Skip to Content
LifeLife
  • Article
  • Open Access

27 January 2026

21 Pages

The Ratio of S2−/SO42− Induces the Transference of Cadmium in Rhizosphere Soil, Soil Pore Water and Root Iron Plaque

,
,
and
1
Key Laboratory for Lake Pollution Control of the Ministry of Ecology and Environment, National Engineering Laboratory for Lake Pollution Control and Ecological Restoration, Chinese Research Academy of Environmental Sciences, Beijing 100012, China
2
Department of Science, Technology, and Finance, Ministry of Ecology and Environment, Beijing 100006, China
*
Author to whom correspondence should be addressed.
This article belongs to the Section Plant Science

Abstract

Rice (Oryza sativa L.) readily accumulates cadmium (Cd), posing dietary exposure risks in populations dependent on rice-based diets. This study investigated how sulfur (S) redox processes regulate Cd mobility in S-deficient, Cd-contaminated paddy soil under waterlogged conditions. A pot experiment was conducted with two S treatments (−S and +S, 30 mg kg−1) throughout the rice growing season. S addition markedly increased pore water S2− concentrations during early growth (tillering) and mid-season (booting) and suppressed the diffusion of SO42− from non-rhizosphere to rhizosphere at later stages (filling–maturity). Consequently, Cd in soil pore water was significantly lower in +S than −S treatments at all stages. Sulfur-amended soil showed a redistribution of Cd from labile fractions (exchangeable and carbonate-bound) to more stable fractions (Fe/Mn oxide-bound). Sulfur application also altered the rhizosphere microbiome: the relative abundance of sulfate-reducing bacteria (SRB) increased at the booting and filling stages, while sulfur-oxidizing bacteria (SOB) became more dominant at maturity. Additionally, +S enhanced Cd sequestration on rice root iron plaque by 32–67% during the grain-filling and maturity stages compared to −S. Throughout the rice growing period, redox-driven shifts in the S2−/SO42− ratio emerged as a key control on Cd behavior, with low pe + pH (strongly reducing conditions) promoting Cd sulfide precipitation and high pe + pH (more oxidizing conditions) causing Cd remobilization.

1. Introduction

Rice (Oryza sativa L.) readily absorbs and accumulates cadmium (Cd) from soil, leading to elevated Cd levels in rice grain and increased dietary Cd exposure in rice-consuming populations [1]. This is a critical food safety issue in many Asian countries where rice is a staple crop [2,3]. In China, soil Cd concentrations have risen in recent decades, with nearly 19% of croplands contaminated to varying degrees [4]. Some paddy soils show Cd levels up to 27.3 mg kg−1 [5], and rice grains from such fields can exceed food safety limits (0.2 mg Cd kg−1) fivefold [6]. Despite the contamination, these lands often remain under cultivation to meet food demand, necessitating mitigation measures to ensure rice safety [7]. Under current regulations (e.g., Chinese risk control standards), slightly Cd-polluted paddy soils can be used for rice if water management or soil amendments are applied to reduce Cd bioavailability.
Sulfur (S) amendments have been proposed as one such measure, as S can effectively decrease Cd availability in flooded paddy soils [8]. In anaerobic conditions, S is microbiologically and chemically reduced to sulfide (S2−) or converted into sulfhydryl compounds, which can precipitate insoluble Cd sulfides (CdS) or bind Cd in stable complexes [9,10,11]. For example, Fulda et al. observed increased CdS and Cd–thiol species in anoxic paddy soil as redox potential (Eh) declined [12]. Other studies reported that S applications shift Cd from readily bioavailable fractions (exchangeable and carbonate-bound) to more geochemically stable forms like Fe oxide-bound or residual fractions [13,14]. These findings suggest that sulfur-driven redox processes can immobilize Cd in flooded soils, yet the effect may not be uniform over time.
Sulfur cycling in paddy soils is dynamic and involves opposing processes of sulfate reduction and sulfide oxidation. Microbial sulfate-reducing bacteria (SRB) generate S2− from SO42− under anoxic conditions, promoting precipitation of metal sulfides such as CdS and thus immobilizing Cd in the solid phase [15]. Conversely, chemical or microbial sulfur-oxidizing activity (via SOB) can re-oxidize S2− back to SO42− when oxic microsites exist (e.g., near roots), potentially releasing previously sequestered Cd [11,16]. These coupled sulfidation–re-oxidation reactions cause temporal swings in dissolved Cd and its solid-phase speciation during rice growth. Notably, as redox conditions oscillate with flooding duration and root activity, Cd may precipitate at one stage and remobilize at another. The extreme example of this is the “Cd rebound” phenomenon: a sudden increase in pore water Cd observed mid-season when conditions become less reducing, which likely results from the oxidation of earlier-formed CdS and other sulfide complexes [17].
Microbial communities mediate these redox transitions. SRB (e.g., Geobacter, Desulfobacca, Desulfovibrio) thrive in reduced zones of the rhizosphere and drive sulfate reduction, whereas SOB (e.g., Rhodopseudomonas) flourish when oxygen or alternative electron acceptors are available and drive sulfide oxidation [18,19]. Changes in the balance or activity of SRB vs. SOB can therefore tilt the S2−/SO42− equilibrium, modulating whether Cd is precipitated as sulfide or remains soluble. Recent studies have highlighted that the composition of the rhizosphere microbiome (including SRB/SOB abundance) can significantly influence Cd fate [16,20]. For instance, inoculation or stimulation of SRB has been shown to decrease plant Cd uptake by promoting CdS formation, while active SOB can accelerate Cd remobilization during oxidative phases of wetland soil cycles [20]. Another critical factor is the formation of iron plaque on rice roots. Root iron plaque consists of Fe(III) oxides that precipitate on root surfaces due to radial oxygen loss from roots into the anoxic soil [21]. This plaque can sorb or co-precipitate Cd and other metals, thus acting as a barrier to Cd entry into the plant. The extent and effectiveness of iron plaque in sequestering Cd depend on redox conditions and interactions with sulfur species. Under strongly reducing conditions, excess S2− might react with Fe2+ to form FeS/FeS2, potentially limiting Fe(III) plaque formation or even dissolving existing plaque by reducing Fe(III) to Fe2+. Conversely, under moderately oxidizing conditions, robust plaque formation can trap Cd but may later release it if conditions shift [22,23,24]. Thus, there is a complex interplay: S-induced sulfide may compete with or modify iron plaque, with consequent effects on Cd partitioning among soil, pore water, and roots.
In summary, the S2−/SO42− ratio in the rhizosphere emerges as a potential “master variable” integrating soil redox status, microbial activity, and mineral interactions, which together govern Cd mobility [25,26,27]. This ratio will vary with soil conditions (flooding regime, organic matter availability) and plant developmental stage (which influences root oxygen release and exudation) [28]. Understanding how the S2−/SO42− balance controls Cd speciation and transfer in the soil–plant system is essential for both biogeochemical insight and for designing effective interventions.
The objectives were to (1) determine how S addition influences Cd partitioning in the rhizosphere at different growth stages; (2) elucidate the mechanistic roles of redox conditions and sulfur-transforming microbes (SRB/SOB) in driving Cd precipitation or release; and (3) evaluate the contribution of root iron plaque to Cd immobilization under varying S2−/SO42− regimes. By integrating geochemical and microbiological data, we aim to clarify how the S2−/SO42− ratio governs Cd mobility, and to identify practical implications for using sulfur amendments to mitigate Cd uptake by rice.

2. Method and Materials

2.1. Soil Preparation and Characterization

An uncontaminated paddy soil (surface 0–20 cm) from Huaining County, Anhui Province, was chosen. The soil was air-dried, gently crushed, and passed through a 2 mm sieve to remove stones, roots, and debris. To simulate a Cd-contaminated paddy field, a portion of this soil was spiked with CdCl2 solution and thoroughly mixed with the remaining soil to achieve a target total Cd concentration of 10 mg kg−1. The spiked soil was incubated (aging) for 4 weeks at 60% water-holding capacity to stabilize Cd before use. After aging, the soil was again air-dried and sieved (2 mm). Initial properties of the prepared soil were as follows: pH 6.56 (1:5 soil/water), available S 16.5 mg kg−1, total Cd 10.16 mg kg−1, and DTPA-extractable Cd 8.96 mg kg−1.

2.2. Pot Experiment Design

A pot experiment was conducted with two treatments (n = 3): −S (no sulfur addition) and +S (30 mg S kg−1 soil). A moderate S rate (≈60–70 kg S ha−1 for the plow layer) was selected to influence sulfur redox processes while minimizing risks of excessive acidification or phytotoxicity. Sulfur was applied as reagent-grade Na2SO4 and thoroughly mixed into the soil of the +S pots before flooding and transplanting; no sulfur was added to the −S pots. Pots were made of polymethyl methacrylate (24 cm × 24 cm × 20 cm). To separate rhizosphere and non-rhizosphere compartments, a 30-μm nylon-mesh inner box (12 cm × 12 cm × 20 cm) was installed at the center of each pot. The mesh permitted water and solute exchange but restricted root growth to the inner compartment, thereby creating a root-influenced rhizosphere inside the box and a root-free non-rhizosphere outside. Each pot received 1.75 kg of Cd-spiked soil in the inner box and 5.25 kg of the same soil in the outer compartment. Basal fertilizers (per pot: 1.4 g KCl, 1.05 g KH2PO4·H2O, and 1.4 g CO(NH2)2) were applied and mixed uniformly before flooding. All pots were flooded with deionized water to 2 cm above the soil surface and pre-incubated for 7 d prior to transplanting. Three pre-germinated rice seedlings were transplanted into the inner compartment, and plants were thinned to one healthy seedling per pot after 7 d. Floodwater depth was maintained at ~2 cm throughout the cultivation period (≈4 months). No mid-season drainage was applied, allowing assessment of Cd−S dynamics under continuous waterlogging, while acknowledging that field drainage practices may modify redox conditions and Cd behavior.

2.3. Sampling and Measurements

Soil pore water and soil samples were collected at four key growth stages of rice: tillering (21 days after transplanting, DAT), booting (53 DAT), flowering/filling (82 DAT), and maturity (112 DAT). At each sampling, we collected rhizosphere pore water, non-rhizosphere pore water, rhizosphere soil, and non-rhizosphere soil separately. Pore water was extracted in situ from each compartment using Rhizon soil solution samplers (Rhizosphere Research Products, Giesbeek, The Netherlands) inserted horizontally. The Rhizon is a porous PTFE probe (0.15 µm pore size) that draws soil solution by vacuum. Approximately 10–15 mL of pore water was obtained from rhizosphere and from non-rhizosphere zones in each pot. To preserve sulfide (S2−) and prevent its oxidation during handling, 5 mL of fresh pore water was immediately mixed 1:1 with 5 mL of an alkaline antioxidant buffer (SAOB: 2 M Na salicylate, 0.41 M ascorbic acid, 2 M NaOH) in the field [29]. This stabilized solution was diluted to 20 mL with de-oxygenated ultrapure water and used for S2− measurement within 2 h of sampling. The remaining pore water was filtered (0.45 µm) and refrigerated (4 °C) for analysis of sulfate (SO42−), dissolved Cd, dissolved Fe, and dissolved organic carbon (DOC). After pore water sampling, soil from the rhizosphere box and the non-rhizosphere zone was sampled using an auger. Rhizosphere soil was defined as the soil inside the mesh where rice roots were present; non-rhizosphere soil was taken from the bulk soil outside the mesh (ensuring no roots in the sample). The soil samples were immediately stored at –80 °C to halt microbial activity. A portion of each soil sample was later thawed for DNA extraction (to characterize sulfur-cycling microbes), and the remainder was freeze-dried for chemical analyses (Cd fractions, basic properties).

2.4. Analysis of Soil and Its Pore Water

We conducted the determination of soil pH using a glass electrode with a soil–water ratio of 1:5 (W/V) (SevenExcellence™, Mettler Toledo, Zurich, Switzerland). The redox potential (Eh) was assessed in situ by determining the potential difference between a platinum (Pt) and saturated calomel electrode (FJA-6, Di Chuan, Najing, China). Soil total Cd was determined by digesting with HCl-HNO3-HF following the detection by using ICP-MS (Thermo Fisher Scientific, ELEMENT™, Waltham, MA, USA) [30]. Available Cd was extracted by the solution of 0.05 M DTPA, 0.1 M TEA, and 0.01 M CaCl2 and analyzed using ICP-MS [31]. Available S was extracted using 0.01 M Ca(H2PO4)2, followed by the determination using ICP-OES (Agilent 5800, Santa Clara, CA, USA) [32].
We employed a sequential extraction procedure to extract different Cd fractions in the soil outlined by Tessier et al. [33]. Briefly, 1 g fresh-dried soil was successively extracted by 8 mL of 1 M MgCl2 for exchangeable Cd, 8 mL of 1 M NaOAc for carbonate Cd, 20 mL of 0.04 M NH2OH·HCl (in the 25% HOAc) for Fe/Mn oxide Cd, and 20 mL of 0.02M HNO3 (in the 30% H2O2) for organic Cd. Cd concentrations in the extractant of each extraction step were determined using ICP-OES. The residual was digested with HCl-HNO3-HF followed by ICP-OES determination of residual Cd.
Soil pore water was acidified with HCl before analysis. DOC concentration was detected by TOC analyzer (Apollo 9000, Thermo Fisher Scientific, USA). The concentrations of Cd and Fe were determined using ICP-MS. The concentration of SO42− was analyzed using ion chromatography (Agilent 6000 Series, USA), and S2− was measured using a silver/sulfide-selective electrode (9616BNWP, Thermo Fisher Scientific, USA).

2.5. DNA Extraction and High-Throughput Sequencing

Total soil bacterial DNA was extracted from 0.5 g of each rhizosphere soil sample (stored at −80 °C) using the DNeasy PowerSoil Kit (MoBio, San Diego, CA, USA) according to the manufacturer’s protocol. DNA yield and purity were checked with a NanoDrop spectrophotometer and by agarose gel electrophoresis; DNA integrity was confirmed on a Bioanalyzer 2100 (Agilent, USA). High-throughput sequencing of the bacterial 16S rRNA gene was employed to profile the community, focusing on sulfur-cycling groups. The V3–V4 hypervariable region of the 16S rRNA gene was PCR-amplified using primers 341F (5′-CCTACGGGNGGCWGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). Unique sample barcodes were ligated to amplicons, and libraries were sequenced on an Illumina MiSeq platform (2 × 250 bp) [34]. Quality control (trimming of low-quality reads and adapter removal) was performed with FastQC. Operational taxonomic units (OTUs) were clustered at 97% sequence identity. Taxonomic assignment was performed against the SILVA rRNA database (v132) [35]. From the sequencing data, we specifically quantified the relative abundance of known sulfate-reducing bacteria (SRB) and sulfur-oxidizing bacteria (SOB) genera in each sample, since these functional groups were of interest.

2.6. Iron Plaque Extraction and Analysis

To evaluate Cd sequestration in root iron plaque, rice roots were harvested at booting, filling, and maturity stages and the Fe-oxyhydroxide plaque on root surfaces was extracted using an ascorbate–citrate–acetate (ACA) solution [36]. The entire root system from each plant was gently washed to remove loose soil, then immersed in 50 mL of ACA solution (consisting of 3.0 g ascorbic acid + 40 mL 0.3 M sodium citrate + 5 mL 10% acetic acid, adjusted to pH 3) for 2 h with occasional shaking. This mild reductive solution dissolves Fe(III) oxides (i.e., the iron plaque) without significantly attacking the root tissue itself. After extraction, the solution was filtered and brought to 100 mL volume. The extracts were analyzed for Fe, Mn, S, and Cd by ICP-OES or ICP-MS (for Cd). The Fe concentration from this extraction is taken as the Fe content of iron plaque, since internal root Fe would remain largely undissolved by the mild ACA solution. The Cd in the extract (termed ACA-extractable Cd) represents Cd that was bound in or on the plaque. After ACA treatment, roots were thoroughly rinsed with deionized water to remove any residual extractant and then oven-dried (60 °C) to constant weight for Cd analysis of root tissues. Iron plaque amount (mass of Fe per root dry weight) and thickness can be estimated by the Taylor & Crowder method [37], but here we primarily use the chemical data (Fe and Cd in plaque) for comparisons.

2.7. Statistical Analysis

Treatment effects and temporal trends were assessed using analysis of variance (ANOVA). For each parameter (e.g., pore water Cd, soil Cd fractions, plaque Cd), we performed two-way ANOVAs where appropriate (factors: S treatment and growth stage) or one-way ANOVAs for comparisons within a given stage. Post hoc comparisons of means were performed with Fisher’s least significant difference (LSD) test at p < 0.05. In figures, significant differences are indicated either by different letters for comparisons among stages or by asterisks for pairwise contrasts between −S and +S treatments, as detailed in each figure caption. All statistical analyses were conducted in SPSS 23 (IBM, Armonk, NY, USA). To explore relationships among variables, we performed redundancy analysis (RDA) and Pearson correlation analyses. RDA was used to relate soil pore water properties (Eh, pH, DOC, S2−, SO42−, Fe) to the distribution of Cd in soil fractions across treatments and stages. We also constructed a structural equation model (SEM) to delineate hypothesized causal pathways linking soil conditions (pe + pH, S2−/SO42− ratio, etc.), microbial populations, and Cd in pore water and plaque. The SEM was evaluated using partial least squares estimation (SmartPLS 4.0), and path coefficients were considered significant at p < 0.05 or 0.01 as marked. All data are presented as mean ± standard deviation (n = 3 pots).

3. Results

3.1. Dynamics of the Soil Pore Water Properties

Soil pore water pH fluctuated between 5.6 and 6.6 over the season (Figure 1c). In all treatments, pH tended to decline from tillering to booting, then rose slightly by maturity. S addition had a minor acidifying effect, resulting in slightly lower pH than control at most stages, but these differences were not statistically significant. Redox potential (Eh) exhibited a more pronounced temporal pattern (Figure 1b): Eh dropped from +250 mV at tillering to as low as –100 mV at booting, indicating strongly reduced conditions, then increased again by the filling stage (+100 to +200 mV), and finally decreased slightly at maturity. This mid-season rebound in Eh (booting to filling) was observed in both −S and +S, likely due to intensified radial oxygen loss from rice roots around flowering. The combined pe + pH followed a similar trend (Figure 1d), ranging from 11 down to 4.3. Notably, at the booting stage, pe + pH reached its minimum (4.3–5.0), which in the +S rhizosphere was below the theoretical threshold (pe + pH ≈ 4.85) for CdS precipitation. This indicates extremely reducing conditions conducive to metal sulfide formation. No significant differences in Eh or pe + pH were found between +S and −S treatments at any stage, indicating that the redox environment was largely governed by flooding and root activity rather than S addition in this experiment.
Figure 1. Dynamics of pore water properties in rhizosphere vs. non-rhizosphere soil at tillering (21 d), booting (53 d), filling (82 d), and maturity (112 d) stages (mean of three replicates). Panels show: (a) DOC, (b) Eh, (c) pH, (d) pe + pH, (e) sulfate (SO42−), (f) sulfide (S2−), (g) dissolved Fe, and (h) dissolved Cd. Square symbols = rhizosphere; Triangle symbols = non-rhizosphere. Black = −S (control), red = +S treatment.
The pore water sulfate and sulfide showed nearly opposite trends (Figure 1e,f), reflecting sulfur redox transformations. In the control (–S), sulfate in rhizosphere pore water gradually declined over time, whereas in non-rhizosphere water it initially decreased, then increased slightly by maturity. With S addition, there was an initial spike of sulfate mainly detectable in the non-rhizosphere: +S non-rhizosphere had higher SO42− than −S through tillering and booting. However, by the filling stage, an interesting redistribution occurred: sulfate in +S rhizosphere remained lower than in −S rhizosphere, while in the non-rhizosphere, the +S treatment showed higher sulfate than −S. In essence, S addition inhibited or satisfied the sulfate demand of the rhizosphere such that less sulfate migrated inward during grain filling. Sulfide concentrations were mostly below detection at tillering, then rose markedly by booting as conditions became strongly reducing (Figure 1f). From booting to filling, S2− levels dropped in all treatments, coinciding with the rise in Eh—indicating sulfide was being oxidized during that interval. Across all stages, rhizosphere pore water had significantly higher S2− than non-rhizosphere (for example, at booting, rhizosphere S2− was 0.6–1.3 mg L−1, non-rhizosphere S2− was below 0.2 mg L−1). S addition further amplified sulfide production in the rhizosphere: at booting, +S rhizosphere S2− was 1.3 mg L−1, nearly double that of −S (0.7 mg L−1). This confirms that added sulfate was rapidly biologically reduced to sulfide in the root zone during early growth. By maturity, sulfide levels were low in all cases due to partial re-oxidation and possibly sulfate depletion.
Iron in pore water (Figure 1g) increased under reducing conditions as Fe(III) in soil was reduced to Fe2+. In rhizosphere pore water, Fe rose sharply from tillering to a peak at filling (30–40 mg L−1 Fe), then subsided by maturity as some Fe re-precipitated. In contrast, non-rhizosphere pore water Fe remained much lower and even decreased during the booting–filling interval. The rhizosphere had significantly higher Fe than the non-rhizosphere at all times, reflecting root-induced mobilization and translocation of Fe. Notably, during the booting-to-filling stage, rhizosphere Fe spiked while non-rhizosphere Fe dropped, indicating a net movement of Fe into the root zone, consistent with root uptake of Fe2+ and plaque formation. S addition consistently led to lower dissolved Fe in the rhizosphere compared to −S. This suggests that with extra S, more Fe2+ was captured as FeS or related precipitates instead of accumulating in solution. In other words, S-driven sulfide likely precipitated some Fe as iron sulfides (FeS, FeS2), thereby lowering soluble Fe in pore water.
Cd in pore water was initially very low at tillering (around 0.01–0.02 mg L−1) but increased dramatically by the filling stage (Figure 1h). In the control −S rhizosphere, pore water Cd surged to 1.0 mg L−1 at filling, before dropping to 0.2 mg L−1 at maturity. This mid-season “Cd rebound” (from booting to filling) was observed in all treatments, but S addition mitigated its magnitude. Throughout the rice growth, +S treatment had significantly lower pore water Cd than −S in both rhizosphere and non-rhizosphere. For example, at the filling stage, rhizosphere Cd was 0.60 mg L−1 in +S and 1.03 mg L−1 in −S. Even at maturity, +S maintained lower Cd in pore water. These results show that sulfur application consistently reduced Cd availability in pore water at every stage. It is also evident that the rhizosphere had higher Cd than the bulk soil in late stages, irrespective of S: at filling, rhizosphere Cd > non-rhizosphere Cd (Figure 1h). By maturity, pore water Cd subsided but remained higher in −S than +S.
Overall, sulfur addition increased pore water S2− (especially at booting), modified SO42− distribution between zones, and consistently reduced dissolved Cd and Fe relative to −S. The temporal patterns identify critical windows: strong reduction from tillering to booting likely favored Cd immobilization, whereas the booting-to-filling redox rebound coincided with sulfide oxidation and a pronounced Cd increase. Although +S pots also exhibited a Cd spike at filling, its magnitude was smaller than in controls, indicating that S-driven immobilization was partially reversed during oxidative phases but still provided net mitigation of Cd availability. These dynamics suggest active Cd transfer among soil solids, pore water, and root-associated sinks, which is further evaluated using solid-phase and microbial analyses.

3.2. Cd Fractions in Paddy Soil

At tillering, the Cd in soil was largely in readily soluble forms due to the initial contamination procedure. Exchangeable Cd was high (in rhizosphere 2.1 mg kg−1, 21% of total Cd in −S soil), and carbonate-bound Cd was also significant (1.8 mg kg−1, 18%). Fe/Mn oxide-bound Cd made up 35%, and the rest was organic/residual (25%). Comparing treatments, +S had a slightly higher exchangeable Cd concentration than −S at tillering. However, the proportion of Cd in an exchangeable form was actually lower with S, because +S also had a reduction in carbonate and Fe/Mn fraction content at tillering.
By booting (after 8 weeks of flooding), total extractable Cd in soil decreased. All fractions showed their lowest concentrations at booting, reflecting that much of the Cd had either been solubilized into pore water or possibly sequestered in forms not captured by these extraction steps. Importantly, at booting, the difference between −S and +S became pronounced: in +S rhizosphere soil, exchangeable Cd and Fe/Mn oxide Cd concentrations were significantly higher than in −S, even though proportions were similar. This suggests that S addition slowed down the depletion of those fractions, possibly by precipitating some Cd as sulfide while some remained exchangeable.
At the filling stage, differences sharpened further. In non-rhizosphere soil, +S treatment showed consistently lower percentages of exchangeable and carbonate Cd and higher Fe/Mn-bound Cd than −S. In rhizosphere soil, +S reduced the exchangeable Cd proportion by 5.5% at filling compared to −S. Thus, sulfur application clearly shifted Cd into more strongly sorbed phases by the mid-to-late season. By the mature stage, this trend was most evident: +S rhizosphere soil had 11% exchangeable Cd, but 22% in −S, and correspondingly more Cd in the Fe/Mn oxide fraction (45 in +S and 34% in −S). A similar pattern was seen in the non-rhizosphere at maturity. These data indicate that S amendment led to a long-term immobilization of Cd in less labile forms in both root-associated and bulk soil.
Statistical analysis (Figure 2) confirmed that S treatment significantly affected Cd fraction distribution at certain stages. For instance, at maturity, exchangeable Cd was significantly lower in +S (both in absolute concentration and %), while Fe/Mn oxide-bound Cd was significantly higher in both zones (p < 0.05). The mechanisms likely involve S-driven sulfide formation and Fe sulfide precipitation, which reduce soluble Cd, as well as possible enhancement of Fe (hydr)oxide surfaces (via iron plaque or soil Fe minerals) that sorb Cd.
Figure 2. Effect of sulfur addition on Cd fractionation in (A) rhizosphere soil and (B) non-rhizosphere soil at the tillering, booting, filling, and mature stages. Bars show the concentration of Cd in each fraction (mg kg−1 soil) under −S and +S, stacked in order: exchangeable, carbonate, Fe/Mn oxides, (plus minor organic and residual not visible). Percent values indicate the proportion of total Cd in that fraction.

3.3. Relative Abundance of SRB and SOB in Rhizosphere

Under control conditions (–S), the composition of SRB and SOB in the rhizosphere changed substantially with plant development. For SRB, Geobacter and Desulfobacca were particularly enriched at the booting stage and then declined later. Desulfosporosinus and Desulfotomaculum showed an opposite trend, being more abundant at maturity than earlier (Figure 3c,d). These patterns suggest that different SRB taxa prosper at different times—possibly the booting stage favors certain fast-growing SRB (like Geobacter), whereas by maturity, conditions (such as substrate availability or redox), others are favored (spore-forming SRB like Desulfosporosinus) (Figure 3). For SOB, Rhodopseudomonas was notable for spiking during the booting-to-filling period and then dropping by maturity. In contrast, genera like Defluviicoccus and Rhodomicrobium increased at maturity.
Sulfur fertilization significantly impacted the abundance of several key genera, especially at stages where sulfur transformation activity was highest (booting and maturity). At the booting stage, +S treatment significantly increased the relative abundance of the two most dominant SRB genera, Geobacter and Desulfobacca, by roughly 40–60% compared to −S. +S also led to a moderate increase in Desulfovibrio at booting. These results indicate that providing sulfate stimulated the proliferation or activity of SRB at mid-season—presumably because these bacteria had more substrate (SO42−) to respire, yielding more energy and growth. Indeed, the ratio of SRB to SOB (total abundance) in the rhizosphere was higher in +S at booting than in −S, reflecting a shift towards net sulfate-reducing conditions. At maturity, +S treatment significantly increased certain SRB that peak late: Desulfosporosinus and Desulfitobacterium were more abundant in +S than −S, as was Desulfotomaculum. For SOB genera, +S had a marked effect at maturity: Defluviicoccus and Rhodomicrobium (which were the dominant SOB in the late season) were significantly higher in +S than −S at maturity. In contrast, at booting and filling, +S did not significantly raise Rhodopseudomonas. One interesting observation is that at the filling stage, some SOB genera (Rhodomicrobium, Rhodobacter) actually showed a decrease in relative abundance under +S compared to −S. This might be because +S dampened the overall extent of sulfide oxidation (less substrate for those SOB) during the booting-filling transition, as evidenced by lower sulfate accumulation in +S rhizosphere (Figure 1e).
Figure 3. Relative abundance (%) of dominant sulfur-cycle bacteria in rhizosphere soil at tillering, booting, filling, and maturity (–S vs. +S). Top: SRB genera (a) Geobacter, (b) Desulfobacca, (c) Desulfovibrio, (d) Desulfosporosinus, (e) Desulfitobacterium, (f) Desulfotomaculum. Bottom: SOB-related genera (g) Defluviicoccus, (h) Rhodopseudomonas, (i) Rhodomicrobium, (j) Hyphomicrobium, (k) Rhodobacter, (l) Pedomicrobium. Bars are means ± SD (n = 3). Different letters above bars indicate significant differences among growth stages within the same treatment (p < 0.05, LSD). Asterisks denote significant differences between −S and +S at a given stage (* p < 0.05; ** p < 0.01).

3.4. ACA Extractable Plaque Fe and Cd

In the −S (control) treatment, iron plaque development followed a typical pattern: the amount of Fe on roots (Figure 4a) increased from booting to a maximum at filling (6.5 g Fe kg−1 root) and then slightly declined by maturity (5 g/kg). The plaque Cd content (Figure 4b) was highest at booting (3.2 mg Cd/kg root), then decreased at filling (2.2 mg/kg) and further at maturity (1.0 mg/kg). This resulted in a booting > filling > maturity trend for plaque Cd in −S. The decrease in plaque Cd over time, despite plaque Fe peaking at filling, suggests that some Cd bound in plaque at booting might have been remobilized or translocated by later stages, or diluted by additional Fe precipitation with lower Cd content.
Under +S, the plaque Fe content was markedly affected. At booting, +S plaque Fe (3.7 g/kg) was significantly lower than −S (5.8 g/kg). At filling, +S plaque Fe remained lower than −S (4.8 and 6.6 g/kg, p < 0.05). Only by maturity did +S plaque Fe catch up or exceed −S (5.5 and 4.9 g/kg; +S was slightly higher, p < 0.05). Overall, S application seemed to suppress iron plaque formation during the early and mid-season, but led to somewhat more plaque formation by the end. Indeed, our pore water data showed lower dissolved Fe in +S (Figure 1g), implying Fe removal by sulfides. The slight rebound of plaque Fe in +S at maturity could be because by late season, some sulfide was oxidized back to sulfate (reducing competition), and roots continued to release oxygen, allowing Fe plaque to accumulate on any remaining active root surfaces.
Plaque Cd content showed an even more interesting behavior. At booting, +S had significantly less Cd on plaque than −S (1.8 and 3.2 mg/kg, a 44% drop, p < 0.01). However, at the filling stage, +S plaque Cd surged to higher than −S (3.0 and 2.2 mg/kg, p < 0.05), and at maturity, +S remained slightly higher (1.3 vs. 1.0 mg/kg, not a large difference but notable). In other words, S application caused a delay and redistribution in plaque Cd sequestration: initially, less Cd was found on roots (perhaps because Cd stayed in soil or pore water as sulfide precipitates instead of on plaque), but during grain filling, a significant Cd accumulation on plaque occurred in the +S treatment. This aligns with our earlier observation of a Cd “transfer” among compartments at filling: soil Cd dropped, pore water Cd spiked (Figure 1h), and plaque Cd increased, especially in +S. The data suggest that as sulfide was oxidized and Cd remobilized at filling, the presence of iron plaque (though smaller in mass in +S at that time) captured a good portion of that Cd, effectively acting as a sink.
Figure 4. Effect of sulfur addition on ACA extractable Fe (a) and Cd (b) (g·kg−1 DW) in iron plaque exposed to Cd-contaminated soil at the booting, filling, and mature stages. Different letters indicate the significance at p < 0.05 (LSD) for different stages in the same treatment. *, ** indicate the significant difference between -S and +S treatments at p < 0.05, 0.01, respectively.

4. Discussion

4.1. Sulfur-Induced Changes in Cd Availability over the Growing Season

Our results demonstrate that sulfur addition significantly decreased Cd availability in the soil–plant system, but the extent of this effect varied with rice growth stages. Throughout the experiment, +S treatment consistently showed lower pore water Cd concentrations than the control, indicating a net immobilization of Cd by S (Figure 1h). This finding aligns with previous studies; Liu et al. observed that supplying 45 mg S kg−1 to a flooded paddy soil lowered pore water Cd levels and subsequent plant uptake [38]. The primary mechanisms for Cd immobilization by S are as follows: (i) precipitation of Cd as Cd sulfide (CdS) or co-precipitation with iron sulfides (FeS, FeS2), and (ii) S-induced changes in the soil solid phase that favor Cd sorption (e.g., conversion of Cd into Fe/Mn oxide-bound forms). Our data provide evidence for both mechanisms [39,40].
Under continuously flooded, reducing conditions, added sulfate was microbially reduced to sulfide, which can precipitate Cd as CdS (a mineral with an extremely low solubility, K_sp ≈ 10−27). At the booting stage, the pe + pH of the rhizosphere in +S dropped to 4.3 (Figure 1d), below the critical value (4.85) for CdS formation [40]. This thermodynamic threshold indicates that Cd2+ and S2− could coexist in a saturation state, favoring the CdS solid. Correspondingly, sulfide (S2−) concentrations were significantly higher in +S at booting (Figure 1f), and pore water Cd was very low at that time. We infer that much of the Cd was temporarily locked up as CdS (and/or related sulfhydryl complexes) during early growth when conditions were strongly reducing. Although we did not directly identify CdS with spectroscopy (a limitation of our study), the indirect evidence is strong: low pe + pH, high S2−, and low soluble Cd, which is exactly the scenario for CdS precipitation. Prior research concurs that CdS can form in flooded soils given sufficiently low redox potential and available S2− [41,42].
Another pathway for Cd immobilization under S addition was the transformation of Cd into more sorbed fractions. By the and of the season, +S soils had a significantly greater fraction of Cd associated with Fe/Mn oxides and less in exchangeable form than −S soils (Figure 2). This suggests that sulfur might have indirectly promoted the adsorption or incorporation of Cd onto iron (hydr)oxides. One possible explanation is that partial oxidation of sulfide (by bioturbation or diffusion of oxygen) may have produced fresh Fe(III) oxides (from Fe2+) that act as new binding sites for Cd. Additionally, sulfide can stimulate soil aging processes that incorporate Cd into less labile forms. The decline in pore water Cd with S was partly attributable to this fraction shift: as labile Cd was scavenged by sulfide precipitation or by sorption to Fe oxides, the immediate bioavailable pool shrank. We observed a positive effect of S on Fe/Mn-bound Cd, especially at maturity (Figure 2), meaning more Cd ended up strongly held in that fraction in S-treated soil. This corroborates findings by Zhang et al. (2020), who reported decreased exchangeable Cd and increased oxide-bound Cd with S amendments [43]. Thus, S application not only causes insoluble sulfide formation but can enhance longer-term Cd fixation in the soil matrix.
Despite these clear immobilization processes, the research recorded a transient increase in pore water Cd from booting to filling stage (Figure 1h). This can be explained by a redox reversal: as the rice entered the reproductive phase (around panicle initiation to flowering), root oxygen release intensified (to meet plant aeration and nutrient uptake needs), raising the redox potential in the rhizosphere. The measured Eh rose by 300 mV from booting to filling (Figure 1b), which corresponds to pe + pH increasing from 4 to 8 (Figure 1d)—an environment no longer favorable for metal sulfides. Consequently, previously precipitated CdS would begin to dissolve, releasing Cd2+ back into solution. Simultaneously, S2− was oxidized to sulfate (Figure 1e,f), further corroborating that oxidative processes occurred. Our data show a clear jump in pore water sulfate and dissolved Cd concurrently at filling in both treatments (though attenuated in +S). This Cd rebound is essentially the undoing of some earlier immobilization: as conditions became less reducing, Cd that had been locked in sulfide or weakly bound forms was remobilized.
Importantly, the sulfur-amended soil suffered a smaller Cd rebound than the control. In the −S rhizosphere, pore water Cd shot up to 1.0 mg L−1 at filling (from 0.02 mg L−1 at booting), whereas in +S it rose to 0.6 mg L−1 (from 0.01 to 0.6)—still a large increase but a 40% lower peak. We attribute this to a few factors: (1) greater Cd immobilization to start with in +S (so less absolute Cd was available to rebound); (2) continued presence of sulfide in +S during filling, which may have buffered some Cd via re-precipitation or kept Cd in complexed form; and (3) enhanced Cd recapture by iron plaque in +S. The implication is that while S cannot completely prevent Cd remobilization if oxidation occurs (indeed, some CdS will inevitably oxidize when exposed to O2), it can mitigate the extent and duration of Cd release.
Another phenomenon observed was the decline in soil total Cd from booting to filling (especially in −S), suggesting a net removal of Cd from the soil solid phase. Part of it was the pore water Cd spike, and eventually some was taken up by the plant or relocated to the root plaque. In +S, though pore water Cd increased, much of that Cd appeared to end up on root plaque rather than translocating to shoots (as evidenced by higher plaque Cd in +S at filling, Figure 4b). This highlights a transfer of Cd among compartments: S application seemed to promote a transfer from soil solid–soil pore water (upon sulfide oxidation) onto root surfaces, rather than into the plant or back into easily exchangeable soil forms. Our conceptual model is that S initially “holds” Cd in insoluble forms in soil; when conditions shift, Cd is released but is then quickly captured by iron plaque, effectively shifting Cd from the soil to a biologically sequestered form on the root exterior. This dynamic transfer was especially evident during the filling stage in +S: soil Cd fractions showed a notable drop, pore water Cd spiked, and plaque Cd rose concurrently. In contrast, without S, more of the released Cd might remain in solution longer or be taken into the plant.
In summary, sulfur amendments reduced overall Cd availability in our system by fostering CdS precipitation and altering Cd partitioning to more sorbed forms. However, the effectiveness of S was modulated by rice growth stage and redox changes: under sustained reducing conditions, S kept Cd immobile, but a mid-season oxidation led to partial remobilization. Our findings reinforce that pe + pH is a master variable controlling Cd solubility: in our SEM analysis (Figure 5), we found that pe + pH itself did not directly predict Cd in solution, but through its effect on the S2−/SO42− ratio, it strongly influenced Cd dynamics. Low pe + pH (strongly reducing) directly promoted low pore water Cd via S2− formation, whereas higher pe + pH indirectly led to higher Cd via sulfide oxidation (and also possibly by reducing iron plaque formation efficiency).
Figure 5. Redundancy analysis of the influence of soil pore water properties on the distribution of Cd fractions at the tillering, booting, filling, and mature stages, respectively. Circle, rhizosphere soil with S supplement; Star, rhizosphere soil without S supplement; Square, non-rhizosphere soil with S supplement; Triangle, non-rhizosphere soil without S supplement.

4.2. Effects of Sulfur Supply on Cd Sequestration in Plaque at Different Growth Stages

The plaque coating rice roots is composed mainly of iron oxides associated with dichromate-reactive colloids, which can immobilize heavy metals via adsorption and/or co-precipitation [44]. In this study, the concentrations of ACA-extractable Fe and Cd in the iron plaque varied with both S application and rice growth stage (Figure 4). This agrees with previous reports showing that iron-plaque elemental composition changes across rice developmental stages [42] and is sensitive to S amendment. Compared with the −S treatment, S addition significantly decreased ACA-extractable Cd in the plaque at the booting stage (Figure 4b), indicating that S application hindered Cd sequestration by iron plaque. This effect may be partly attributed to the reduced ACA-extractable Fe under S addition (Figure 4a), because S application lowered pore water Fe concentrations, likely due to FeS precipitation triggered by sulfate reduction. Consistently, rhizospheric pore-water S2− was higher under S addition than under −S at the booting stage (Figure 1f). Previous studies suggest that S2− can be toxic to rice roots, potentially suppressing root oxygen release and the secretion of oxidizing compounds in the rhizosphere, thereby limiting Fe2+ oxidation. Meanwhile, accumulated S2− may further promote reductive dissolution of Fe(III) oxides on the root surface, contributing to plaque disassembly. At a broader scale, S application may also promote Cd redistribution within the soil matrix through CdS precipitation and/or co-precipitation of Cd2+ with FeS/FeS2, particularly under strongly reducing conditions [45]. Accordingly, we observed a higher Cd remaining in paddy soil under S addition than under −S at the booting stage (Figure 2). Moreover, pore water S showed a significant positive correlation with soil Cd fractions, further supporting a close linkage between sulfur transformations and Cd partitioning.
However, S addition significantly increased ACA-extractable Cd in the iron plaque at the filling stage (Figure 4b). This differs from previous reports suggesting that S promotes Cd accumulation on root surfaces mainly by increasing the amount of iron plaque [23]. In our study, S addition markedly decreased ACA-extractable Fe in the plaque (Figure 4a), indicating that the enhanced plaque Cd at filling was not simply driven by greater plaque mass. Instead, from booting to filling, the progressive increase in Eh likely oxidized previously formed sulfide precipitates (CdS and/or Cd co-precipitated with FeS/FeS2), releasing both Cd and sulfate (SO42−). This interpretation is supported by the concurrent increases in pore water Cd and SO42− at filling (Figure 1e). Under Cd stress, rice roots may also increase S demand to synthesize thiol-containing compounds (e.g., glutathione and phytochelatins) to detoxify Cd, which can enhance sulfate uptake and/or accumulation at the root surface [46]. Because sulfate can form aqueous complexes with Cd (e.g., CdSO40), it may facilitate Cd mobilization and transport toward the root surface, thereby increasing Cd sequestration onto the plaque during the filling stage [24].
Across the entire growing season, ACA-extractable Fe was positively correlated with ACA-extractable Cd (Figure 6), suggesting that Cd retention in the plaque is closely coupled to Fe deposition. This supports the view that iron plaque can act as a barrier that reduces Cd entry into root tissues by scavenging Cd at the root–soil interface [47]. Nonetheless, iron plaque has also been considered a potential reservoir that may later supply heavy metals to rice under changing redox conditions [16]. We further observed a significant positive correlation between ACA-extractable Fe and SO42− (Figure 6), implying that sulfate dynamics were linked to plaque formation. One possible mechanism is that SO42− promotes Fe2+ transport to the root surface via complexation and charge balance, thereby facilitating subsequent oxidation and precipitation of Fe(III) (hydr)oxides as plaque [41]. Overall, our results suggest that iron-plaque formation is directly governed by dissolved Fe2+/Mn2+ availability in pore water, rhizosphere Eh, and root oxygen loss (ROL) rate [48], while also being indirectly modulated by sulfur speciation and transformations [24,47].
Figure 6. Heat map of correlation analysis between S-redox related bacteria (SRB and SOB) and phytochemical properties in rhizosphere soil, pore water, and iron plaque under S application and without S application. * indicates significant correlation at p < 0.05.

4.3. Redox of S Induced by SRB/SOB Driving Transfer of Cd in the Rhizosphere

The transformation of sulfur between S2− and SO42− in flooded soils is largely a biological process, governed by the activities of SRB and SOB communities. Our results clearly show that changes in SRB/SOB abundance are linked with shifts in the S2−/SO42− ratio and, by extension, Cd precipitation or release [16].
During early growth (tillering to booting), conditions became strongly reducing (pe + pH dropped), enabling SRB to thrive. In both the −S and +S rhizospheres, we detected a rise in total SRB relative abundance by booting. However, with S addition, SRB were further stimulated—notably Geobacter and Desulfobacca bloomed at booting under +S (Figure 3a,b). These genera are known, efficient acetate-oxidizing SRB that reduce sulfate to sulfide. The high SRB activity in +S at booting corresponds to rapid sulfate consumption and accumulation of sulfide in pore water (Figure 1f). Our correlation analysis (Figure 6) indeed showed a positive correlation between the abundance of SRB and S2− concentration in the rhizosphere, as well as a positive correlation between the SRB/SOB ratio and S2−/SO42− ratio. This confirms that an active SRB population is a driver of high S2−/SO42−. In turn, this high S2−/SO42− at booting was directly linked to low pore water Cd—as evidenced by our SEM results, which indicated S2−/SO42− had a significant negative direct effect on pore water Cd (standardized path coefficient −0.65, p < 0.01). Thus, we can say SRB activity indirectly lowered Cd bioavailability by raising S2− levels and precipitating Cd [18,19]. Sulfur addition amplified this effect because it ensured ample sulfate substrate for SRB. In control soil, native sulfate might become limiting (our initial available S was only 16.5 mg/kg), whereas in +S we added 30 mg S/kg, so SRB had more electron acceptors to respire. The positive response of Geobacter and Desulfobacca in +S at booting is a strong indication that these SRB took advantage of the extra sulfate. In any case, throughout the season, the total SRB/SOB ratio can serve as an indicator of the redox of sulfur.
At the booting-to-filling transition, we saw a shift: Eh rose, and SOB activity surged, especially Rhodopseudomonas (Figure 3h). Rhodopseudomonas are versatile purple non-sulfur bacteria that can oxidize sulfide or thiosulfate under microaerobic/light conditions. The fact that their relative abundance spiked during filling in −S and also increased in +S, indicates an aerobic window occurred, enabling sulfide oxidation to sulfate. SOB use the sulfide produced earlier as an energy source, converting it back to sulfate. Our chemical data confirm that from booting to filling, S2− fell and SO42− rose (Figure 1e,f). The presence of oxygen (from roots) was the likely trigger, and the SOB were the agents facilitating the reaction beyond what abiotic oxidation did [17,20]. As SOB oxidized sulfide at filling, they created sulfate in the rhizosphere. Two consequences followed: (1) CdS and other metal sulfides were oxidatively dissolved, releasing Cd to pore water (hence the Cd rebound); (2) the newly produced sulfate could diffuse away or be taken up by roots. We saw evidence for root uptake: the accumulation of sulfate on root surfaces and in tissues at filling was proposed [24,47], and aligns with our observed sulfate pattern. SOB activity thus fed the plant’s sulfur assimilation pathway (as sulfate is the form plants uptake). It is intriguing that total SOB abundance correlated positively with pore water sulfate and somewhat with pore water Cd in our correlation heatmap (Figure 6). Specifically, our data indicated that pore water Cd had a positive correlation with SOB abundance and with sulfate concentration (p < 0.05) (Figure 7). This underscores that when SOB dominated, sulfide was being converted to sulfate, which coincided with Cd release.
After filling and into maturity, conditions in the rhizosphere likely trended back to more reducing as the plant’s activity waned (we observed Eh drop again from filling to maturity). At this stage, SRB regrouped—especially the spore-formers and other slower-growing genera increased in relative abundance in +S (Figure 3). The microbial data suggest that by maturity, the community had changed in composition compared to earlier, reflecting adaptation to the accumulated effects of a season of flooding and S amendment. The late dominance of certain SRB in +S implies that as long as some sulfate was around, SRB continued to reduce it—possibly targeting not only sulfate but any oxidized sulfur species, including those from plant inputs (e.g., root-derived sulfate or organosulfur). This late SRB activity could cement Cd into sulfide or stable complexes once more at the tail end of the season.
The structural equation model (SEM) we constructed (Figure 8) ties this all together: it showed that the S2−/SO42− ratio had direct negative effects on pore water Cd and also indirectly affected Cd in plaque via affecting pore water Cd and plaque Fe. The SEM highlighted that the S redox state is a central node influenced by SRB/SOB and influencing Cd outcomes. In the SEM, we also found that SRB abundance (total) correlated positively with S2− and negatively with pore water Cd, while SOB abundance (total) correlated positively with SO42− and pore water Cd (consistent with our discussion above). The ratio SRB/SOB was positively linked to plaque Cd in correlations, meaning when SRB were more dominant (thus more sulfide formation), eventually more Cd ended up in plaque—a seemingly paradoxical result that makes sense when considering the pathway: more SRB—more sulfide—more Cd precipitated early—more Cd later on plaque after partial oxidation.
Finally, it is worth noting a practical insight: while 16S rRNA gene sequencing gave us valuable indications of SRB and SOB trends, it does not confirm their functional activity. Future work could use qPCR or metagenomics to directly track functional sulfur metabolism genes, which would strengthen the causal links. Nevertheless, the coherence between our microbial relative abundance data and the chemical patterns strongly suggests that the SRB and SOB were key drivers, not just passengers, in the S–Cd system.
Figure 7. Heat map of correlation analysis among the phytochemical properties in rhizosphere soil, pore water, and iron plaque under S application and without S application. * indicates significant correlation at p < 0.05.
Figure 8. Structural equation model (SEM) describing the direct and indirect effects of the properties in rhizosphere soil on ACA extractable Cd in iron plaque and dissolved Cd in the soil pore water. *, ** indicate the significance of pathway at p < 0.05, 0.01, respectively.

5. Conclusions

This study shows that the rhizosphere sulfide-to-sulfate ratio (S2−/SO42−) is a key regulator of cadmium (Cd) transfer among soil solids, pore water, and root iron plaque in flooded rice systems. In a pot experiment with Cd-contaminated paddy soil, sulfur addition (+S; 30 mg S kg−1 as Na2SO4) consistently reduced dissolved Cd relative to the −S control, but the magnitude and pathways of Cd retention were strongly growth-stage-dependent. During early growth under strongly reducing conditions, +S increased rhizosphere S2− and lowered pe + pH to levels favorable for metal-sulfide stability, coinciding with lower pore water Cd and a shift in soil Cd from labile (exchangeable/carbonate) pools toward more stable Fe/Mn oxide-associated fractions. A pronounced mid-season “Cd rebound” occurred from booting to filling as rhizosphere Eh increased, consistent with partial oxidation of reduced sulfur species and transient Cd remobilization; however, +S attenuated this rebound (peak pore water Cd 40% lower than −S), indicating that sulfur buffering and/or rapid secondary capture limited Cd release. Root iron plaque functioned as a dynamic Cd sink: sulfur initially suppressed plaque Fe and plaque-bound Cd at booting, but enhanced plaque Cd sequestration during filling and maturity, contributing to lower soluble Cd during later stages. Microbial community shifts supported these redox-driven processes, with sulfur amendment favoring sulfate-reducing taxa at key stages and increasing sulfur-oxidizing taxa during the booting–filling transition; the SRB/SOB balance tracked S2−/SO42− and aligned with model results identifying this ratio as a direct control on dissolved Cd. Overall, moderate sulfur amendment can mitigate Cd availability in flooded rice soils by regulating sulfur redox and plaque-mediated capture, but effective application requires awareness of stage-dependent redox oscillations. Key limitations include the pot setting, a single S rate, 16S-based functional inference, and a lack of direct solid-phase confirmation of CdS, warranting future field trials and functional/mineralogical validation.

Author Contributions

Conceptualization, G.W. and X.J.; methodology, K.W.; software, K.W.; validation, K.W. and G.W.; formal analysis, X.J.; investigation, K.W.; resources, X.J.; data curation, G.W.; writing—original draft preparation, Y.L.; writing—review and editing, Y.L.; visualization, G.W.; supervision, X.J.; project administration, G.W.; funding acquisition, G.W. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the Open Research Fund of Key Laboratory for Lake Pollution Control of the Ministry of Ecology and Environment (2024HPYKFYB03).

Institutional Review Board Statement

Not applicable.

Data Availability Statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare no competing interests.

References

  1. Wang, P.; Chen, H.; Kopittke, P.M.; Zhao, F.J. Cadmium contamination in agricultural soils of China and the impact on food safety. Environ. Pollut. 2019, 249, 1038–1048. [Google Scholar] [CrossRef] [Scilit]
  2. Kosolsaksakul, P.; Farmer, J.G.; Oliver, I.W.; Graham, M.C. Geochemical associations and availability of cadmium (Cd) in a paddy field system, northwestern Thailand. Environ. Pollut. 2014, 187, 153–161. [Google Scholar] [CrossRef] [Scilit]
  3. Li, Z.; Liang, Y.; Hu, H.; Shaheen, S.M.; Zhong, H.; Tack, F.M.G.; Wu, M.; Li, Y.F.; Gao, Y.; Rinklebe, J.; et al. Speciation, transportation, and pathways of cadmium in soil-rice systems: A review on the environmental implications and remediation approaches for food safety. Environ. Int. 2021, 156, 106749. [Google Scholar] [CrossRef] [Scilit]
  4. Wang, C.C.; Zhang, Q.C.; Yan, C.A.; Tang, G.Y.; Zhang, M.Y.; Ma, L.Q.; Gu, R.H.; Xiang, P. Heavy metal(loid)s in agriculture soils, rice, and wheat across China: Status assessment and spatiotemporal analysis. Sci. Total Environ. 2023, 882, 163361. [Google Scholar]
  5. Lu, A.X.; Li, B.R.; Li, J.; Chen, W.; Xu, L. Heavy metals in paddy soil-rice systems of industrial and township areas from subtropical China: Levels, transfer and health risks. J. Geochem. Explor. 2018, 194, 210–217. [Google Scholar] [CrossRef] [Scilit]
  6. Zou, M.; Zhou, S.; Zhou, Y.; Jia, Z.; Guo, T.; Wang, J. Cadmium pollution of soil-rice ecosystems in rice cultivation dominated regions in China: A review. Environ. Pollut. 2021, 280, 116965. [Google Scholar] [CrossRef] [Scilit]
  7. Yu, L.; Zhu, J.; Huang, Q.; Su, D.; Jiang, R.; Li, H. Application of a rotation system to oilseed rape and rice fields in Cd-contaminated agricultural land to ensure food safety. Ecotoxicol. Environ. Saf. 2014, 108, 287–293. [Google Scholar] [CrossRef] [Scilit]
  8. Li, S.S.; Chen, S.B.; Wang, M.; Lei, X.; Zheng, H.; Sun, X.; Wang, L.; Han, Y. Iron fractions responsible for the variation of Cd bioavailability in paddy soil under variable pe + pH conditions. Chemosphere 2020, 251, 126355. [Google Scholar]
  9. Wang, Z.; Liu, W.; Liu, J.; Liu, X.; Liu, R.; Zhao, Y. Differences and mechanism of dynamic changes of Cd activity regulated by polymorphous sulfur in paddy soil. Chemosphere 2022, 291, 133055. [Google Scholar] [CrossRef] [Scilit]
  10. Liu, J.L.; Yu, Q.; Showalter, A.R.; Bunker, B.A.; Swanson, J.S.; Reed, D.; Rong, X.G.; Fein, J.B. Cadmium and proton adsorption onto a halophilic archaeal species: The role of cell envelope sulfhydryl sites. Geochim. Cosmochim. Acta 2020, 276, 186–197. [Google Scholar]
  11. Huang, H.; Ji, X.B.; Cheng, L.Y.; Zhao, F.J.; Wang, P. Free radicals produced from the oxidation of Ferrous sulfides promote the remobilization of Cadmium in paddy soil during drainage. Environ. Sci. Technol. 2021, 55, 9845–9853. [Google Scholar] [CrossRef] [Scilit]
  12. Fulda, B.; Voegelin, A.; Kretzschmar, R. Redox-controlled changes in Cadmium solubility and solid-phase speciation in a paddy soil as affected by reducible sulfate and copper. Environ. Sci. Technol. 2013, 47, 12775–12783. [Google Scholar] [CrossRef] [Scilit]
  13. Zhang, D.; Du, G.; Chen, S.G.; Rao, W.; Li, X.; Jiang, Y.; Liu, S.; Wang, D. Effect of elemental sulfur and gypsum application on the bioavailability and redistribution of cadmium during rice growth. Sci. Total Environ. 2019, 657, 1460–1467. [Google Scholar]
  14. Zhao, M.; Qian, E.; Zhang, F.; Liu, R.; Liu, X.; Zhao, Y.; Liang, X. Spatiotemporal dynamics of labile Cd in soil during rice growth. Sci. Total Environ. 2020, 738, 139832. [Google Scholar] [CrossRef] [Scilit]
  15. Yang, F.; Zhai, W.; Li, Z.; Huang, Y.; Manzoor, M.; Yang, B.; Hou, Y.; Lei, L.; Tang, X. Immobilization of lead and cadmium in agricultural soil by bioelectrochemical reduction of sulfate in underground water. Chem. Eng. J. 2021, 422, 130010. [Google Scholar] [CrossRef] [Scilit]
  16. Zhang, D.; Gong, Y.; Qin, R.; Zaib, S.; Shang, K.; Xia, J.; Zhou, L.; Xie, Y. Effect and mechanism of Rice-Solanum nigrum L. rotation system inoculated with sulfate-reducing bacteria (SRB) and sulfur-oxidizing bacteria (SOB) for the remediation of Cd-contaminated paddy soil. J. Hazard. Mater. 2025, 496, 139540. [Google Scholar]
  17. Wang, G.X.; Hu, Z.Y.; Li, S.Y.; Wang, Y.; Sun, X.L.; Zhang, X.R.; Li, M. Sulfur controlled cadmium dissolution in pore water of cadmium-contaminated soil as affected by DOC under waterlogging. Chemosphere 2020, 240, 124846. [Google Scholar] [CrossRef] [Scilit]
  18. Demin, K.A.; Prazdnova, E.V.; Minkina, T.M.; Gorovtsov, A.V. Sulfate-Reducing Bacteria Unearthed: Ecological Functions of the Diverse Prokaryotic Group in Terrestrial Environments. Appl. Environ. Microbiol. 2024, 90, e01390-23. [Google Scholar] [CrossRef] [Scilit]
  19. Ghosh, W.; Dam, B. Biochemistry and Molecular Biology of Lithotrophic Sulfur Oxidation by Taxonomically and Ecologically Diverse Bacteria and Archaea. FEMS Microbiol. Rev. 2009, 33, 999–1043. [Google Scholar] [CrossRef] [Scilit]
  20. Huang, H.; Lv, Y.W.; Tian, K.K.; Shen, Y.; Zhu, Y.L.; Lu, H.Y.; Li, R.H.; Han, J.G. Influence of sulfate reducing bacteria cultured from the paddy soil on the solubility and redox behavior of Cd in a polymetallic system. Sci. Total Environ. 2023, 901, 166369. [Google Scholar] [CrossRef] [Scilit]
  21. Maisch, M.; Lueder, U.; Kappler, A.; Schmidt, C. From Plant to Paddy—How Rice Root Iron Plaque Can Affect the Paddy Field Iron Cycling. Soil Syst. 2020, 4, 28. [Google Scholar] [CrossRef] [Scilit]
  22. Maisch, M.; Lueder, U.; Kappler, A.; Schmidt, C. Iron lung: How rice roots induce iron redox changes in the rhizosphere and create niches for microaerophilic Fe (II)-oxidizing bacteria. Environ. Sci. Technol. Let. 2019, 6, 600–605. [Google Scholar] [CrossRef] [Scilit]
  23. Li, H.B.; Zheng, X.W.; Tao, L.X.; Gao, L.; Xiong, J. Aeration increases cadmium (Cd) retention by enhancing iron plaque formation and regulating pectin synthesis in the roots of rice (Oryza sativa) seedlings. Rice 2019, 12, 28. [Google Scholar] [CrossRef] [Scilit]
  24. Cao, Z.Z.; Qin, M.L.; Lin, X.Y.; Zhu, Z.W.; Chen, M.X. Sulfur supply reduces cadmium uptake and translocation in rice grains (Oryza sativa L.) by enhancing iron plaque formation, cadmium chelation and vacuolar sequestration. Environ. Pollut. 2018, 238, 76–84. [Google Scholar] [CrossRef] [Scilit]
  25. Zhang, X.X.; Zhang, X.X.; Zheng, Y.J.; Wang, R.P.; Chen, N.C.; Lu, P.X. Accumulation of S, Fe and Cd in rhizosphere of rice and their uptake in rice with different water managements. Environ. Sci. 2013, 34, 2837–2846. [Google Scholar]
  26. Sun, L.; Song, K.; Shi, L.; Duan, D.; Zhang, H.; Sun, Y.; Qin, Q.; Xue, Y. Influence of elemental sulfur on cadmium bioavailability, microbial community in paddy soil and Cd accumulation in rice plants. Sci. Rep. 2021, 11, 11468. [Google Scholar] [CrossRef] [Scilit]
  27. Zhang, F.; Peng, D.; Liu, L.; Jiang, H.; Bai, L. Cultivar-dependent rhizobacteria community and cadmium accumulation in rice: Effects on cadmium availability in soils and iron-plaque formation. J. Environ. Sci. 2022, 116, 90–102. [Google Scholar] [CrossRef] [Scilit]
  28. Wei, T.; Liu, X.; Dong, M.; Lv, X.; Hua, L.; Jia, H.; Ren, X.; Yu, S.; Guo, J.; Li, Y. Rhizosphere iron and manganese-oxidizing bacteria stimulate root iron plaque formation and regulate Cd uptake of rice plants (Oryza sativa L.). J. Environ. Manag. 2021, 278, 111533. [Google Scholar] [CrossRef] [Scilit]
  29. Balasubramanian, S.; Pugalenthi, V. A comparative study of the determination of sulphide in tannery waste water by ion selective electrode (ISE) and iodimetry. Water. Res. 2000, 34, 4201–4206. [Google Scholar] [CrossRef] [Scilit]
  30. Álvarez, S.M.; Llamas, N.E.; Álvarez, M.B.; Marcovecchio, J.E.; Garrido, M.; Domini, C.E. Rapid determination of total metals: Synergic effect of ultrasound energy and ionic liquids on the digestion of sediment samples. J. Soil. Sediment. 2021, 21, 3692–3703. [Google Scholar] [CrossRef] [Scilit]
  31. Ma, P.; Tian, T.; Dai, Z.; Shao, T.; Zhang, W.; Liu, M. Assessment of Cd bioavailability using chemical extraction methods, DGT, and biological indicators in soils with different aging times. Chemosphere 2022, 296, 133931. [Google Scholar] [CrossRef] [Scilit]
  32. Bettiol, A.C.T.; Brasos, L.B.; Lopes, I.G.; Andriolli, I.A.; Ferreira, M.E.; da Cruz, C.P. Evaluation of potentially available nitrogen by biological and chemical methods in soil cultivated with maize in succession to cover crops. J. Plant Nutr. 2022, 45, 1919–1932. [Google Scholar] [CrossRef] [Scilit]
  33. Tessier, A.; Campbell, P.G.C.; Bisson, M. Sequential extraction procedure for the speciation of particulate trace metals. Anal. Chem. 1979, 51, 844–851. [Google Scholar] [CrossRef] [Scilit]
  34. Hou, J.Y.; Wu, L.H.; Liu, W.X.; Ge, Y.Y.; Mu, T.T.; Zhou, T.; Li, Z.; Zhou, J.W.; Sun, X.; Luo, Y.M.; et al. Biogeography and diversity patterns of abundant and rare bacterial communities in rice paddy soils across China. Sci. Total Environ. 2020, 730, 139116. [Google Scholar] [CrossRef] [Scilit]
  35. Quast, C.; Pruesse, E.; Yilmaz, P.; Gerken, J.; Schweer, T.; Yarza, P.; Peplies, J.; Glöckner, F.O. The SILVA ribosomal RNA gene database project: Improved data processing and web-based tools. Nucleic Acids Res. 2013, 41, 590–596. [Google Scholar] [CrossRef] [Scilit]
  36. Gao, M.X.; Hu, Z.Y.; Wang, G.D.; Xia, X. Effect of elemental sulfur supply on cadmium uptake into rice seedlings when cultivated in low and excess cadmium soils. Commun. Soil Sci. Plant Anal. 2010, 41, 990–1003. [Google Scholar] [CrossRef] [Scilit]
  37. Taylor, G.J.; Crowder, A.A. Uptake and accumulation of heavy metals by Typha latifolia in wetlands of the Sudbury, Ontario region. Can. J. Bot. 1983, 61, 63–73. [Google Scholar] [CrossRef] [Scilit]
  38. Zhou, H.; Zeng, M.; Zhou, X.; Liao, B.H.; Peng, P.Q.; Hu, M.; Zhu, W.; Wu, Y.J.; Zou, Z.J. Heavy metal translocation and accumulation in iron plaques and plant tissues for 32 hybrid rice (Oryza sativa L.) cultivars. Plant Soil 2015, 386, 317–329. [Google Scholar] [CrossRef] [Scilit]
  39. Liu, Z.; Wang, Q.Q.; Huang, S.Y.; Kong, L.X.; Zhuang, Z.; Wang, Q.; Li, H.F.; Wan, Y.N. The risks of sulfur addition on cadmium accumulation in paddy rice under different water-management conditions. J. Environ. Sci. 2022, 118, 101–111. [Google Scholar] [CrossRef] [Scilit]
  40. Fortin, D.; Beveridge, T.J. Microbial sulfate reduction within sulfidic mine tailings: Formation of diagenetic Fe sulfides. Geomicrobiol. J. 1997, 14, 1–21. [Google Scholar] [CrossRef] [Scilit]
  41. Zhang, Q.; Chen, H.F.; Huang, D.Y.; Guo, X.B.; Xu, C.; Zhu, H.H.; Li, B.; Liu, T.T.; Feng, R.W.; Zhu, Q.H. Sulfur fertilization integrated with soil redox conditions reduces Cd accumulation in rice through microbial induced Cd immobilization. Sci. Total Environ. 2022, 824, 153868. [Google Scholar] [CrossRef] [Scilit]
  42. Huang, G.; Ding, C.; Guo, F.; Zhang, T.; Wang, X. The optimum Se application time for reducing Cd uptake by rice (Oryza sativa L.) and its mechanism. Plant Soil 2018, 431, 231–243. [Google Scholar] [CrossRef] [Scilit]
  43. Zhang, X.; Zhang, X.; Lv, S.; Shi, L.; Wang, R. Migration and transformation of cadmium in rice—Soil under different nitrogen sources in polymetallic sulfide mining areas. Sci. Rep. 2020, 10, 2418. [Google Scholar] [CrossRef] [Scilit]
  44. Khan, N.; Seshadri, B.; Bolan, N.; Saint, C.P.; Kirkham, M.B.; Chowdhury, S.; Yamaguchi, N.; Lee, D.Y.; Li, G.; Kunhikrishnan, A.; et al. Root Iron Plaque on Wetland Plants as a Dynamic Pool of Nutrients and Contaminants. Adv. Agron. 2016, 138, 1–96. [Google Scholar]
  45. Weber, F.A.; Voegelin, A.; Kretzschmar, R. Multi-metal contaminant dynamics in temporarily flooded soil under sulfate limitation. Geochim. Cosmochim. Acta 2009, 73, 5513–5527. [Google Scholar] [CrossRef] [Scilit]
  46. Wu, Z.Y.; Zhang, C.H.; Yan, J.L.; Yue, Q.; Ge, Y. Effects of sulfur supply and hydrogen peroxide pretreatment on the responses by rice under cadmium stress. J. Plant Growth Regul. 2015, 77, 299–306. [Google Scholar] [CrossRef] [Scilit]
  47. Greipsson, S. Effects of iron plaque on roots of rice on growth and metal concentration of seeds and plant tissues when cultivated in excess copper. Commun. Soil Sci. Plan. 1994, 25, 2761–2769. [Google Scholar] [CrossRef] [Scilit]
  48. Wang, J.; Yu, L.; Qin, L.Y.; Sun, X.Y.; Zhou, W.N.; Wang, M.; Chen, S.B. Low pe + pH inhibits Cd transfer from paddy soil to rice tissues driven by S addition. Chemosphere 2023, 335, 139126. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.