The removal of veterinary antibiotics in the high-rate anaerobic bioreactor: continuous and batch studies Open Access

In China, more than 0.5 billion pigs entered the market in the year 2020 (Ministry of Agriculture and Rural Affairs 2021). Veterinary antibiotics are often used both for therapy and health care as additive of pig's feed to prevent disease and promote pig's growth to sustain such huge husbandry. Negative effects of antibiotics abuse in husbandry have caused public concern, like antibiotic resistant bacteria (Gao et al. 2022). A series of policies have been promulgated in China to control and reduce the use of veterinary antibiotics (Hao et al. 2015). Currently, however, pig farmers still heavily rely on veterinary antibiotics resulting in severe antibiotics abuse. As one of the consequences, large amounts of swine wastewater containing veterinary antibiotics were produced.

Sulfonamides (SAs) and tetracyclines (TCs) are the most widely used veterinary antibiotics and fluoroquinolones (FQs) emerges as a concerned pollutant (Cheng et al. 2018). TCs are usually added into pig's feed for the disease prevention and growth promotion. SAs and FQs are frequently used in therapeutic treatment. Due to the incomplete degradation of veterinary antibiotics in the pig's metabolism, some veterinary antibiotics enter environment through pig's feces and urine. It has been reported that over 14% TCs were excreted into feces (Xu et al. 2020) and up to 70% unmetabolized FQs were released to the environment (Van Doorslaer 2014).

High-rate anaerobic bioreactors, such as upflow anaerobic sludge blanket (UASB) reactor, expanded granular sludge blanket (EGSB) reactor and internal circulation (IC) reactor, have been applied to treat swine wastewater (Zeng et al. 2019). The well-settling granular sludge is a key support for high-rate anaerobic bioreactors. The functional microorganisms in anaeroblic granular sludge (AnGS) constitute the high-efficient microbial community by self-selection, self-assembly and self-immobilization, which give rise to the good adsorption capacity, the degradation activity and the settleability of granular sludge (Yu et al. 2019).

Adsorption and biodegradation are two important mechanisms for antibiotics removal in biological wastewater treatment systems (Li & Zhang 2010). In the anaerobic bioreactor, antibiotics molecules have different dissociation forms according to their dissociation constants (pKa) and environmental pH. They can be adsorbed onto charged particles, like microorganism cells, by electrostatic interaction and hydrophobic effect (Ahmed et al. 2015). Microbial degradation process of organic pollutants in swine wastewater is synergistically mediated by different microbial groups (Vrieze & Verstraete 2016). Hence the abundant anaerobic microorganisms which colonized within AnGS provide multiple possible metabolic routes for antibiotics biodegradation. Previous studies, mainly based on the batch and completely mixed systems, showed anaerobic treatment can remove TCs and SAs effectively. Liu's results showed sulfon-amides were most effectively removed by anaerobic digestion, followed by quinolones and tetracyclines (Liu et al. 2018). Cheng's results showed the tetracycline, oxytetracycline and chlortetracycline q_m of anaerobic sludge were 169, 185 and 185 mg/gMLSS, respectively (Cheng et al. 2020). Oliveira's results showed, in batch system, when sulfamethazine concentration was low (<50 μg/L), biodegradation was the primary removal route, and when the concentration was high (>50 μg/L), adsorption became dominated (Oliveira et al. 2016). Mohring's results showed after 5 weeks' anaerobic digestion of swine manure, sulfadiazine, sulfamerazine, sulfamethoxazole, sulfadimethoxine, and trimethoprim were completely eliminated while sulfathiazole, sulfamethazine, and sulfamethoxypyridazine showed persistence (Mohring et al. 2009). However, in the continuous and high-rate anaerobic granular sludge bed bioreactor, the behaviors of these veterinary antibiotics remain unclear (Cheng et al. 2018, 2020). The aims of this study were: (1) to investigate the elimination routes of target antibiotics in the high-rate anaerobic bioreactor over long-term; (2) to characterize the adsorption and biodegradation aspects of AnGS in order to study the behavior of antibiotics in the bioreactor.

In batch isotherm experiments, SMZ and ENR were dissolved in pH = 7.0 phosphate buffer solution, rising from 10 mg/L to 160 mg/L. CTC was dissolved in the range of 200–1,200 mg/L. One gram sterilized AnGS (wet weight) was added into a 50 mL-solution-containing flask. Flasks were fixed in a thermostat shaker at 35 °C and 150 rpm. The adsorption time was set to 48 h. In batch kinetics experiments, the initial concentrations of SMZ, ENR and CTC were set at approximately 100 mg/L, 100 mg/L and 500 mg/L. 5 g sterilized AnGS (wet weight) was added into a 200 mL-solution-containing flask which was fixed at a 35 °C and 150 rpm thermostat shaker. Each experiment was repeated in triplicate. The isotherm data was fitting with Langmiur model and Freundlich model and the kinetics data was fitted with pseudo-first-order model and pseudo-second-order model by Origin 2019b. See supplementary material: adsorption kinetics models.

In each 250 mL flask, 5 g AnGS was added into 200 mL synthetic swine wastewater which did not contain organic carbon source. Three experimental groups were set: in group I, the sole organic carbon source was antibiotics with concentration of 50 mg/L; in group II, the organic carbon source was glucose (5,000 mg/L) and antibiotics (50 mg/L); in group III, the organic carbon source was acetate (5,000 mg/L) and antibiotics (50 mg/L). After stripping with N₂ for 3 min, each flask was sealed with butyl rubber plug and fixed in water bath at 35 °C. Samples were taken by syringes during the degradation process. Each experiment was repeated in triplicate. The degradation data was fitted with first-order reaction kinetics by Origin 2019b. See supplementary material: degradation kinetics model.

TS, VS, pH and COD were measured by the standard methods (APHA 2012). Before volatile fatty acids (VFA) and antibiotics determination, water samples were filtered through 0.45 μm membrane to reduce suspended solids interference.

VFA concentrations, including acetate, propionate and butyrate, were determined by gas chromatograph equipped with flame ionization detector (GC-FID). GC on a 6890N Network GC System (Agilent Technologies, USA) installed an Agilent DB-FFAP column (30.0 m, 0.32 mm, 0.25 μm). FID worked at 250 °C, hydrogen flowed at 40.0 mL/min, air flowed at 400 mL/min and the carrier gas was nitrogen. The running time was 18 min per sample.

Antibiotics concentrations, including SMZ, ENR and CTC, were determined by LC-20AD high performance liquid chromatograph (HPLC) (SHIMADZU, Japan), equipped with SPD-20AV detector and InerSutain C18 column (4.6 × 250 mm, 5 μm). The running condition of SMZ and ENR: 65% 0.2 M fomate and 50 mM ammonium solution, and 35% methanol; retention time of SMZ and ENR was 10 min and detection wavelenghth was 275 nm. The running condition of CTC: 55% 20 mM oxalic acid solution, 22.5% methanol and 22.5% acetonitrile; retention time of CTC was 15 min and detection wavelenghth was 370 nm.

The operation was divided into three stages: stage I at low OLR and low antibiotics level; stage II at high OLR and low antibitoics level; stage III at high OLR and high antibitotics level. During stage I (days 0–9), the average concentrations of SMZ, ENR and CTC in influents were 5.5 mg/L, 3.7 mg/L and 4.7 mg/L, respectively. The OLR of ASF bioreactor was 6 kg/(m³·d). The COD removal efficiency was over 90%, and the COD concentration of effluents was 150 mg/L with pH of 7.2–7.5. During stage II (days 10–27), the OLR of ASF bioreactor was lifted to 20 kg/(m³·d) with antibiotics concentrations unchanged. The COD concentration of effluents surged to 1,200 mg/L, with the decrease of COD efficiency to 73% and effluents pH to 6.8. During stage III (days 28–47), the average concentrations of SMZ, ENR and CTC in influents were elevated to 11.2, 7.9 and 7.2 mg/L with OLR maintained 20 kg/(m³·d). The COD concentration of effluent rose to 1,800 mg/L with 63% COD removal efficiency and pH around 6.5. The results showed the higher antibiotics concentrations in the effluents, the more severe effect on the performance of ASF bioreactor. The effluents’ pH dropped to 6.5 which indicated the VFAs accumulation. However, with such a combined effect of multi-antibiotics, the COD efficiencies of ASF bioreactor was over 50%, indicating the robustness of ASF bioreactor.

No VFAs accumulation was observed in the stage I, indicating the anaerobic microorganisms functioned in balance, even though multi-antibiotics affected. When the OLR was elevated, in stage II, the acetate, propionate and butyrate in effluents accumulated rapidly, with average total concentration of 483 mg/L. With increasing OLR, concentrations of acetate and butyrate declined to 150 and 35 mg/L, but propionate rose sharply to 700–800 mg/L. Propionate degradation was mediated by propionate oxidation bacteria and methanogenic archaea (Bok et al. 2001). This sytrophic community was important but fragile, because the anaerobic oxidation of propionate was most thermodynamically unfavourable among the degradation of VFAs (Stams & Plugge 2009). With the co-existence of antibiotics SMZ, ENR and CTC, the concentration of propionate rose but acetate declined, suggesting that these antibitotics inhibited the metabolism of propionate oxidation bacteria. When entering stage III, the concentration of acetate rose gradually, accompanying a decrease of propionate. This result showed the activities of acetotrophic methanogens was inhibited, and the rate of acetate consumption was unable to keep pace with propionate oxidation. This imbalance led to the shift from propionate to acetate accumulation. It was perceived that when OLR was 6 kg/(m³·d), the ASF bioreactor was so robust that countered the effects of antibiotics, and worked well without VFAs accumulation. When the OLR was elevated to 20 kg/(m³·d), suffered from the dual stress of high organic loadings and antibiotics inhibition, the performance of ASF bioreactor deteriorated with the change of propionate accumulation into acetate accumulation.

During stages I and II, the average concentrations of SMZ, ENR and CTC in effluent were 5.2 mg/L, 1.0 and 0 mg/L, with the corresponding removal efficiencies of 0, 72 and 100% for SMZ, ENR and CTC. During stage III, with higher antibiotics concentrations in the influents, the removal efficiencies of SMZ, ENR and CTC were 0, 54 and 100%. ASF bioreactor could achieve a good and remarkable removal capacity for CTC and ENR, respectively, but showed no removal capacity for SMZ.

During the initial 20 HRT (120 h), the concentration of ENR in effluents gradually rose from 0 to 1.4 mg/L. After 20 HRT, the concentration of ENR stayed at 1.4 mg/L. The ‘apparent’ breakthrough curve of ENR spanned 20 HRT, and the parameters were shown in Table 3. The reason for using ‘apparent’ breakthrough curve was that the removal of ENR obviously included two mechanisms: adsorption and biodegradation. The ‘real’ breakthrough curve was shorter than the apparent one. The hypothetical breakthrough curve was calculated by using Thomas model where k_Th and q₀ were the same as the apparent one, changing with C₀ of 3.7 mg/L (the influents concentration); see Figure 4. Within incipient 4 HRT (24 h), the contribution of adsorption and biodgradation was 59% and 34% in removal efficiency, with the average removal efficiency of 92%. Within 4–8 HRT (24–50 h), the granular sludge bed was approaching saturation state, the removal efficiency of adsorption declined. After 8 HRT, biodegradation dominated the removal process, accounting for 62% in the removal efficiency of ENR.

The CTC concentrations in effluents during the whole experiment were almost undetectable which indicated the ASF bioreactor had a good performance for CTC removal. Based on the results from isotherm and kinetics experiments, AnGS had an outstanding adsorption capacity for CTC, therefore the granular sludge bed had a large adsorption capacity. According to the Thomas model, when the capacity q₀ was hypothetically assigned 50 mg/g-VSS (100 times SMZ capacity), k_Th = 0.02 L/(h·mg), C₀ = 5 mg/L, Q = 0.5 L/h, X = 110 g-VSS, the exhaustion time (C = 0.95C₀) was 372 HRT (2,232 h) which was much longer than that of SMZ and ENR; see Figure 4.

Once an anaerobic bioreactor was used to treat swine wastewater, the physical and chemical interactions occurred between antibiotics and AnGS. These interactions led to the antibiotics removal by adsorption. After fitted by the Langmuir and Freundlich models, the optimal fitting isotherm was chosen according to higher correlation coefficient (R²). At 30 °C, pH around neutral, the experimental data and adsorption isotherms for SMZ, ENR and CTC on AnGS were shown in Figure 1, and the fitting parameters were shown in Table 4.

The SMZ and CTC adsorptions by AnGS were well fitted to Langmuir model, but the ENR adsoption was better fitted to Freundlich model. The saturated SMZ adsorption capacity of AnGS was q_m = 1.74 mg/g-VSS, R² = 0.990. AnGS had a good adsorption capacity of CTC over 800 mg/g-VSS, but a poor adsorption capacity of ENR around 20 mg/g-VSS. Tetracycline antibiotics have high water solubility (0.008–0.062 mol/L) and lower octanol–water partition coefficient (logKow from −1.25 to −1.12), which defines their hydrophilic property. In contrast, ENR and SMZ have relatively hydrophobic property. More hydrophilic surface area on AnGS conferred a good adsorption capacity of CTC, but poor adsorption capacities of ENR and SMZ (Van Doorslaer 2014).

At the initial concentration 100 mg/L, AnGS had highest removal efficiencies of 5 and 24% for SMZ and ENR, respectively. In contrast, CTC removal efficiency was 98% at 400 mg/L initial concentration. The half-saturation time of AnGS for SMZ, ENR and CTC were 3.1 h, 2.1 h and 3.3 h, respectively. According to the previous studies, the half-saturation time of activated carbon for sulfonamides, flouroquinolones and tetracyclines were 5 min, 3–15 min and 6 h, respectively (Calisto et al. 2015; Genç & Dogan 2015; Zhang et al. 2015). Compared with activated carbon, AnGS had a higher CTC adsoption rate, but much lower rate of SMZ and ENR adsorption.

In the SMZ removal process, the adsoption accounted for 11% of total removal efficiency, and the biodegradation accounted for 89%. In the ENR removal process, the adsorption and biodegradation contributed 17 and 83%, respectively. It is possible to speculate that the removal of SMZ and ENR depends on the biodegradation in the AnGS system. Owing to the good CTC adsorption capacity and high adsorption rate of AnGS, the CTC concentration declined sharply at 0.2 mg/L within 24 h, which implied the adsoption was the main CTC removal mechanism in the AnGS system.

When glucose was introduced into the system, after cultivation for 120 h, the removal efficiencies of SMZ, ENR and CTC were 7.7, 31 and 100%, seeing Figure 6. From the first-order reaction kinetics models, the half-life of SMZ and ENR was elevated by 35 and 17%, which meant glucose hindered the biodegradation of SMZ and ENR. As for CTC, due to the good adsorption capacity of AnGS, the removal efficiency was not affected by glucose.

When acetate was introduced into the system, removal effiencies of SMZ, ENR and CTC had a notable decrease, with 2.3%, 10.4% and 97.6%; see Figure 6. The inhibition effect of acetate on the antibiotics biodegradation was stronger than glucose.

Another results difference between the high-rate and batch experiments was the biodegradation part. In the batch experiments, the removal efficiencies by degradation of SMZ and ENR were 2–9% and 10–31%, respectively. In the high-rate bioreactor, these became nearly 0 and 62%. The short HRT condition has an adverse effect on slow reaction in which the intrinsic reaction rate is low. When HRT is lower than the intrinsic reaction rate, the biodegradation reaction will cease going (Alvarino et al. 2014). The rate constant k can reflect the intrinsic reaction rate. Therefore, more recalcitrant, lower k, and the biodegradation process in the high-rate bioreactor is poorer. On the other hand, under multiple selective pressures, some functional microorganisms can possibly be riched, and the removal efficiency of the specific compound in high-rate bioreactor becomes higher than the batch system.

In the high-rate anaerobic bioreactor, SMZ removal process relied on the adsorption but it was very weak; ENR removal process mainly was biodegradation; CTC removal process nearly was adsorption because of the big capacity of AnGS. These results were helpful to create a rational basis for designing more suitable treatment systems as feasible barriers to the release of antibiotics into the environment.

The removal processes of veterinary antibiotics: sulfamethizole (SMZ), enrofloxacin (ENR) and chlortetracycline (CTC), in the high-rate anaerobic treatment system were investigated. Anaerobic granular sludge (AnGS) was proven to remove antibiotics by two main mechanisms: adsorption and biodegradation. After operation for 48 days, the average removal efficiencies of anaerobic bioreactor for SMZ, ENR and CTC were 0, 54 and 100%, respectively. By using fixed-bed adsorption models, the saturation time for SMZ, ENR and CTC was 4 HRT (24 h), 8 HRT (48 h) and 372 HRT (2,232 h). In the high-rate anaerobic bioreactor, SMZ removal process mainly relied on the adsorption but it was very weak; ENR removal process was based on the adsorption and biodegradation; CTC removal process was based to a large extent on the adsorption because of the large capacity of AnGS.

Zhuo Zeng: Experimental design, data curation and writing-original draft. Ping Zheng: Conceptualization, supervision and writing-review and editing. Da Kang: Methodology and visualization. Wen-Ji Li: Resources. Dong-Dong Xu: Methodology. Wen-Da Chen, Chao Pan and Lei-Yan Guo: Investigation.

This research was financially supported by the Science and Technology Program of Sichuan Province (2021YJ0382), the Fundamental Research Funds for the Central Universities (2682021CX066), and the Sichuan Youth Science and Technology Innovation Team funding (2022JDTD0005).

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

The authors declare there is no conflict.