA (79.2%) and GL (76.8%) soils. Despite the sharp rise in SH soil during the anaerobic stage, it also exhibited a pronounced decline under aerobic conditions (75.9%). Contribution of ROS to the Oxidative Removal ROS quenching experiments were performed to clarify the specific roles of individual ROS in TC degradation and tetA attenuation. As shown in Figure b and S3, the addition of specific scavengers significantly suppressed both TC removal and the decline in tetA abundance. Notably, •OH was identified as the dominant reactive species, accounting for 35.4–50.3% of TC removal and 54.5–62.1% of tetA attenuation, while O2 •– and H2O2 also made appreciable contributions. Collectively, these findings demonstrate that multiple ROS generated during redox oscillation synergistically drive the elimination of antibiotics and resistance genes, with •OH serving as the principal oxidant. Physicochemical Drivers of ROS Generation during Redox Cycles During the 30 day flooding period, key physicochemical indicators of the five paddy soils were systematically monitored (detailed data in Text S7). The main results were summarized as follows: all soils remained under strongly reducing conditions (Figure S5a). Initially acidic QH and MA soils exhibited a moderate pH increase to near-neutral values (pH 6.7–7.2) (Figure S5b). The content of 0.5 M HCl-extractable Fe decreased in the order SH > MA > QH > CZ > GL, and the Fe­(II) fraction increased progressively with flooding duration, reaching a maximum at approximately Day 15 across all five soils, before stabilizing or declining slightly thereafter (Figures a and S6). WSP compounds followed trends similar to those of dissolved organic carbon (DOC), with significantly higher levels in MA and SH soils (Figures b and S7a). Dissolved Fe­(II) concentrations were higher in QH and MA soils and increased progressively with flooding duration (Figure S7b). Soil DHA was significantly activated upon flooding, with activities across soils following QH > MA > SH > CZ > GL (Figure c). In contrast, catalase (CAT) activity was highest in SH soil but decreased significantly after flooding, whereas urease activity (UE) remained relatively stable throughout the flooding process (Figure S8). Temporal changes in 0.5 M HCl-extractable Fe­(II) (a), WSP compounds (b), and DHA (c) in five paddy soils during 30 days of anaerobic incubation. Linear regressions between cumulative •OH production and 0.5 M HCl-extractable Fe­(II) (d), WSP compounds (e), DHA (f), ROS (H2O2 and O2 •–) (g), and pH (h). Linear regression between the consumed amount of 0.5 M HCl-extractable Fe­(II) and cumulative yield of •OH (i). To further examine the factors associated with ROS generation, correlations between ROS production and soil physicochemical indicators were first evaluated using Pearson correlation analysis, and key relationships were subsequently verified using linear regression (Figures d–i and S9). ROS production showed the strongest positive correlation with iron fractions, particularly those extracted by 0.5 M HCl, followed by significant correlations with dissolved organic carbon, WSP compounds, and specific soil enzyme activities (Figure S9). In contrast, soil pH and Eh were negatively correlated with ROS generation. Specifically, •OH production displayed a strong positive correlation with 0.5 M HCl-extractable Fe­(II) (p < 0.001; Figure d) and was also markedly linked to WSP compounds and DHA (both p < 0.001; Figure e,f). Further significant positive associations were observed with H2O2 (p < 0.01; Figure g) and O2 •– (p < 0.001), whereas a distinct negative correlation emerged with pH (p < 0.05; Figure h). H2O2 generation was most closely associated with 0.5 M HCl-extractable Fe (r = 0.85; Figure S9) and also positively correlated with WSP compounds and UE (r = 0.67). Similarly, O2 •– production showed the strongest link to 0.5 M HCl-extractable Fe­(II) (r = 0.79), with positive correlations to WSP compounds (r = 0.69) and DHA activity (r = 0.66), but a significant negative correlation with CAT (r = −0.42). Notably, both WSP compounds and DHA were positively correlated with 0.5 M HCl-extractable Fe­(II) content, implying their synergistic role in Fe­(II) generation and transformation during redox cycling. To further verify the critical role of iron species in •OH production, an anaerobic–aerobic alternation experiment was conducted using CZ soil subjected to 15 day flooding followed by 12 h aerobic oxidation. As shown in Figures i and S10, a highly significant positive correlation was observed between the consumption of 0.5 M HCl-extractable Fe­(II) and the •OH cumulative production (p < 0.001; Figure i). This relationship was stronger than for other iron species, confirming that weakly crystalline Fe­(II) dominated •OH generation during oxidation. Microbial Taxa Associated with Anaerobic Fe­(II) Precursor Formation Microbial Diversity and Community Structure High-throughput sequencing of the 16S rRNA gene was performed to analyze the microbial communities of five paddy soils at different time points (Day 0, 5, 15, and 30) during 30 days of anaerobic incubation. A total of 3,175,752 high-quality bacterial sequences were obtained from 60 samples, with an average read length of 376 bp after quality control. The α-diversity indices (Sobs, Shannon, and Chao) exhibited generally consistent dynamic fluctuations throughout the flooding period (Figure S11). Among the soils, GL exhibited the highest microbial α-diversity, whereas SH had the lowest. Non-metric multidimensional scaling (NMDS) further demonstrated distinct clustering of microbial communities among different soil samples, with significant inter-group differences (Adonis test, R 2 = 0.89, p = 0.001, Stress = 0.13) (Figure S12, Table S1). At the phylum level, the original soils were dominated by Bacillota, Chloroflexota, Pseudomonadota, and Actinomycetota, which collectively accounted for over 70% of the total community (Figure S13). After flooding, the relative abundance of Bacillota decreased markedly by approximately 17%, while Thermodesulfobacteriota, Planctomycetota, and Myxococcota showed substantial enrichment, reaching 2–5 times their initial levels. These results indicated that prolonged flooding significantly altered the microbial community composition and enhanced the prevalence of anaerobic and facultative anaerobic taxa, reflecting a shift in the soil metabolic potential under reducing conditions. Specific Distribution of Key Microbial Taxa in Paddy Soils Based on the microbial community analysis, a multi-level LEfSe discriminant analysis (LDA ≥ 2) was conducted to identify differential microbial taxa across the five paddy soils and four incubation stages, resulting in 252 significantly enriched genera (Figure S14). To further resolve the functional contributions of these taxa to anaerobic metabolite generation, we quantified the absolute abundances of microbial groups associated with Fe­(III) reduction, WSP compounds-linked metabolism, and DHA throughout the 30 day flooding period (Figure a–c). The estimated absolute abundance of each taxon was calculated by multiplying its relative abundance derived from 16S rRNA gene amplicon sequencing by the total bacterial 16S rRNA gene copy number quantified by qPCR for the same sample. Abundance of iron-reducing bacteria (a), WSP compounds-linked taxa (b), and DHA-associated taxa (c) in five paddy soils during 30 days of anaerobic incubation (calculated as 16S rRNA gene × relative abundance). Linear regressions of 0.5 M HCl-extractable Fe­(II) vs. Iron-reducing bacterial abundance (d); WSP compounds vs. WSP compounds-linked taxa abundance (e); DHA vs. DHA-associated taxa abundance (f). For Fe­(III)-reducing taxa, Clostridium and Geomonas were predominant. The estimated absolute abundance of Geomonas showed the highest cumulative in SH soil, reaching 1.07 × 109 16S rRNA gene copies g/dry soil at Day 5 and remaining elevated, whereas CZ soil consistently displayed minimal levels (< 2.0 × 107 16S rRNA gene copies g/dry soil). Similarly, substantial enrichment of Clostridium and Geomonas were detected in MA soil, increasing by 8.2- and 21.1-fold by Day 15, respectively. Fe­(III)-reducing abundances generally peaked between Days 5 and 15 before modest declines, except in MA and SH soils where Geomonas persisted. About WSP compounds-linked taxa, primarily Clostridium, Bacillus, and Lysinibacillus, were most abundant in SH and MA soils. The estimated absolute abundance of Lysinibacillus reached 1.2 × 107 16S rRNA gene copies g/dry soil in SH at day 15 and 8.0 × 106 16S rRNA gene copies g/dry soil in MA at Day 15, representing 3- to 5-fold higher levels than in QH, CZ, and GL soils, while CZ soil consistently harbored the lowest abundances. After Day 15, WSP compounds-linked taxa tended to plateau or decline. For DHA-associated taxa, Pseudomonas and Azospirillum were key indicators. QH soil exhibited a transient peak at Day 5 (6.0 × 107 16S rRNA gene copies g/dry soil for both), whereas MA and CZ soils showed notable increases in Azospirillum at Day 15 (4.4 × 107 16S rRNA gene copies g/dry soil each). SH and GL soils maintained lower, more stable DHA-associated taxa throughout flooding. Overall, SH and MA soils supported higher functional microbial abundances, contrasting with the limited functional potentials of CZ and GL soils. Differential Microbial Taxa and Their Environmental Correlation Linear regression analysis was performed between the estimated absolute abundances of putative functional microbial groups and the three key environmental indicators to further evaluate their specific associations (Figure d–f). Among the Fe­(III)-reducing genera, Anaeromyxobacter exhibited the strongest positive correlation with 0.5 M HCl-extractable Fe­(II) (p < 0.001; Figure d), followed by Desulfosporosinus (p < 0.001) and Desulfitobacterium (p < 0.001), whereas Clostridium (p < 0.001) and Geomonas (p < 0.001) showed moderate associations. Regarding WSP compounds, Lysinibacillus displayed the highest correlation (p < 0.001; Figure e), with Clostridium (p < 0.001) and Streptomyces (p < 0.001) also demonstrating significant positive relationships. For DHA, Bacillus (p < 0.001; Figure f), Pseudomonas (p < 0.001), and Azotobacter (p < 0.001) were the most closely linked taxa. These linear relationships quantitatively support a potential functional coupling between specific microbial populations and the accumulation of Fe­(II), WSP compounds, and DHA under anaerobic conditions. Metagenomic Evidence for Functional Potential Related to Fe Cycling and Microbial Redox Metabolism Metagenomic sequencing on Day 30 further revealed the distribution of functional genes associated with iron reduction, WSP compounds metabolism, and DHA activity across the five paddy soils (Figure ). For iron reduction, MtrA was the dominant gene in all soils, with the highest abundances in SH and MA (RPKM: SH, 142.5; MA, 115.6). FccA was detected only in SH, QH, and MA soils, peaking in MA (RPKM: 18.0), while CymAb was exclusively found in SH and MA, with MA showing significantly higher levels than SH (RPKM: 7.6 vs. 1.4). Genes involved in WSP compounds metabolism, such as xyn, were most abundant in SH and MA (RPKM: SH, 1308.4; QH, 802.2), followed by MA (RPKM:768.3), and lowest in CZ and GL (RPKM: ∼10). DHA-associated genes, including nuoN, exhibited the highest enrichment in SH, MA and QH (RPKM: 1132.5, 769.5 and 686.1, respectively), whereas SH, CZ, and GL soils showed > 90% lower abundances. These genetic profiles confirm that MA and SH soils possess enhanced functional potential for Fe­(II) mobilization and ROS precursor generation, while CZ and GL soils display limited iron-reducing and oxidative metabolic capacities. RPKM abundances of functional genes: Iron-reducing gene (a), WSP compounds-linked gene (b) and DHA-associated gene (c) in the five paddy soils on Day 30 of anaerobic incubation. Discussion Microbial-Driven Iron Reduction Induces ROS Formation during Anaerobic–Aerobic Alternation In paddy soils subjected to anaerobic–aerobic alternation, ROS production is driven primarily by the accumulation of reactive Fe­(II) during flooding and its subsequent oxidation. During the flooded phase, reducing conditions promote the reduction of Fe­(III) to Fe­(II), a process regulated by electron transfer chains. WSP compounds act as both electron shuttles and reductants in this process, while DHA enhance electron transfer and help sustain the reducing conditions. During the drainage and reoxygenation phase, the abundant accumulated Fe­(II) undergoes rapid oxidation, triggering Fenton-like reactions that generate •OH and other ROS. The reaction equations for the above process are as follows 1 Fe(II)+O2→Fe(III)+O2·− 2 Fe(II)+O2·−+2H+→Fe(III)+H2O2 3 Fe(II)+H2O2→Fe(III)+·OH+OH− As revealed by the reaction equations, Fe­(II) is the central species sustaining this chain reaction. A strong positive correlation was observed between the consumption of 0.5 M HCl-extractable Fe­(II) and cumulative •OH production (p < 0.001, Figure i), indicating that poorly crystalline Fe­(II) extracted by 0.5 M HCl serves as the principal precursor for ROS generation during the aerobic phase. During the anaerobic phase, Fe­(II) accumulation is primarily driven by iron-reducing bacteria, with electron transfer mediated by WSP compounds and DHA; both of these factors are themselves regulated by specific microbial taxa. This pathway was further corroborated by structural equation modeling (Figure , Text S8), which showed that iron-reducing bacteria primarily promoted the accumulation of reactive Fe­(II), whereas water-soluble phenolics and dehydrogenase acted as auxiliary drivers, amplifying Fe­(II) enrichment by sustaining electron-shuttling capacity and a strongly reducing conditions, respectively. These synergistic processes collectively maximized the reactive Fe­(II) reservoir under flooded conditions. Upon transition to aerobic conditions, the pre-accumulated Fe­(II) served as the dominant substrate, triggering Fenton-like reactions that generate •OH. (a) Piecewise SEM of •OH generation, R 2 indicates the proportion of variance in ROS production explained by the model. The values on the arrows are standardized path coefficients. Blue solid arrows indicate positive correlations, while purple dashed arrows indicate non-significant correlations (p > 0.05). Numbers adjacent to arrows are standardized path coefficients. *p < 0.05; **p < 0.01; ***p < 0.001. (b) Standardized effects on •OH production. From a functional perspective, three distinct microbial guilds coordinately drive the anaerobic accumulation of active Fe­(II), the principal precursor for subsequent ROS generation, through both direct and auxiliary pathways. Members of Anaeromyxobacter and Clostridium are known to participate in Fe­(III) reduction via respiratory and fermentative pathways, respectively. WSP compounds-linked taxa such as Lysinibacillus and Streptomyces contributed to the production of water-soluble phenolics (e.g., ferulic acid) through feruloyl esterase activity, thereby sustaining an active electron-shuttling capacity that facilitated Fe­(III) reduction and Fe­(II) stabilization. Concurrently, DHA-associated taxa including Bacillus and Pseudomonas enhanced the overall reducing milieu by elevating soil DHA, which indirectly promoted the reductive dissolution of Fe­(III) oxides and expanded the Fe­(II) reservoir. Collectively, this functional synergy support a conceptual model in which microbial Fe­(III) reduction, phenol-mediated electron transfer, and enhanced microbial redox activity jointly favored the accumulation of reactive Fe­(II) under anaerobic conditions, thereby establishing the foundation for Fenton-like •OH production upon reoxygenation. Intrinsic ROS Production Potential of Different Paddy Soils Under anaerobic–aerobic alternation, the ultimate ROS yield in paddy soils can be deconstructed into two sequential yet mechanistically distinct stages: microbially driven accumulation of reactive Fe­(II) during the anaerobic phase, which defines the “production potential”, and soil property-governed conversion efficiency of Fe­(II) to •OH during the aerobic phase, which dictates the “realized capacity”. During flooding, poorly crystalline Fe­(III) oxides serve as readily available electron acceptors, undergoing iron reduction by iron-reducing bacteria to generate surface-adsorbed and low-crystallinity-associated Fe­(II) that constitutes the intrinsic production potential. Microbially secreted WSP compounds (electron shuttles and ligands) accelerate Fe­(III) reduction, while DHA-associated taxa sustain reducing conditions, together amplifying Fe­(II) accumulation. This microbial synergism is particularly pronounced in weakly acidic soils rich in poorly crystalline iron, such as MA and QH, endowing them with exceptionally high ROS production potential. Notably, SH soil also exhibited substantial Fe­(II) accumulation capacity, yet its realized •OH yield was disproportionately low relative to its potential. This discrepancy between potential and realized capacity arises from the differential modulation of Fe­(II)-to-•OH conversion by soil geochemical factors during the aerobic phase. Recent evidence demonstrates that reactive Fe­(II) residing on low-crystallinity mineral surfaces and ligand-complexed sites, rather than aqueous Fe­(II), constitutes the dominant substrate driving Fenton-like •OH production. Soil pH directly governs the proportion of Fe­(II) available for productive conversion by altering its oxidation kinetics and solution speciation: acidic conditions (MA and QH) retard homogeneous Fe­(II) oxidation, thereby enhancing •OH yield, whereas the near-neutral pH of SH soil accelerates Fe­(II) oxidative precipitation, diminishing the effective substrate reservoir. The role of organic matter exhibits a pronounced two-phase paradox: although it may indirectly facilitate iron reduction as an electron donor under anaerobic conditions, during the aerobic phase dissolved organic matter and its functional groups collectively suppress net ROS yield through competitive •OH scavenging (with second-order rate constants of 108–109 M–1 s–1) and surface passivation effects. , Additionally, the near-neutral pH of SH soil (7.1) accelerates Fe­(II) oxidative precipitation, diminishing the reactive Fe­(II) pool available for Fenton reactions, in contrast to the acidic conditions of MA and QH soils (pH 5.6 and 5.3) that retard Fe­(II) oxidation and preserve its reactivity. Consequently, the combination of high organic matter content and near-neutral pH in SH soil jointly constrained its conversion efficiency, accounting for the observed decoupling between its high production potential and low realized •OH output. For CZ and GL soils, however, both meager poorly crystalline iron pools and alkaline pH limited production potential from the outset, resulting in correspondingly negligible realized capacity. ROS-Mediated Mitigation of Antibiotic and ARG Pollution in Paddy Fields ROS, particularly •OH, generated during the alternating anaerobic–aerobic process in paddy soil, effectively degrade TC and suppress the proliferation of tetA gene. The contribution rates of •OH to TC degradation and tetA removal were 35.4–50.3% and 54.5–62.1%, respectively (Figure b), which are primarily attributed to its high redox potential (E 0 = 2.8 V) and non-specific oxidative capacity. The degradation of TC is achieved mainly through attacks on key functional groups such as phenolic hydroxyl and dimethylamino groups via electron transfer, hydrogen abstraction, and addition reactions. In the case of tetA gene, •OH directly damages DNA molecules, causing irreversible alterations including base modification and phosphodiester bond cleavage, ultimately leading to attenuation of gene function. In the soil environment, ROS demonstrate a multi-pronged control effect by reducing antibiotic residues and suppressing the activity and dissemination of ARGs. Studies have shown that radicals such as •OH can effectively degrade typical antibiotics including fluoroquinolones, tetracyclines, and sulfonamides, significantly reducing their environmental persistence and biological activity. Meanwhile, ROS disrupt the structure of extracellular DNA (eDNA), directly impairing the transferability of ARGs. In addition, oxidative damage caused by ROS compromises microbial cell integrity, interferes with metabolic functions, and inhibits DNA repair mechanism, thereby reducing the survival of host bacteria and the stability of the ARGs they carry. Although soil components such as organic matter, pH buffering systems, and adsorption sites may attenuate the practical efficiency of ROS through matrix quenching effects, optimizing environmental conditions to regulate ROS generation and transformation offers a promising approach for achieving synergistic control of antibiotic removal and ARG dissemination. The 12 h aerobic phase employed in this study captures the most dynamic period of ROS generation during the flooding-drainage transition as the most intense ROS production occurs during the initial stage of drainage when O2 rapidly reacts with the accumulated Fe­(II) pool. This temporal pattern is representative of field drainage events, where ROS production diminishes once the Fe­(II) precursor is consumed. In recent years, the “frequent flooding-short drainage alternation” has been widely adopted for water conservation and pollution reduction in paddy systems. Our findings further elucidate its environmental benefits: the microbially driven reductive environment formed during flooding induces substantial ROS generation upon drainage, which effectively degrades residual TC and reduces ARG abundance. This dual mechanismdirect TC decomposition and suppression of ARG disseminationsubstantially mitigates potential risks to ecosystem integrity and human health. − Conclusion This study systematically elucidates the mechanisms and environmental effects of reactive oxygen species (ROS) generation in paddy soils under alternating wetting and drying conditions. During the anaerobic flooding phase, microbial synergy promoted the accumulation of reactive Fe­(II). Iron-reducing bacteria were the primary contributors to the reductive dissolution of poorly crystalline Fe­(III) oxides, while WSP compounds-linked taxa secreted water-soluble phenol that functioned as electron shuttles and reductants; DHA-associated taxa further facilitated Fe­(II) enrichment by sustaining a strongly reducing milieu. Collectively, these three functional guilds maximized the Fe­(II) reservoir under anaerobic conditions. Upon subsequent reoxygenation, the accumulated Fe­(II) species efficiently generated •OH and other ROS via Fenton-like reactions, which significantly accelerated tetracycline degradation and the attenuation of corresponding antibiotic resistance genes. The ROS production potential varied markedly among soil types, primarily governed by poorly crystalline Fe­(III) content and the abundance of Fe­(III)-reducing microorganisms, with alkaline pH acting as the predominant suppressive factor and elevated organic matter exerting additional inhibitory effects. From a biogeochemical coupling perspective, this work clarifies ROS formation patterns in dynamic redox environments and deepens the mechanistic understanding of iron cycle–mediated natural attenuation of pollutants. Supplementary Material