Adsorption of tylosin in wastewater by iron-rich farmland soil and the effect of iron reduction and common cations

Livestock wastewater reused in farmland may cause tylosin to stay in farmland soil. Under the influence of some factors, such as irrigation and fertilization, tylosin may desorb and diffuse into the water environment. Batch adsorption experiments and soil column flooding experiments were set up to investigate the effects of several cations and iron reduction on the adsorption, desorption and oxidation removal of tylosin in iron-rich farmland soils (red soil). The results showed that tylosin can be adsorbed by the red soil. The coexistence of these cations significantly reduced its adsorption capacity. The order of influence was as follows: Ca2þ>Mg2þ> K>NH4 >Na þ. This means that some agricultural farming measures, such as the application of chemical fertilizers, would release the adsorbed tylosin into the farmland. Anaerobic iron reduction and massive production of ferrous ions did not affect the adsorption and desorption of tylosin in the red soil column. Moreover, the ferrous iron could activate persulfate to generate hydroxyl radicals and sulfate radicals which oxidized and removed the tylosin adsorbed in the soil column. Therefore, the iron reduction that occurred during flooding was not a factor causing tylosin release, which provided a way for tylosin in iron-rich farmland soils to be oxidized and removed.


INTRODUCTION
As an antibiotic widely used in poultry and livestock, tylosin (TYL,C 46 H 77 NO 17 Hoese et al. () found that tylosin could reach 0.3 mg/g in the swine manure. There was 1-30 mg/L tylosin in some manure storage tanks (Kolz et al. ). In many areas, livestock wastewater was reused for farmland irrigation and fertilization after anaerobic treatment. However, antibiotics, such as tylosin, were rarely eliminated during anaerobic treatment (Ben et al. ) and would diffuse to farmland soil with wastewater reuse. It is reported that the adsorption distribution coefficient (K d ) of tylosin for some types of farmland soil in the United States was 24-65 L/kg (Hu & Coats ) and 10.4-387 L/kg in the Netherlands (Ter Laak et al. b), indicating that farmland soil has a strong adsorption capacity for tylosin. Part of the tylosin remaining in farmland soil might be degraded (Hu & Coats ; Sassman et al. ), whereas some factors might cause its further diffusion to natural water environments, endangering the safety of aquatic organisms (Wollenberger et al. ).
The mineral particles in the soil, especially those containing rich Fe(III), play an important role in the adsorption of tylosin (Guo et al. , ; Call et al. ). However, when the soil is flooded (wetland soil or paddy farming) and forms an anaerobic environment, dissimilatory iron reduction (DIR) reactions are likely to occur. During DIR, some bacteria can use Fe(III) in the soil as an electron acceptor, coupled to oxidation of organic matter. The energy released in iron reduction is captured by the bacteria for growth (Lovley ). Studies found that DIR caused the release of phosphate and heavy metals absorbed on the surface of soil mineral particles containing Fe(III) (Wang et al. ; Upreti et al. ). It is unclear whether tylosin would also be released into the soil solution due to DIR and then spread to downstream rivers with the discharge of water.
Meanwhile, as ion exchange was the main mechanism of adsorption of tylosin on iron minerals, it is easily affected To prevent tylosin from entering farmland soil and natural water environment, the tylosin in the wastewater needs to be treated before reuse. DIR in the soil may also provide a potential way for the oxidation and removal of tylosin. Ferrous iron generated by DIR could be adsorbed on the surface of the soil and form the Fe(II)/Fe(III) solid-phase interface. It is well known that Fe(II) and Fe(III) are the most efficient transition metals for activating hydrogen peroxide (Anipsitakis & Dionysiou ). To verify the above speculation, red soils rich in iron oxides were used to conduct the following investigations: (1) the adsorption characteristics of tylosin on iron-rich
It was air-dried and sieved through 0.15 mm for use. The total iron content in red soil was 53.7 g/kg, and the main iron minerals were hematite (Fe 2 O 3 ) and magnetite (Fe 3 O 4 ).

Effects of different cations
The batch adsorption experiments using 2.0000 ± 0.0005 g red soil and 50 mL tylosin solution were conducted in Erlenmeyer flasks. The adsorption isotherm experiments had different initial concentrations ranging from 0.5 to 100 mg/L. Common cations (Na þ , K þ , NH þ 4 , Ca 2þ and Mg 2þ ) contained in fertilizers were selected as coexisting ions, and their effects on the adsorption of tylosin were investigated. The concentration of these ions was set at 0.01 mol/L and was, respectively, formed by adding NaCl, KCl, NH 4 Cl, CaCl 2 and MgCl 2 to the experimental system. The initial concentration of tylosin was set at 5 mg/L.
According to the results of this experiment, the calcium ion having the greatest influence on the adsorption of tylosin was selected for the desorption experiment.

Effect of DIR
A 2.5 L glass bottle was used to build a sealed flooded soil system (Supplementary Figure S1). Because the red soil particles were tiny and the pore water in the soil column was limited, quartz sand was added to increase pore water. Preliminary experiments showed that tylosin was not adsorbed by quartz sand (Supplementary Figure S2). The glass bottle was filled with about 10 cm soil column, including 90 g red soil, 1,500 g quartz sand and 10 g paddy soil, providing iron-reducing bacteria. During the experiment, nitrogen was purged into the bottle to form the anaerobic environment. The adsorption efficiency of tylosin in the soil column system was verified by preliminary experiments (Supplementary Figure S3).
To explore the effect of DIR on the adsorption of tylosin, three experimental groups were set up. They were named as DIR pretreatment, Continuous DIR and No DIR. The soil columns were submerged with a solution with glucose concentration at 2,000 mg/L in both DIR pretreatment and Continuous DIR systems, and No DIR was treated using a solution without glucose. Except for glucose, the other components of the solutions remained the same (Supplementary Table S1). All three systems were kept in anaerobic incubation, and the ferrous ion in pore water was monitored.
When ferrous ion concentration no longer increased, the incubation was terminated and the solution was discharged.
Then, one solution containing 2 mg/L tylosin was added into the No DIR and the DIR pretreatment to flood the soil columns. Another solution added to the Continuous DIR included both 2 mg/L tylosin and 2,000 mg/L glucose.
Subsequently, pore water in three systems was taken at 1 h, 2, 4, 6, 8 and 10 days to determine the concentrations of tylosin and ferrous ion.

Fe(II)-activated persulfate oxidation
The experimental device was the same as that described in the section 'Effect of DIR'. Four groups were set up this time. All groups were pretreated to produce ferrous ions and to allow tylosin adsorption in the soil column similar to that in DIR pretreatment in the section 'Effect of DIR'.
Then, the soil column in four groups was flooded again by different concentrations (0, 3, 6 and 9 mmol/L) of sodium persulfate (Na 2 S 2 O 8 ) solution when the pretreatment solution was discharged. After 1 day of reaction, sodium persulfate solution was also discharged, and 0.02 mol/L calcium chloride solution was added for desorption. Before the sodium persulfate solution and calcium chloride solution were drained, pore water samples in the soil column of four groups were taken to determine the tylosin concentration.

Radical verification
To verify the types of radicals involved in the Fe(II)activated persulfate oxidation reaction, methanol and tertbutyl alcohol (TBA) were used as the scavenger of • OH and SO À 4 . It was reported that methanol can react with both • OH and SO À
The injection volume was 100 μL. The column temperature was 30 C.    ), the proportion of reduced iron oxides was relatively small (Supplementary Table S3), so it could not affect the adsorption sites of tylosin on the red soil.

Effects of different cations
The continued reduction of iron oxides in the red soil column of Continuous DIR did not lead to an increase of tylosin in the pore water, which indicated that the adsorbed tylosin did not undergo a large amount of desorption. This further showed that the DIR, which can cause changes in the chemical form of ferric iron in red soil, had no effect on the adsorption and desorption of tylosin. It was speculated that the inhibitory effect of tylosin on iron-reducing bacteria makes iron oxides adsorbing tylosin difficult to reduce. Tylosin has a great influence on soil microbial activity and microbial community (Westergaard et al. ; Aldén Demoling & Bååth ). Therefore, the reduction of iron oxides may be part of the soil particles that did not adsorb tylosin. In addition, it seemed that ferrous ions did not compete for tylosin adsorption sites of red soil particles like divalent cations such as calcium and magnesium ions. The reason may be the low concentration of ferrous ion, which was lower than that of calcium and magnesium ions in the section 'Effects of different cations'. For the actual iron-rich farmland soil, although the types of organic matter that could induce iron reduction contained in most farmlands were diverse, the iron concentrations in pore water were far lower than that in this study (Garnier et al. ). This means that the iron reduction and ferrous ions caused by flooding in the soil would be lower than the level of this study.
Therefore, it can be inferred that DIR in farmland soil would not affect the adsorption and desorption of tylosin.  reduced. However, the removal of tylosin did not increase with increase of persulfate. Therefore, for the amount of tylosin adsorbed by the soil columns, 3 mmol/L persulfate could achieve the maximum removal. In addition, the variation of ferrous iron before and after reaction in the first group soil column was also much lower than those in the other three groups ( Figure 6). Obviously, the ferrous iron in the soil column of the other three groups was involved in the reaction. It can be judged that the Fe(II)-activated persulfate oxidation reaction between ferrous iron and persulfate in these soil columns was likely to occur; the removal of tylosin should be the result of free radical oxidation produced by the reaction. Meanwhile, the variation of ferrous content in red soil columns increased with the increase in persulfate concentration, attributed to the reaction between excessive persulfate and ferrous iron, which was similar to a previous study (Xu & Li ). Figure 7 shows the changes of tylosin in pore water of the soil column of different systems before and after persulfate exposure. After oxidation, the tylosin concentration in the methanol or TBA coexisting system was significantly higher than that in the control system. This means that they prevented the removal of tylosin. Methanol is usually used as a scavenger for • OH and SO 4 À , and TBA was for SO 4 À . Therefore, the result confirmed that the two radicals, (1)

Radical verification
This further confirmed that the ferrous iron formed by the anaerobic reduction of the iron-rich soil could activate the persulfate to form an Fe(II)-activated persulfate oxidation system, which in turn caused the adsorbed tylosin to be oxidized and removed. Since free radical was not selective, other organic pollutants could also be removed in this reaction system. Therefore, it is also a potential way to remove other antibiotics in livestock wastewater. The removal was achieved in the following way. Before reusing, the wastewater first entered into a horizontal flow constructed wetland (HFCW) using iron-rich soil as substrate and periodically adding persulfate during operation.

CONCLUSION
Tylosin was retained in the iron-rich red soil by adsorption.
The adsorption isotherm was well described by the Henry model and the Freundlich model. A variety of cations contained in chemical fertilizers had an impact on the adsorption of tylosin. Divalent calcium ions and magnesium ions had the greatest impact, which could cause a large amount of adsorbed tylosin to be released. Therefore, tylosin entering the farmland soil was unstable. Anaerobic iron reduction and ferrous iron production had no significant effect on the adsorption and desorption of tylosin in the iron-rich red soil column. The ferrous iron from iron reduction could activate the persulfate to form free radicals, causing the adsorbed tylosin in the soil column to be oxidized and removed. This result can be used as a way to remove tylosin.

ACKNOWLEDGEMENT
This work was supported by the National Natural Science Foundation of China (Nos. 51778245 and 51378226).

DECLARATION OF INTEREST
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.