Simultaneous nitrification and denitrification in a PCL-supported constructed wetland with limited aeration

Considerable advances have been made in the substrate design and operation strategies of constructed wetlands to facilitate nitrogen elimination. However, few studies examined the complicated interaction between solid organic substrates and limited aeration on nitrogen removal. A vertical flow constructed wetlands in gradient distribution of inorganic and solid organic substrates (polycaprolactone/PCL) (P-VFCW) and a controlled vertical flow constructed wetland without PCL (C-VFCW) were developed for the tertiary treatment of municipal tailwater. Results indicated that ammonia was nearly converted to nitrate, while the total nitrogen removal efficiencies (TNREs) in C-VFCW were negligible. In P-VFCW, however, optimal TNREs approached 95% with an aeration rate of 0.06 mL·min−1 and hydraulic retention time (HRT) of 24 h, and simultaneous nitrification and denitrification process (SND) in aerobic conditions was confirmed. As for the spatial microbial community structure evolution, Comamonas, which is associated with heterotrophic nitrification and anoxic/aerobic denitrification, was enriched along the vertical profiles of P-VFCW. Autotrophic nitrifier (Nitrospira), aerobic denitrifier (Bradyrhizobium and Azospira), and anoxic denitrifier (Ignavibacterium and Methyloversatilis) were dominated in different depths of P-VFCW, respectively. Besides, Canna indica biomass in P-VFCW was significantly larger than that in C-VFCW, which was attributed to the plant adaption response to diverse nitrogen. The P-VFCW in gradient distribution of inorganic and solid carbon sources under limited aeration is a promising technology for advanced nitrogen removal.


Introduction
Nitrogen is a prevalent pollutant present in surface water and groundwater, which would result in water environment quality's deterioration. Diverse procedures were performed to remove nitrogen from wastewater, including adsorption, ion exchange, reverse osmosis, air stripping, and electrochemical and biological process (Dehghani et al. 2018;Hossini et al. 2015;Jonoush et al. 2020). Among them, constructed wetland (CW) is widely applied for nitrogen removal due to its low cost, easy maintenance requirements, and eco-friendliness (Gunes et al. 2021;Parde et al. 2021;Vymazal et al. 2021 Wallace andWaltham 2021). The nitrogen removal mechanisms in CWs include plants' uptake, substrates' adsorption, and biotransformation, with the microbial nitrification and denitrification being the most dominant pathway (Tang et al. 2020;Zhou et al. 2020). In general, oxygen deficiency is always present in the CWs (Nguyen et al. 2019), and the scarcity of electron donors further restricts the denitrification process. Due to the limited oxygen and electron donor, CWs consistently exhibit suboptimal total nitrogen removal efficiency (TNRE) when treating wastewater with a low C/N ratio.
Previous studies have demonstrated that operation mode and substrate design of CWs are effective strategies to improve its nitrogen removal. Aeration was frequently used to improve the overall treatment efficacy of CWs (Boog et al. 2019;Wang et al. 2020), and it was found that aeration can significantly boost the total nitrogen removal of Responsible Editor: Alexandros Stefanakis CWs (Jehawi et al. 2020;John et al. 2020;Li et al. 2021). Compared to the CW without aeration, the aerated CW presented superior nitrogen removal capability, with a NH 4 + -N removal efficiency of 94.3% and TNRE of 91.5% . Meanwhile, nitrate arising from the nitrification process cannot be eliminated completely owing to the insufficient electron donor for denitrification, resulting in nitrate accumulation accompanied by a low TNRE in CWs. From this, how to eliminate the nitrate produced by the nitrification process is a crucial issue. Several researchers have recently attempted to tackle the insufficient electron supply for denitrification in CWs by employing solid organic substrates that served as both substrates and the carbon source (Si et al. 2018). It has been demonstrated that biodegradable polymers with a steady and robust electron supply capacity considerably improve the TNREs of CWs (Jia et al. 2021;Shen et al. 2014). These biodegradable polymers, mainly including polycaprolactone (PCL) and poly (3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV), have been investigated (Li et al. 2016;Xu et al. 2019;Yang et al. 2020). However, the cost of PHBV (21.0-37.2 €/kg NO 3 − -N) is significantly higher than that of PCL (5.4-8.9 €/kg NO 3 − -N) (Zhang et al. 2016). In light of this, PCL is more economically viable.
As such, newly, some research has attempted to combine the benefits of aeration with biodegradable polymers to promote the TNREs of CWs. In a PHBV/PLA-and ceramsite-supported CW, Yang ) discovered that under the limited aeration, higher ammonia removal efficiency (91%) and TNRE (97%) were higher than in the absence of aeration. Similarly, the NH 4 + -N and TN removal efficiencies in an aerated PHBV-supported CW were much greater than in non-aerated PHBV-supported CW, reaching 90.4% and 92.1%, respectively (Sun et al. 2018). However, in the experiments described above, the biodegradable polymers were placed at the bottom of CWs, which would result in two distinct types of issues. On the one hand, substrates inextricably link with the other key operation parameters and significantly influence the performance of CWs (Ji et al. 2021). In the presence of abundant organic matter, the competition of aerobic heterotrophic microorganisms for oxygen would bring adverse effects on nitrifiers (Tran et al. 2018, Zhu andChen 2001). On the other hand, aeration would increase the consumption of biopolymers by aerobic heterotrophic microorganisms, resulting in the clogging of CWs, secondary effluent pollution, and a decline in the in polymers' efficient utilization. The distribution pattern of biopolymers imposed considerable influences on nitrogen removal in aerated CWs. In the case of aeration, it was critical to develop a proper distribution of biopolymer and other inorganic substrates in CWs. Additionally, the underlying mechanism about simultaneous ammonia and nitrate removal in aerated polymer-supported CW was still unclear.
In this study, two VFCWs with limited aeration were developed: a VFCW without PCL polymer (C-VFCW) and PCL-supported VFCW (P-VFCW). This study aims to (i) investigate whether P-VFCW would show a good treatment performance during a long-term continuous operation and (ii) explore the spatial evolution of microbial community structure, and thereby elucidating the underlying mechanism about simultaneous ammonia and nitrate removal.

Materials
In the present study, two VFCWs were built with multiple inorganic/organic media: volcanic rock, ceramsite, and PCL polymer. Volcanic rock and ceramsite granules were purchased from GongYi Songxin Filter Material Industrial Enterprise, and the brand of PCL granules was American Solvay Advanced polymers 6800. The sizes of volcanic rock and ceramsite granules ranged from 4 to 6 mm and from 2 to 3 mm, respectively. The physical properties of PCL granules were cylindrical shape, a diameter of 3.0 mm, and a height of 1.00 mm.
Two VFCWs were intended for the advanced treatment of municipal tailwater. According to the "municipal wastewater treatment plant pollutants discharge standard" class A (GB18918-2002), synthetic wastewater containing 8 mg·L −1 NH 4 + -N and 7 mg·L −1 NO 3 − -N was prepared by adding NaNO 3 , NH 4 Cl, and KH 2 PO 4 into a 50 L container, and the characteristics of influent during different phases are shown in Table 1. In phase I, due to a lower operating temperature, the influent DO concentrations were higher, and the influent nitrite concentrations were lower. There was no chemical oxygen demand in the influent to determine whether

Experimental apparatus and procedures
The structure of each VFCW is shown in Fig. 1. Two VFCWs were both performed using a Plexiglas cylinder with 20-cm diameter and 80-cm length. The substrates were supported by a Plexiglas mesh disc with a 2-mm pore size. A titanium aeration disc was at the end of the experimental devices, which connected with an air pump. Six sampling ports were set at 10-cm intervals from 0 to 60 cm along the reactors. Multi-dimensional gradation of substrates was adopted in this study: a 10-cm ceramsite in the top layer to promote the plant growth, followed by 30-cm PCL polymer (750.00 g)/ceramsite mixture or ceramsite and 20-cm inorganic substrate-volcanic rock in the bottom layer to facilitate the nitrification process. Meanwhile, Canna indica (average height of 48.36 ± 1.42 cm, average weight of 262.62 ± 6.02 g) was chosen and planted in each VFCW. Activated sludge taken from the secondary sediment tank of the Quyang municipal wastewater treatment plant in Shanghai, China, was used as an inoculation source with a final biomass concentration of 800 mg MLSS·L −1 for each VFCW. After inoculation, synthetic wastewater was pumped into these systems from the bottom of the VFCWs at a flow rate of 13.08 mL·min −1 . A rotameter was to adjust aeration rates of air pump (Hailea, ACO-5504) between 0.06 mL·min −1 and 0.15 mL·min −1 . The experiment was divided into three phases to explore the effects of aeration rate and HRT on the effluent water quality. Two VFCWs were exposed to natural surroundings (such as air, temperature, light). The experiment operated from June 1 to August 30 and lasted for 3 months. The influent characteristics and operation parameters are listed in Table 1.

Analytical methods
Influent and effluent samples were collected daily. All the samples were filtrated through a 0.45 μm filter membrane to remove suspended solids. Total organic carbon (TOC) was determined by a TOC analyzer (TOC V-CPN, Shimadzu, Japan). NO 3 − -N, NO 2 − -N, and NH 4 + -N concentrations were measured according to the standard methods by UV spectrophotometer (UV 1900i, Shimadzu). Dissolved oxygen (DO) and pH were performed by a handled multi-parameter water quality analyzer (Hach, USA). All the tests were conducted in triplicate, and the measurement data were presented as mean values.

Effluent characteristics in two different VFCWs
The difference in treatment performance between the two VFCWs could be disclosed from the effluent quality. Figures 2, 3, and 4 depict the physical characteristics of pH and DO, nitrogen removal performance, and TOC variations for each VFCW during a long-term operation.

Effluent pH and DO characteristics
Figure 2 depicts the variations of pH and DO in the influent and effluent. During the whole operation process, the influent and effluent pH were 6.73-7.56 and 6.63-7.70 in P-VFCW and C-VFCW, respectively. On the whole, the difference in pH between influent and effluent in each VFCW was negligible. In general, the effluent pH always decreased in the biodegradable polymer-supported denitrification system, which was ascribed to the acidic intermediate products resulting from the hydrolysis of the polymers (Xu et al. 2018a). It is known that H + and OH − are produced during the nitrification and denitrification processes, respectively; hence, SND could alleviate pH fluctuations (Hu et al. 2021). Meanwhile, the influent DO in phase I remained reasonably stable, ranging from 6.09 to 7.16 mg·L −1 . The DO concentrations in the influent tank were 0.17-5.97 mg·L −1 and 0.32-5.92 mg·L −1 , respectively, which fluctuated with the exchange frequency of fresh wastewater in the tank. Whereas, the effluent DO in three stages slightly fluctuated. The effluent DO in C-VFCW varied between 5.00 and 8.46 mg·L −1 with an average of 7.61 ± 0.55 mg·L −1 . The effluent DO in P-VFCW ranged from 7.48 to 8.41 mg·L −1 , 4.35 to 7.32 mg·L −1 , and 3.72 to 7.55 mg·L −1 , with averages of 8.07 ± 0.28, 6.03 ± 0.93, and 5.44 ± 0.90 mg·L −1 for phases I, II, and III, respectively. Compared to the C-VFCW, the lower effluent DO in P-VFCW may be attributed to the DO consumption by aerobic heterotrophic bacteria. Additionally, there was no substantial suppression of the nitrification when DO exceeded 1.7-2.0 mg·L −1 ; nevertheless, concerning the denitrification process, the DO should be maintained below 0.3-0.5 mg·L −1 (Lin et al. 2020). Thus, high DO concentrations in this CW may inhibit denitrification, thereby decreasing the TNREs. During the whole operation, the influent NH 4 + -N varied from 3.82 to 8.38 mg·L −1 with an average of 6.71 ± 1.10 mg·L −1 , while the eff luent NH 4 + -N concentrations was negligible in each VFCW (Fig. 3a). Clearly, the NH 4 + -N removal efficiencies in all phases exceeded 95%, and it revealed that aeration rates of 0.06-0.15 mL·min −1 resulted in excellent nitrification efficiency. Compared to the non-aerated CW, the intermittent-aerated CW had a higher NH 4 + -N removal efficiency of 94.3% , which further indicated that the DO supply was advantageous for nitrogen removal in CW.

Effluent nitrogen characteristics
However, in phase I, the effluent nitrogen was mostly nitrate in each VFCW ( Fig. 3b and d). Due to the residual organic carbon source in the inoculated activated sludge or the substrates' adsorption, the TN slightly decreased in each VFCW in the first few days. The adsorption capacity of the substrates for NO 3 − -N was extremely low, and its adsorption capacity for NH 4 + -N was about 0.01 mg NH 4 + -N/g substrates, indicating the substrates have a weak adsorption capacity for nitrogen. -N concentrations averaged at 12.54 ± 1.84 mg·L −1 and 7.91 ± 0.56 mg·L −1 with low TNREs of 8% and 40% in C-VFCW and P-VFCW, respectively. Meanwhile, an excessive DO would create a high redox potential and further seriously hinder the denitrification process. When the aeration rate decreased from 0.15 to 0.06 mL·min −1 (phase II), effluent nitrogen primarily in the form of NO 3 − -N remained high in C-VFCW, but it sharply declined in P-VFCW. After a 16-day adaption period, the effluent NO 3 − -N concentrations were extremely low, and the effluent TN averaged at 1.50 ± 1.34 mg·L −1 with a remarkable TNRE of 90% in P-VFCW. Similarly, in an intermittent-aeration CW, a lower DO of 3 mg·L −1 and shorter HRT of 2 days resulted in a larger TNRE than a DO of 4 mg·L −1 and HRT of 3 days , suggesting that a lower DO was closely related to a higher TNRE. It was observed that the nitrate reduction enzyme-Nap locating at periplasm is not sensitive to the inhibition of DO; thus, Nap can still drive the reduction of NO 3 − -N ). Nevertheless, DO was still a crucial factor influencing the denitrification performance under aerobic conditions. In general, the optimal DO concentration for aerobic denitrification was generally found in the range of 3.0-5.0 mg·L −1 (Hu et al. 2021). Collectively, it was inferred that the SND process and aerobic denitrification contributed to a considerable TNRE in P-VFCW.
When the HRT decreased from 24 h (phase II) to 12 h (phase III), there was no noticeable difference in the effluent of C-VFCW. However, the effluent NO 3 − -N and the TN concentrations both increased and averaged at 2.63 ± 1.01 mg·L −1 and 3.28 ± 1.08 mg·L −1 in P-VFCW, respectively; thus, the TNRE decreased from 90 to 76%. It indicated that TNRE decreased with a decline of HRT in P-VFCW. Despite this, the TNRE was unexpectedly high under a shorter HRT and aerobic condition (DO > 4 mg·L −1 ). As an intermediate product of nitrification and denitrification processes, effluent nitrite accumulation was both negligible in each VFCW (Fig. 3c), revealing that these processes were largely complete. In addition, the cost of PCL was about 2-7 times that of using methanol and ethanol, but the same as that of using acetic acid (Wang and Chu 2016). However, there are no safety concerns in terms of toxicity and inflammability for PCL during transport and operation. If the price of PCL can be reduced further, it is reasonable to assume that an aerated P-VFCW for advanced nitrogen treatment will become more economically viable.

Effluent TOC characteristics
The effluent TOC concentrations are illustrated in Fig. 4, which could determine whether TOC accumulation occurred in the effluent of VFCWs. In general, TOC accumulation occurred in biodegradable polymer-supported denitrification systems for nitrate removal in previous studies (Xu et al. 2019(Xu et al. , 2018b. During the whole experiment, the effluent TOC was negligible in C-VFCW, and the effluent TOC concentrations were in the range of 1.11-5.95 mg·L −1 in P-VFCW. The average effluent TOC concentrations for phases I, II, and III were 2.53 ± 0.66, 2.71 ± 1.22, and 2.13 ± 0.68 mg·L −1 ,respectively. According to the environmental quality standards for surface water (GB3838-2002), the effluent residual TOC concentrations in the P-VFCW were too low to pose a threat of secondary pollution to the surface water. However, in an aerated PHBV/PLA-supported SND system, effluent TOC rapidly increased up to a maximal concentration of 36.40 mg·L −1 . Meanwhile, in an aerated PHBV and ceramsite-supported VFCW where PHBV polymer deposited at the bottom of systems, the COD concentrations were about 19.44 ± 7.91 mg·L −1 , which was still higher than that in the present study (Sun et al. 2018). It indicated that the gradient structure of biopolymer and other inorganic substrates in this study may have an obvious advantage in reducing effluent organics. Additionally, a lower effluent TOC in P-VFCW was attributed to the proper distribution pattern of organic and inorganic substrates under aerobic conditions.

Growth characteristic of Canna indica
Plants require substantial amounts of nitrogen to support growth, and hence, nitrogen acquisition is typically a growth-limiting factor. It is believed that plants in CWs make an important contribution in terms of contaminant uptake (Schierano et al. 2020;Zhang et al. 2021). In general, approximately 60-70% of TN is removed via denitrification, while 20-30% of TN can be consumed by plant uptake in CWs (Spieles and Mitsch 2000). After 90 days of cultivation, the average height of Canna indica increased from 37.55 ± 5.53 to 75.20 ± 4.08 cm and 105.97 ± 15.36 cm for C-VFCW and P-VFCW, respectively, and its wet weight rose from 64 ± 6.10 to 173.13 ± 2.16 g and 733.96 ± 4.04 g in C-VFCW and P-VFCW, respectively. Their dry weights were about 31.16 g and 132.06 g, respectively, and it contained 15 g N /Kg Canna indica. The nitrogen absorption amount of Canna indica was about 300.84 mg and 1848.09 mg, respectively. This result indicated that some of influent NH 4 + -N and NO 3 − -N were removed by Canna indica for its growth in each VFCW, and the contribution ratio of Canna indica for nitrogen removal was about 13.53% in P-VFCW.
Nitrogen response mainly refers to the plant adaptation responses to diverse nitrogen conditions or fluctuations in the nitrogen supply, which is a critical component of the regulatory network controlling plant growth (Ueda et al. 2017). NO 3 − -N and NH 4 + -N are the primary forms of nitrogen accessible to plants for assimilation. The relative growth rate of Canna indica exceeded 0.01 g·day −1 in the absence of L NH 4 + -N , indicating that a low effluent NH 4 + -N concentration would not inhibit the growth of Canna indica in either C-VFCW or P-VFCW. Meanwhile, it was found that when plants were supplied with both NO 3 − -N and NH 4 + -N, they preferred NH 4 + -N uptake more than NO 3 − -N. For example, when a candidate plant for wetland Arundo donax was fed with NO 3 − -N, it had significantly less above-ground biomass than when it was fed with NH 4 + -N ( Tho et al. 2017). Besides, high nitrate concentrations (> 5 mM) also negatively affected axial root growth in maize (Tian et al. 2008). Consequently, a higher NO 3 − -N concentration likely inhibited the growth of Canna indica, resulting in a lower biomass growth in C-VFCW than in P-VFCW.

Microbial community structures' spatial evolution
Due to a negligible TNRE in C-VFCW, the microbial community structures' spatial evolution in P-VFCW was deeply explored, which was closely associated with its good TNRE under an aerated condition. Illumina high-throughput sequencing for 16S rRNA genes was to reveal the microbial community structures of the biofilm samples collected along the vertical regions of P-VFCW. Table 2 summarizes the sequencing analysis results and the diversity of microbial communities. The sequences without sampling obtained from five samples were respectively 72,251, 60,581, 69,620, 71,239, and 72,995 with a length range of 414-420 bp. They were resampled to 54,460 for the subsequent analysis. The rarefaction curves indicated that the sequencing depth was adequate for the microbial community analysis. The Chao and Shannon indices  can be used to estimate the richness and diversity of the microbial community, respectively. The richness and diversity among these five samples differed significantly, according to the different microbial community structures.
The microbial communities were further identified at phylum and genus levels (Fig. 5). The relative abundances of major phyla (sequence percentage > 1% in at least one sample) are shown in Fig. 5a. The main phyla among all samples were divided into 12 phyla, but their relative abundances differed. Proteobacteria was the predominant phylum in all samples, with significantly higher relative abundances in samples 20_30C (60%), 30_40D (60%), and 40_50E (52%) than in samples 0_10A (32%) and 10_20B (40%) (Fig. 5a). In terms of the biological nitrogen removal processes, ammonia-oxidizing bacteria and denitrifying bacteria in PCL-supported system both lay in Proteobacteria (Bucci et al. 2021;Zhang et al. 2016). The difference in the relative abundances of Proteobacteria among five samples might be associated with distinct nitrogen biological processes in vertical regions of P-VFCW.
Top 30 genera are depicted in Fig. 5b, and the microbial community structures significantly differed with the vertical profile of P-VFCW. On the whole, Comamonas predominated in all samples, with relative abundances of 6.22%, 14.35%, 28.17%, 26.87%, and 13.79% in samples 0_10A, 10_20B, 20_30C, 30_40D, and 40_50E, respectively. Comamonas has been observed in a SND bioreactor under oxygen-rich conditions (Lai et al. 2020). Some strains of Comamonas were identified based on the denitrifying gene coding for nirS in a PHBV-supported denitrification system (Shams Tabrez Khan et al. 2007), while some of which were also capable of heterotrophic nitrification and aerobic heterotrophic denitrification (Cao et al. 2021;Liu et al. 2021;Moura et al. 2018). These results revealed that Comamonas played a crucial role in nitrification and denitrification processes in this system.
Aeration intensity regulates the bacteria growth and the distribution of bacteria species with depth in CWs (John et al. 2020), which further confirmed the genera variations with the vertical profile of P-VFCW. Nitrospira was found to be much more abundant in the samples 0_10A (2.06%) and 10_20B (1.21%) than in other samples. Similarly, Nitrospira, as a kind of autotrophic nitrifying microorganism (Xiang et al. 2020), played a critical role in the nitrification process in an aerated PHBV-supported denitrification system (Sun et al. 2019). As stated, Nitrospira also contributed to the nitrification process in the 0-20-cm zone of the system. Meanwhile, denitrification-associated microbes including Bradyrhizobium and Azospira (Horn et al. 2006;Vymazal et al. 2021) were observed in each sample, while their relative abundances in samples 0_10 A, 10_20B, and 20_30C were higher than in other samples. In  and Methyloversatilis were associated with anoxic denitrification (Ali et al. 2016, Liu et al. 2020, Shams Tabrez Khan et al. 2007. Bdellovibrio was heterotrophic aerobic bacteria involving in the degradation of organic matters (Iannacone et al. 2021), which was related to the organic substrates-PCL polymers in this study. Collectively, it was concluded that a spatial microbial community structure evolution mainly including the nitrifiers and aerobic/anoxic denitrifiers was responsible for a considerable TNRE in P-VFCW. Additionally, a PCA analysis was performed and is presented in Fig. 5c. The PCA analysis revealed the resemblance of community structures among these five samples. Environmental factors such as DO concentration, nitrogen concentrations and form (mainly including NH 4 + -N and NO 3 − -N), and substrate (inorganic or organic substances) were believed to be responsible for the microbial community structures. There was no visible similarity between 0_10A and 10_20B samples despite the fact that both samples were volcanic. This could be attributed to different concentrations and form of nitrogen. However, samples 20_30C, 30_40D, and 40_50E (the blue oval) were in close proximity, indicating that they shared a partial portion of microbial structure. The similarity among these three samples was contributed to the same substrates (PCL polymer and ceramsite), while the longitudinally decreasing NO 3 − -N concentrations brought subtle differences. Overall, the difference in the microbial structure could be a result of the distinct nitrogen metabolic capacities with the vertical profile of P-VFCW.

Simultaneous nitrification and denitrification mechanism
Aeration and organic substrate-PCL polymers were discovered to be critical factors for microbial community structures' spatial evolution. The association of aeration and PCL polymers resulted in a spatial microbial distribution involving in nitrification and denitrification processes (Fig. 5).
It was reported that in an aerated PHBV/PLA-supported biofilter, simultaneous nitrification and denitrification occurred in the first 5 cm of the column, but denitrification happens exclusively in higher sections of the column . Due to the DO consumption by the nitrification process, the DO concentration was highest at the bottom of P-VFCW and reduced with the increasing depth of the system. Nitrospira and Comamonas, both autotrophic/heterotrophic microbes, were responsible for the nitrification process in the 0-20-cm region of the system. Denitrification also occurred simultaneously in the 10-20 cm zone due to the presence of abundant aerobic denitrification bacteria such as Comamonas, Azospira, and Stenotrophobacter. PCL granules as a high-molecular-weight biopolymer were hydrolyzed and degraded into some low-molecular-weight substances (Fang et al. 2020). These dissolved intermediate degradation products of PCL granules would diffuse into the adjacent region of 10-20 cm, resulting in an aerobic heterotrophic denitrification process. Meanwhile, there is no doubt that PCL granules facilitated the denitrification process in 20-50-cm zone, which were consistent with the abundant aerobic denitrifying microbes and anoxic heterotrophic microbes. Besides, the mixture pattern of PCL and ceramsite would increase the effective utilization ratio of PCL polymers for denitrification, given that both the biofilm on the PCL polymer and ceramsite played an essential role in nitrate reduction. As a result, the underlying mechanism about SND in an aerated PCL-supported CW was deduced and is illustrated in Fig. 6.

Conclusion
Results showed that SND occurred in the P-VFCW; both aeration rate and HRT were crucial factors for the TNREs. The TNRE of P-VFCW still reached 76% even at a shorter Fig. 6 Simultaneous nitrification and denitrification mechanism in an aerated P-VFCW HRT of 12 h and a higher aeration rate of 0.15 mL·min −1 . Illumina sequencing analysis implied that the dominant microbes (mainly including ammonia-oxidizing bacteria, aerobic or anoxic denitrification bacteria) differed with the vertical profile of P-VFCW, which was associated with nitrification or denitrification processes. After the whole operation, a bigger biomass growth of Canna indica (increasing by 58.41 cm and 471.97 g) occurred in P-VFCW than that in C-VFCW, which was ascribed to a plant response to diverse nitrogen. This study developed a novel VFCW for advanced nitrogen removal from municipal tailwater.
Funding This work was supported by Shanghai Sailing Program (20YF1400200), the Fundamental Research Funds for the Center Universities (20D111324), and the National key research and development project (2019YFC0408603).

Data availability
The manuscript has not been published elsewhere. The article is not a consideration in any other journal in full or in part.

Declarations
Ethics approval and consent to participate Not applicable.

Competing interests
The authors declare no competing interests.