Can simultaneous immobilization of arsenic and cadmium in paddy soils be achieved by liming?

Liming acidic paddy soils to near-neutral pH is the most cost-effective strategy to minimize cadmium (Cd) accumulation by rice. However, the liming-induced effect on arsenic (As) (im)mobilization remains controversial and is called upon for further investigation, particularly for the safe utilization of paddy soils co-contaminated with As and Cd. Here, we explored As and Cd dissolution along pH gradients in flooded paddy soils and extracted key factors accounting for their release discrepancy with liming. The minimum As and Cd dissolution occurred concurrently at pH 6.5–7.0 in an acidic paddy soil (LY). In contrast, As release was minimized at pH < 6 in the other two acidic soils (CZ and XX), while the minimum Cd release still appeared at pH 6.5–7.0. Such a discrepancy was determined largely by the relative availability of Fe under overwhelming competition from dissolved organic carbon (DOC). A mole ratio of porewater Fe/DOC at pH 6.5–7.0 is suggested as a key indicator of whether co-immobilization of As and Cd can occur in flooded paddy soils with liming. In general, a high mole ratio of porewater Fe/DOC (≥ 0.23 in LY) at pH 6.5–7.0 can endow co-immobilization of As and Cd, regardless of Fe supplement, whereas such a case is not in the other two soils with lower Fe/DOC mole ratios (0.01–0.03 in CZ and XX). Taking the example of LY, the introduction of ferrihydrite promoted the transformation of metastable As and Cd fractions to more stable ones in the soil during 35 days of flooded incubation, thus meeting a class I soil for safe rice production. This study demonstrates that the porewater Fe/DOC mole ratio can indicate a liming-induced effect on co-(im)mobilization of As and Cd in typical acidic paddy soils, providing new insights into the applicability of liming practice for the paddy soils.


Introduction
Paddy soils confer global production of rice and rice-based cereals (Zou et al. 2021). From both geologic and anthropogenic sources, arsenic (As) and cadmium (Cd) accumulation in paddy soils has occurred widely (Khanam et al. 2020;Xue et al. 2017). Being class I carcinogens, As and Cd preferably transfer from paddy soils to rice and further to humans along the food chain, which poses severe public health 1 3 risks, e.g., cardiovascular diseases, diabetes, physiological disorder, itai-itai disease, and cancers (Luo et al. 2020;Lv et al. 2020;Rinklebe et al. 2019). Particularly, in densely populated Asia with limited reserved paddy resources (Mu et al. 2019;Shahriar et al. 2020), co-contamination of As and Cd in paddy soils has jeopardized the quality and productivity of rice (Mawia et al. 2021;Zhao and Wang 2020), potentially aggravating global food crisis.
To mitigate As and Cd accumulation by rice and reduce their exposure risk, it is critical to lower As and Cd bioavailability in paddy soils . In intensive farming systems, soil acidification as a result of long-term ammonium-based fertilization as well as acid precipitation has largely promoted Cd mobilization (Guo et al. 2018;Meng et al. 2018;Zhao et al. 2018). For Cd-contaminated paddy soils, liming has been widely applied to mitigate Cd bioavailability to rice due to its low cost and high acidneutralizing capacity (Huang et al. 2021;Yang et al. 2018). With liming, soil Cd partitioning tends to transform from soluble and easily exchangeable fractions into carbonateand Fe/Mn oxide-bound fractions which are nonlabile at elevated soil pH (Huang et al. 2022b;Kong et al. 2021;Zhang et al. 2018). As summarized in Fig. 1a, Cd solubility in a range of paddy soils from China, Japan, Czech, India, and Bangladesh shows a consistent decline with increasing soil pH in a strong linear pattern (r = −0.721, p < 0.05).
For paddy soils with As and Cd co-contamination, however, liming has been reported to increase As bioavailability due to enhanced competition of OH − toward anion sorption sites on iron (Fe) (hydr)oxides (Li et al. 2021b), the most important As-hosting constituent in paddy soils (Bandara et al. 2021;Yao et al. 2022). On the other hand, for each unit increase in pH, Fe 3+ activity in soil porewater tends to decrease by 3 orders of magnitude (Lindsay 1988), which is expected to favor As capture from porewater and thus enhance As immobilization into soil solid phase . The two contradicting behaviors that porewater pH regulates soil As mobility can be reflected when porewater As concentration in paddy soils is plotted against porewater pH (Fig. 1b). In general, at porewater pH below ~6.5, soil As release decreases with increasing pH, whereas a steep elevation in porewater As occurs at pH > 7.0. This fitted relationship suggests that liming could co-mitigate Cd and As release in paddy 1 3 soils with increasing pH toward a near-neutral condition. Despite this, geochemical processes of As and Cd in paddy soils can be highly affected by other important properties, such as soil organic matter (as a strong competitor for As by binding with available Fe) besides porewater Fe and pH. This begs an important and open question: Can simultaneous immobilization of As and Cd in paddy soils be achieved by liming?.
Herein, we, if it's the case, further targeted two specific questions: (1) under which condition can As and Cd immobilization in paddy soil by liming be maximized concurrently? and (2) under which condition is As and Cd co-contaminated paddy soil in need of exogenous Fe amendment along with liming? We therefore investigated the changes in porewater As, Cd, and Fe and closely relevant chemical parameters in three acidic paddy soils along pH gradients with liming. The key determinants of soil As and Cd release during the liming process were identified and analyzed in detail. Then, the effectiveness of exogenous Fe, as represented by ferrihydrite, in strengthening As and Cd co-immobilization was explored under liming conditions. Our findings are expected to provide new insights into the applicability of lime in comitigating As and Cd bioavailability in paddy soils with and without exogenous Fe supplement.

Paddy soils preparation
Paddy soils with Cd and As co-contamination were collected from Liuyang (LY), Chenzhou (CZ), and Xiangxi (XX) in Hunan province, South Central China (Table 1). The soil samples were naturally air-dried and passed through a 2-mm sieve. Soil pH was measured at a ratio of 1 : 2.5 (soil to deionized water (DIW), w/v) using an ORP depolarization automatic analyzer (FJA-6, Nanjing, China). To determine total As and Cd, aliquots (0.25 g) of the soil samples were digested using a microwave digestion oven (CEM MARS 6, Matthews, NC, USA) according to EPA 3051a with a slightly modified heating program (Yan et al. 2021).
Soil organic matter (SOM) content was determined by the potassium dichromate oxidation volumetric method ).

Soil incubation experiments
Expt. 1: identifying potentials of soil Fe for As and Cd co-immobilization by adjusting pH from 4 to 9 In a typical procedure, 300 g of soil sample was mixed with 300 mL of DIW to prepare a soil slurry, and its pH was adjusted to desired values (4.0-9.0) by adding 0.1 or 1 M HCl or NaOH dropwise. The resulting soil slurry was incubated at room temperature for 1 week while performing a daily pH measurement and re-adjustment to sustain the preset values if necessary. The soil was then flooded for another week with a 2-3 cm depth overlying the water layer to mimic the flooded paddy condition (Lin et al. 2018), during which porewater pH was kept constant. At the end of flooding, soil porewater was sampled for chemical analysis using porewater samplers (Rhizon MOM-19.21.21F, Rhizosphere Research Products, Wageningen, Netherlands). All the soils studied in this work were performed in triplicates.

Expt. 2: screening of Fe-based materials for As and Cd co-immobilization with lime in CZ and XX
To screen Fe-based materials with relatively high immobilizing capacity for As and Cd, three Fe-based amendments, i.e., ferric oxide (Fe 2 O 3 ), zero-valent iron (ZVI), and ferrihydrite (Fh), were applied in paddy soils (CZ and XX). Specifically, Fh was synthesized in our lab using Fe(NO 3 ) 3 •9H 2 O and NaOH according to a previous protocol (Schwertmann and Cornell 2008). Briefly, Fe(NO 3 ) 3 •9H 2 O was dissolved in DIW, and then, 1 M NaOH was carefully introduced to raise the solution pH to 7.0-8.0 under vigorous stirring condition. After that, the suspension was centrifuged and washed with anhydrous alcohol (70%), air-dried, passed through a 0.15-mm sieve, and stored at room temperature in the dark. The final Fh product had a pH value of 6.3 and pH pzc of 8.0.
Fixing at soil pH 6.5, 300 g of air-dried soil was mixed with 300 mL of DIW, followed by the addition of lime (CaO) and/or Fe amendment, as summarized in Table 2. Then, the soil was flooded for 40 days, with a 2-3 cm flooding water depth. At 5-day intervals, soil pH was measured, and porewater (ca. 15 mL) was collected throughout the flooding period. All the treatments were conducted in triplicates.

Expt. 3: applying exogenous Fh for further decreasing As and Cd in LY with lime
The selected LY was subject to co-treatments of CaO with Fh at an increasing addition ratio ( Table 2). The soil was Table 1 Basic physicochemical properties of three tested paddy soils Note: LY, CZ, and XX refer to the paddy soils from Liuyang, Chenzhou, and Xiangxi in Hunan province, respectively; SOM stands for soil organic matter. Screening value of soil pollution risk (GB 15618-2018): pH ≤ 5.5, Cd = 0.3 mg/kg, As = 30 mg/kg; 5.5 < pH ≤ 6.5, Cd = 0.4 mg/kg, As = 30 mg/kg Soils pH Cd (mg/kg) As (mg/kg) Fe (g/kg) SOM (g/kg) then flooded for 35 days. Soil pH and Eh measurements and porewater collection were conducted weekly. At the end of flooding, fractionation of soil As and Cd was carried out using sequential extraction procedures according to Tessier et al. (1979) and Wenzel et al. (2001) (Table S1).

DGT deployment
To visualize the two-dimensional changes of bioavailable As and Cd in paddy soils under different treatments, the diffusive gradients in thin films (DGT) technique with a ZrO-Chelex binding gel was employed (60 mm × 80 mm, EasySensor, Nanjing, China). The sampling time was set as the 48th h, and ambient temperature was measured every 6 h to calculate As and Cd diffusion rate (Zhang et al. 2014). After retrieval, the ZrO-Chelex gels were rinsed thoroughly with DI water and cut carefully into 10 mm × 10 mm pieces. The ZrO-Chelex gels were first used for the Cd extraction with 1 M HNO 3 for 16 h, followed by a 2-h soaking in DIW and eventually used for the As extraction with 1 M NaOH for 24 h (Ding et al. 2016;Wang et al. 2017).

Chemical analysis
Aqueous As concentration was determined using an atomic fluorescence spectrometer (AFS, Haiguang 6500, China), whereas Cd was analyzed by an atomic absorption spectrophotometer (AAS, PerkinElmer AAnalyst 900T). Total Fe and Fe(II) in solution samples were measured with a 1,10-phenanthroline colorimetric method using an UV-visible spectrophotometer (Evolution 260 Bio, Thermo Fisher Scientific, USA) (Amonette and Charles Templeton 1998). Dissolved organic carbon (DOC) was analyzed using a total organic carbon analyzer (TOC-Vcph, Shimadzu, Japan). Spatial distributions of As, Cd, and Fe in LY with and without Fh addition were examined using a field emission scanning electron microscopy (SEM, Hitachi, SU8020) equipped with energy dispersive X-ray spectroscopy (EDS, HORIBA EX-250, Japan).

Statistical analysis and QA/QC
All the data were presented as average values with standard deviation (n = 3) and plotted with OriginPro 2018. Pearson's correlation analysis was performed in SPSS 25.0. The differences between treatments were determined with analysis of variance (ANOVA) and Duncan's multiple range test with p < 0.05 adopted as the criterion of significance (SPSS 25.0). A digestion blank and a standard soil sample (GBW10010, i.e., GSB-1, from the National Standard Material Center) were included in the digestion of every 18 samples. The linear correlation coefficient for all calibration curves was ≥ 0.999 (n = 6) for both As and Cd measurements. An analytical blank, a calibration verification standard, and a continuing calibration standard were also included in every 20 samples during analysis. The limits of detection are 0.02 μg/L for As and 0.001 μg/L for Cd.

As and Cd dissolution along soil pH gradient and identification of key controlling factors
The pH values of three tested paddy soils ranged from 5.0 to 6.0 (Table 1), showing acidic to weakly acidic soil conditions. Soil pH plays a key role in controlling both As and Cd release in paddy soils (Fan et al. 2020;Honma et al. 2016). As shown in Fig. 1b, a "V"-shape change in porewater As concentration as a function of pH was observed, with the turning point at pH 6.0-7.0. Meanwhile, porewater Cd in paddy soils exhibited a monotonous decline with pH increasing from 5.5 to 7.3 (Fig. 1a). Both porewater As and Cd exhibited a positive linear decrease as soluble Fe declined when porewater pH increased (Fig. 1c, d). On the basis of this, we hypothesize that increasing soil pH value (e.g., liming-induced pH effect) can favor As as well as Cd immobilization in acidic paddy soils, most likely by enhancing the precipitation of Fe (hydr)oxides which functions as As-and Cd-hosting solid phase. To verify the above hypothesis, As and Cd dissolution over a porewater pH range of 4.0-9.0 was investigated with three paddy soils (i.e., LY, CZ, and XX). For the three paddy soils, the lowest porewater As concentrations were found in three different pH ranges, while porewater Cd concentrations reached their minimum values at a fixed pH of 6.5-7.0 identified as the optimal pH condition in most paddy soils for minimizing Cd availability to rice (Fig. 2a, c, e) Du et al. 2018).
Specifically, porewater As and Cd concentrations in LY developed "V"-shaped curves as pH increased from 4.3 to 8.4 (Fig. 2a) and remained at their lowest levels at pH ~6.5-7.0. This indicates that As and Cd release in LY can be simultaneously minimized in the optimal pH range of ~6.5-7.0. At pH < 6.6, porewater As concentration decreased from 607 to 159 μg/L, and porewater Cd concentration decreased from 233 to 0 μg/L. Meanwhile, we observed a dramatical decline in porewater Fe concentration from 1133 to 271 mg/L (Fig. 2b), whereas porewater DOC concentration first decreased from 374 to 223 mg C/L and then increased to 247 mg C/L. In such a paddy soil, three types of components, i.e., Fe-based components, organic carbon (OC), and H + (or OH − ), may participate in regulating As and Cd (im)mobilization via different mechanisms as below: (1) Fe (hydr)oxides can sequestrate As and Cd via adsorption, thus leading to As and Cd immobilization; (2) OC have two forms, DOC and solid OC. DOC not only can compete with As and Cd toward Fe-based binding sites, but also can complex with As and Cd, thereby triggering As and Cd desorption. However, solid OC can immobilize As and Cd via its complexation, but can compete with As and Cd toward Fe-based binding sites to suppress As and Cd immobilization; (3) elevated OH − can favor As desorption via its competition with As but Cd immobilization via its precipitation and weakened competition with H + . Based on the results and analysis, we suggest that an enhanced Error bars represent the standard deviations of three replicates precipitation of soluble Fe (with pH increasing) plays a dominant role in capturing porewater As below pH 7.0. This is further supported by a relatively high porewater Fe/DOC mole ratio that indicates considerable Fe (hydr)oxides available for As immobilization (Fig. 2g). For increasing Cd immobilization, both elevated Fe (hydr)oxides and OH − (or decreased H + ) should be primarily responsible. When pH was further increased to 8.4, porewater As rebounded to 375 μg/L, and a slight rebound of soluble Cd also occurred. At the same time, porewater Fe concentration further decreased to 51 mg/L and then rebounded to 175 mg/L, which is not consistent with the changes in porewater As and Cd concentrations. This indicates that Fe (hydr)oxides should not be primarily responsible for As and Cd liberation above pH 7.0. Interestingly, we found that DOC concentration further increased to 1401 mg C/L (Fig. 2b), which agrees with the changes in porewater As and Cd concentrations. Furthermore, we found a significant correlation between DOC and porewater As and Cd at pH > 7.0 (Fig. S1c, r = 0.663 for As, r = 0.863 for Cd, p < 0.05). In such a case, more DOC can trigger As and Cd desorption (Hussain et al. 2021;Wang et al. 2019). In other words, DOC plays a main role in As and Cd liberation above pH 7.0. Overall, As and Cd release in LY can be co-minimized at the pH 6.5-7.0, where the precipitation of soil endogenous Fe approached its maximum binding capacities of As and Cd. At lower pH values, Fe (hydr)oxides from porewater Fe at increasing pH can serve as the main scavenger for soluble As and Cd. Beyond this threshold pH range, DOC concentration exhibited a substantial increase and reached up to 4 orders of magnitude higher than porewater As and Cd, thereby functioning as an overwhelming complexing agent toward As, Cd, and Fe (hydr)oxides (the most important As-and Cd-hosting phase in paddy soils).
In stark contrast, the lowest porewater As concentrations in CZ and XX occurred at pH 4.7-5.9 and 3.9-5.0, respectively, which were approximately 1-2 units lower than the optimal pH 6.5-7.0 for Cd immobilization (Fig. 2c, e). When pH was increased from the threshold pH value to ~7.0, the porewater As concentrations rebounded, whereas porewater Fe concentration decreased to the minimum value (Fig. 2d, f). This result suggests that Fe content is not the sole determinant of soil As-binding capacity. As previously discussed, DOC may play important roles in such processes; we therefore monitored DOC concentrations in CZ and XX. As shown in Fig. 2d, f, we observed a similar DOC trend in CZ and XX at pH 4.0-9.0, which is consistent with that in LY (Fig. 2b). It is worth noting that the highest As immobilization in CZ and XX appeared at pH ~5, corresponding to a porewater Fe/DOC mole ratio of 0.12-0.15. However, the mole ratio of porewater Fe/DOC dramatically decreased to 0.01-0.03 when increasing pH to 6.0-7.0, which is nearly one order of magnitude lower than that (0.23) in LY (Fig. 2g). This indicates that DOC-triggered As desorption level is beyond the Fe (hydr)oxide-driven As immobilization level at pH 5.0-7.0. Furthermore, DOC plays an increasing role in the As liberation as pH elevated to ~8.0, due to a significant increase of DOC concentration. With respect to the change in porewater Cd concentration in CZ and XX, it followed the same trend as that in LY. Therefore, both elevated Fe (hydr)oxides and OH − (or decreased H + ) are mainly responsible for Cd immobilization below pH ~7.0, while DOC primarily accounts for Cd liberation above 7.0 (r = 0.863, p < 0.05). We also noted that the mole ratios of porewater Fe/DOC in all the tested paddy soils were at least 6 orders of magnitude lower than that of Fe/As above the pH range of 4.0-9.0 ( Fig. 2g-i), which highlights the importance of DOC in regulating As and Cd (im)mobilization. Overall, our results demonstrate multiple mechanisms mentioned above that determine As and Cd (im)mobilization in the three paddy soils over a wide pH range of 4.0-9.0.

Comparison of Fe-based materials for As and Cd co-immobilization
To achieve simultaneously maximum As and Cd co-immobilization in CZ and XX (with limited available Fe due to DOC competition) at porewater pH ≥ 6.5 by liming, three Fe-based materials, i.e., Fe 2 O 3 , ZVI, and Fh, were supplemented to increase final total Fe content in the soils equivalent to that (30.19 g Fe/kg) in LY. With the CaO treatment alone, both soil pH values increased to 6.5 due to the production of OH − via hydrolysis of CaO while increasing As dissolution but lowering Cd dissolution in porewater (Fig. 3). This agrees with our previous observation that NaOHinduced pH effect is responsible for As liberation and Cd immobilization (Fig. 2c, e). Besides OH − , lime hydrolysis can provide soil Ca. On one hand, Ca can co-precipitate with As particularly under alkaline conditions; on the other hand, Ca can bind with DOC to suppress As and Fe binding with DOC. As a result, supplying Ca via lime hydrolysis can improve As immobilization below pH 7.0. In contrast, supplying Ca can lead to Cd liberation via its competition with Cd toward Fe-based binding sites below pH 7.0. Such supplying Ca should be against our observation on As liberation and Cd immobilization after liming, indicating an effect of Ca may be negligible. Hence, the effect of Ca should not be considered in the following discussion. In comparison, Fh exhibited the most prominent immobilizing capability, decreasing porewater As and Cd by 82.5% and 100%. This result agrees with the extraordinary binding capability of Fh toward As and Cd, as found in various environmental media (Fan et al. 2021;Fu et al. 2021;Li et al. 2021a;Ouyang et al. 2021).
It is noteworthy that the Fh addition alone showed a better controlling effect on As release, as compared with CaO+Fh in XX (Fig. 3c), which could result from the lower porewater pH with Fh alone than with CaO+Fh (pH 7.0 vs. 7.3) (Fig. 3d). As shown in Fig. 2a, porewater As maintained its lowest concentrations at pH 6-7 and started a sharp increase from pH 7 to 8 at 236 μg/L/unit. Therefore, these results confirm that pH above 7 is unfavorable for soil As immobilization.
We further checked the dynamic effects of Fh and CaO+Fh on As and Cd immobilization in CZ and XX during an extended flooding period (40 days, Fig. S2). In both treatments, the porewater As concentrations remained at 8.0 μg/L in CZ and 1.4 μg/L in XX, which were lower than those in CK (16.5-85.2 μg/L in CZ and 0.8-8.0 μg/L in XX). Both treatments led to slightly lower or comparable porewater Cd concentration under a flooded condition, as compared with CK. In spite of this, a striking difference in soil pH between liming and CK tends to appear during paddy drainage (Du et al. 2018;Wang et al. 2022), especially in grain filling stage which accounts for 98% of grain-Cd accumulation (Huang et al. 2022a).

Further immobilizing active As and Cd in LY with Fh during liming
Despite a relatively high Fe availability in LY, the lowest porewater As (159.1 μg/L) concentration at the optimal pH (~6.5) remained high (Fig. 2a), causing damages to public health through the food chain. Based on the lime calculation, a varied amount (0.25%, 0.5%, and 1%) of Fh was applied with CaO addition, elevating porewater pH to 6.4-7.1 over a 35-day period of flooding. With increasing the Fh addition, porewater As concentration showed a proportional decrease, and a similar trend was also determined with porewater Cd (Fig. 4a, b). Particularly, with the treatment of CaO+1%Fh, porewater As and Cd concentrations decreased by ~79% and ~51%, respectively. The resulting porewater As was 15.7-27.5 μg/L, which was lower than that of a class I paddy soil under the same flooded condition (19.7-99.1 μg/L, collected downstream of LY). This prominent inhibition of soil As release by CaO+1%Fh can be ascribed to larger Fh supplement and lower porewater pH within 6.5-7.0 (or a higher Eh) under a flooded condition (Fig. 4c, e). For example, with Fh addition increasing from 0.25 to 1%, porewater pH exhibited a proportional decline from 7.0 to 6.8, which was accompanied by an elevated Eh from the 14th day onwards during flooding (Fig. 4e). Correspondingly, the average As release into porewater decreased by 58% and 83% when Fh addition was 0.5% and 1%, respectively, relative to that with the 0.25% Fh. This suggests that liming to increase porewater pH to > 6.5 tends to negatively impact As immobilization, primarily due to the dissolution of Fe (hydr)oxides (Fig. 4d, f) as soil Eh decreased (Fig. 4e). A similar trend was also observed Fig. 3 Effects of CaO, CaO+Fe 2 O 3 , CaO+ZVI, CaO+Fh, and Fh on porewater As, Cd, and pH for CZ (a, b) and XX (c, d) soil on the 10th day during flooding. Different letters indicate a significant difference between treatment means (p < 0.05). Error bars represent the standard deviations of three replicates for the porewater Cd, which displayed a consistent increase with elevated Fe dissolution at a lower Eh (or a higher pH). These results identify a fine control of Eh in the pH range of 6.5-7.0 on As and Cd binding onto Fe (hydr)oxides under a flooded condition, which was determined largely by the amount of lime and Fh applied. Figure 4g, h shows specific changes in As and Cd fractions in LY with the CaO+1%Fh addition. There was a significant transformation of As from labile fraction (F1+F2, decreased by 5.2%) to species binding with amorphous and crystalline Fe oxides (F3+F4, increased by 6.0%), as compared with CK (Fig. 4g). This highlights the significance of Fe (hydr)oxides for maintaining high solid-phase content of As in paddy soils, which is prone to lower soil Eh and elevated DOC under the over-liming condition as analyzed above. With regard to Cd in LY, its exchangeable fraction has the highest plant availability dominated (59.2%), followed by the fraction bound on Fe/Mn oxides (22.4%). With CaO+1%Fh, exchangeable Cd content decreased by 24.6%, and a 10.7% elevation in Fe/Mn oxide-bound Cd was determined (Fig. 4h).
The above changes can be confirmed by SEM-EDS mapping of the treated soil (LY) (Fig. 5): elements, As and Cd, were highly enriched, and their spatial distributions were consistent with that of element Fe, which strikingly differed from the CK sample. This supports that Fe (hydr)oxides function as the principal hosting phase of As and Cd by capturing them from porewater.
Another aspect of note is that the percentage of exchangeable and carbonate-bound Cd (F1+F2, 56.2%) in the treated LY (CaO+1%Fh) was 11-fold higher than that of sorbed As (F1+F2, 4.8%), despite soil total Cd accounting for only 6.3% of total As. The high percentage of labile Cd in soils indicates a nonnegligible release potential of Cd even at near-neutral pH through cation exchange and/or acid dissolution by root exudates, which should be carefully constrained to minimize Cd uptake by rice, especially in grain filling stage . After 35 days of flooding, the profiles of As and Cd lability in LY were evaluated with 2D DGT mapping, which can provide a reliable indication of plant availability of target elements in soils (Guan et al. 2015;Huang et al. 2019;Luo et al. 2021). As shown in Fig. 6, the DGTlabile As and Cd decreased in a Fh-dependent manner with liming. Relative to control, DGT-As was reduced by 10.0-28.7% and DGT-Cd by 31.9-52.9% with increasing Fh from 0.25 to 1% (in the presence of lime). Particularly, with the CaO+1%Fh treatment, the averaged DGT concentrations of As (195.7 μg/L) and Cd (31.5 μg/L) were close to a class I soil (113.3 μg As/L, 24.9 μg Cd/L, As and Cd levels in rice grains< national standard) under the same flooded condition, which represents the highest soil quality category (namely, priority protection) and can sustain safe rice production.

Conclusion and implications
This study demonstrates whether the maximum As immobilization can simultaneously occur with Cd at near-neutral pH (~6.5-7.0) in flooded paddy soils that are highly dependent upon the relative availability of porewater Fe. At pH > 6.0, elevated DOC acts as a prominent competitor of dissolved As for sorption sites on Fe (hydr)oxides precipitated from porewater with liming. This availability of Fe can be indicated by the mole ratio of porewater Fe/DOC, serving as a key indicator of the release consistency between As and Cd in limed soils. For flooded paddy soils with relatively high mole ratios of porewater Fe/DOC at pH 6.5-7.0 (e.g., ≥ 0.23 in LY), the maximum co-immobilization of As and Cd can occur. Additional application of exogenous Fe (e.g., Fh) in with (e-h) CaO+1%Fh. Note: additional As and Cd was spiked into both soils in order to meet the detection limit of SEM-EDS Fig. 6 Profiles of DGT-labile As (a-e) and Cd (f-j) in LY soil (a, f) with CaO+0.25%Fh (b, g), CaO+0.5%Fh (c, h), and CaO+1%Fh (d, i). A class I soil was included for comparison (e, j) 1 3 these soils is optional and can achieve higher standards of soil quality, securing safe rice production. For flooded paddy soils with relatively low mole ratios of porewater Fe/DOC at pH 6.5-7.0 (e.g., 0.01-0.03 in CZ and XX), the maximum As immobilization takes place at a lower pH (≤ 6.0). Exogenous Fe (e.g., Fh) needs to be supplemented along with liming to concurrently maximize soil As and Cd coimmobilization at pH 6.5-7.0. In field-scale practice, raising porewater pH to above 7.0 by liming should be strictly controlled, which would otherwise promote DOC and Fe dissolution and thus induce the co-release of As and Cd. Overall, this study highlights the controlling process of DOC on As capture by Fe (hydr)oxides formed during soil liming and advances the knowledge for assessing lime applicability in As and Cd co-contaminated paddy soils.