Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding

Chemical Engineering Journal xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding Zhen Wang ⇑, Menglu Huang, Ran Qi, Shisuo Fan, Yi Wang, Ting Fan School of Resources and Environment, Anhui Agricultural University, Hefei 230036, China

h i g h l i g h t s
Shunt ratio significantly influenced N transformation in TFCWs.
The TFCW with a shunt ratio of 1:2 removed pollutants most effectively.
Denitrification and anammox were intensified as the shunt ratio was increased.
Multiple N removal processes were active in the TFCW at the optimal shunt ratio.

a r t i c l e

i n f o

Article history: Received 7 September 2016 Received in revised form 5 November 2016 Accepted 7 November 2016 Available online xxxx Keywords: Tidal flow constructed wetland (TFCW) Shunt ratio Nitrogen removal Nitrogen transformation functional gene

a b s t r a c t This study was conducted to explore nitrogen transformation and associated microbial characteristics in a modified single-stage tidal flow constructed wetland (TFCW) at five different shunt ratios. Shunt ratio significantly affected nitrogen removal during operation of the TFCW with adoption of a modified stepfeeding mode. When the shunt ratio was 1:2, TFCW had the best pollutant removal performance and was particularly effective for total nitrogen (TN). The abundance of nitrogen transformation microorganisms was also affected by the shunt ratio of the system. Molecular biological analyses demonstrated that both denitrification and anammox were enhanced when the shunt ratio was greater than 0:1, resulting in the development of multiple and complete nitrogen removal pathways in the TFCW. These results show that the optimal shunt ratio of the modified single-stage TFCW for effective nitrogen removal was 1:2. Ó 2016 Elsevier B.V. All rights reserved.

1. Introduction Constructed wetlands are effective means of treating various types of contaminated water. In comparison with conventional treatment systems, constructed wetlands are more easily maintained and operated, require reduced input, and consume less energy [1]. Nevertheless, CWs often provide inconsistent nutrient reduction, in particular for nitrogen [2]. Therefore, the nitrogen removal capacity of CWs must be improved because nitrogen is a major contributor to water eutrophication. The primary nitrogen removal process in subsurface flow CWs is generally the microbial metabolic pathway, which removes approximately (89–96)% of nitrogen [3]. Several metabolic pathways, including conventional/partial nitrification–denitrification and anaerobic ammonium oxidation (anammox), are involved in nitrogen removal. Various environmental parameters and operating conditions influence the nitrogen treatment performance of ⇑ Corresponding author.

CWs; the ratio of COD to TN (C/N) and DO concentration are crucial for nitrogen transformation [4,5]. Tidal flow constructed wetlands (TFCWs) are a fairly new technology that can be used to enhance nitrogen removal from the environment via a novel method of oxygen transfer [6–8]. However, many studies of TFCWs have found that total nitrogen (TN) removal rates were not as great as expected because of a weak anoxic environment or inadequate organic carbon abundance, increasing the concentration of NOx-N in the effluent [9]. Achieving stable and high-rate nitrogen removal is usually considered as a formidable challenge in TFCWs. One major challenge is that it is extremely difficult to regulate oxygen supply and maintain appropriate level of DO in TFCWs. On the other hand, competition for substrates (e.g. NOx-N, organics, etc.) usually occurs among the various species of microorganisms (e.g. anammox bacteria and denitrifiers) present in TFCWs [10–12]. As C/N ratio and DO content have substantial impacts on nitrification, anammox, and denitrification in a CW system, an appropriate microenvironment must be created for each microorganism involved in nitrogen transformation in TFCWs.

E-mail address: [email protected] (Z. Wang). http://dx.doi.org/10.1016/j.cej.2016.11.060 1385-8947/Ó 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

2

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

Step-feeding is an effective method of improving TN removal in CWs [13]. In a previous study, we showed that the oxygen supply and C/N ratio in a subsurface vertical flow constructed wetland (VSSF) can be regulated by the adoption of step-feeding [14], enhancing nitrogen removal. Based on this finding, a single-stage TFCW with a modified step-feeding mode, which was expected to enhance TN removal, was established. Step-feeding in a CW is generally accomplished by installing a shunt pipe [15]. The shunt ratio of a CW is a crucial parameter that influences its TN treatment performance. Little effort has been expended to achieve satisfactory NH+4-N and TN removal in single-stage TFCWs; improved NH+4-N and TN removal have been achieved with multiple-stage TFCWs. The optimum shunt ratio for wastewater treatment by a single-stage TFCW with step-feeding remains unclear. Therefore, the effects of the shunt ratio on wastewater treatment in TFCWs and nitrogen removal mechanisms at the molecular level merit further investigation. In this study, the domestic sewage treatment performance of a modified single-stage TFCW with step-feeding was assessed at five different shunt ratios. For each TFCW, the absolute abundance of genes involved in nitrogen removal and their ecological associations were assessed to evaluate the relationship between the shunt ratio and the microbial community performing nitrogen removal. 2. Materials and methods 2.1. System descriptions TFCWs were constructed at the greenhouse of Anhui Agricultural University (AHAU) in Hefei, China, where the temperature was maintained at (25 ± 2) °C. Each single-stage TFCW (a polyethylene tank with a diameter of 20 cm and depth of 80 cm) was filled with 70 cm of oyster shell (particle size: 2–5 mm) as the substratum layer, as well as 10 cm of gravel (particle size: 10–15 mm) as the bottom under-drainage layer. The bed had a total volume of 25.12 L and a working volume of 10.12 L (initial porosity 40.30%). A ‘‘X”-shaped perforated inlet pipe was installed on the top of each TFCW, whereas another perforated pipe was installed at the bottom of the system as the shunt pipe. A vertical perforated PVC pipe (80 cm in length and 3.5 cm in diameter) was inserted into the bed in the middle of each TFCW to measure various physical and chemical parameters of wastewater in situ. Four orifices (25 mm internal diameter for each), which were used for collecting substratum samples, were respectively excavated in a line at different depths (14, 28, 42, and 56 cm) of side wall below the top of the bed. Every orifice was sealed by a rubber plug. The emergent plant employed in this study was reed (Phragmites australis), which is usually considered as a popular wetland plant [16]. Four reeds (initial height of approximately 30 cm) were planted in each system, and each of them has a main stem and two or three new shoots.

which occurred every 12 h, was divided into 6 phases in chronological order, consisting of 10 min Feeding Phase 1, 4.50 h Flood Phase 1, 20 min Feeding Phase 2, 3.00 h Flood Phase 2, 15 min Drain Phase, and followed by 3.75 h Rest Phase (as shown in Fig. 1). (a) Feeding Phase 1: a portion of wastewater was rapidly loaded into the TFCW in batch mode via the inlet pipe; (b) Flood Phase 1: after the first feeding, the lower section of the substratum layer was kept saturated for 4.50 h, while the upper section of the substratum layer was unsaturated; (c) Feeding Phase 2: another portion of wastewater was continuously fed to the TFCW via the shunt pipe by a peristaltic pump, in order to reduce oxygen transfer, wastewater was fed into the TFCW in an up-flow pattern rather than the original down-flow pattern; (d) Flood Phase 2: the entire bed was kept saturated for 3.00 h after the second feeding; (e) Drain Phase: all wastewater was drained rapidly via the outlet pipe installed at the bottom of the TFCW; (f) Rest Phase: the entire bed was allowed to rest in an unsaturated state for 3.75 h. Ten liters of wastewater were added to each TFCW for each cycle [corresponding to an HLR of 0.64 m3 (m2 d)1], including the wastewater added to the TFCW through the shunt pipe. The shunt ratio was defined as the ratio between the inflow volume through the shunt pipe and the inflow volume through the inlet pipe. In our study, five different shunt ratios were adopted: 0:1, 1:4, 1:3, 1:2, and 1:1. Correspondingly, heights of water level in the systems were respective 79, 63, 59, 53, and 40 cm above the bottom of the bed during Flood Phase 1 at the five different shunt ratios. Notably, the TFCWs with a shunt ratio of 0:1 were regarded as the control group with an operation mode representing that of conventional TFCWs. Prior to the experiments, five groups of TFCWs with different shunt ratios were fed wastewater for three months to allow the development of plants and biofilms in the bed. The experimental period was one year. 2.3. Analytical procedure Water samples were collected in triplicate once per week from the inlet and outlet of each TFCW and analyzed immediately. In order to provide insight into pollutant conversion mechanisms, a cyclic study was performed when the system was considered to be at a steady state. Water samples, extracted at the midpoint of the water depth from the vertical perforated pipe, were taken for chemical analysis with sampling intervals of 30 min during Flood Phase 1 and each 30 min during Flood Phase 2. As all the systems were operated for 720 cycles, each TFCW was excavated to allow substratum samples to be collected for experiments in the laboratory. During the sampling event, substrate samples excavated from the sampling orifices were mixed evenly, placed in an ice incubator, and immediately sent to the laboratory for DNA extraction. Soil DNA kits (D5625-01; Omega, USA) were used to extract and purify total genomic DNA. The genomic DNA extracted from the soil samples was detected by 1% agarose gel electrophoresis and stored at 20 °C. Besides, the average height of the plants in each TFCW was measured every week throughout the experiments.

2.2. Experimental conditions The TFCWs received schoolyard domestic sewage from AHAU after the sedimentation pretreatment. The water quality parameters of the sewage were as follows: TSS, (65.12 ± 24.51) mg L1; COD, (299.40 ± 17.72) mg L1; BOD, (117.83 ± 12.57) mg L1; 1  NH+4-N, (34.17 ± 5.11) mg L1; NO 2 -N, (3.37 ± 1.67) mg L ; NO3 -N, (1.72 ± 0.38) mg L1; TN, (43.56 ± 7.47) mg L1; TP, (13.64 ± 2.39) mg L1; and pH, (7.74 ± 0.58). Each TFCW was operated in a modified step-feeding mode (designated as the modified ‘‘tidal flow’’ principle using a two-time feeding mode) in this study. Specifically, the entire operation cycle,

2.3.1. Water quality analyses Water quality analyses were conducted for temperature, pH,  DO, ORP, TSS, COD, BOD, TN, NH+4-N, NO 2 -N, NO3 -N, and TP. The +  analyses (including TSS, COD, BOD, TN, NH4-N, NO 2 -N, NO3 -N, and TP) were performed according to standard methods for assessing water and wastewater [17] and pH values of the water samples were determined by a digital pH meter (PB-10, Sartorius, Germany). During a typical cycle of each TFCW, water temperature, DO, and ORP were all measured in situ at the midpoint of the water depth from the vertical perforated pipe, using a portable HACH HQ30d Multi-Parameter analyzer.

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

3

Fig. 1. Schematic of the modified single-stage TFCWs during a typical cycle (including Feeding Phase 1, Flood Phase 1, Feeding Phase 2, Flood Phase 2, Drain Phase, and Rest Phase).

2.3.2. Quantification of nitrogen functional genes The abundance of bacterial 16S rRNA, archaeal 16S rRNA, and nitrogen functional genes (amoA, nxrA, narG, napA, nirS, nirK, qnorB, nosZ, and anammox 16S rRNA) were determined by real-time PCR utilizing SYBR-green and a MyiQ2 Real-Time PCR Detection System (Bio-Rad, USA). Each 20-lL reaction mixture included 10 lL of SYBR Green I PCR master mix (Applied Biosystems, USA), 1 lL of template DNA (sample DNA or plasmid DNA for standard curves), 0.5 lL of forward primers, 0.5 lL of reverse primers, and 8 lL of sterile water. Each amplification consisted of 40 cycles. The primer sequences and real-time PCR protocol were adopted from those reported by Ji et al. [18]. 2.4. Data analysis Removal efficiencies and transformation or accumulation rates  of COD, TP, NH+4-N, NO 2 -N, NO3 -N, and TN were calculated according to the formulas reported by Zhi and Ji [12]. One-way ANOVA was performed using SPSS 21.0. The threshold of significance was p < 0.05. Differences between means were evaluated using Duncan’s test. Pearson’s correlational analysis was used to identify significant linear relationships. Notably, stepwise regression models were built to determine the quantitative response relationships between nitrogen transformation rates and functional genes using SPSS 21.0. 3. Results and analysis 3.1. Temporal variation of DO and ORP in TFCWs during a typical cycle DO plays a crucial role in microbial activity in wetlands [19]. Fig. 2 depicts the temporal variation curve of DO in each TFCW dur-

ing a typical cycle. The concentration of DO began to decrease at the beginning of Flood Phase 1; most DO was depleted by the end of Flood Phase 2. Nevertheless, there were differences between each curve that were due to the use of different shunt ratios in each TFCW. At the beginning of Flood Phase 1, the DO concentration ranged from 3.97 to 4.12 mg L1 for all TFCWs, indicating improved reaeration capacity in comparison with that of traditional CWs [20]. Subsequently, the DO concentration decreased because of degradation of organics and oxidation of NH+4-N during Flood Phase 1. Moreover, DO consumption decreased as the shunt ratio was increased during Flood Phase 1, perhaps because of the decrease in the size of the first feeding as the shunt ratio was increased. At the end of Flood Phase 1, the DO concentrations of the water samples increased from 0.15 to 0.86 mg L1 as the shunt ratio was increased from 0:1 to 1:1. The DO concentration continued to decrease at each shunt ratio during Flood Phase 2, which began after the second feeding. At the end of Flood Phase 2, the DO concentrations of the water samples were as follows: 0.13 mg L1 at the shunt ratio of 0:1, 0.10 mg L1 at the shunt ratio of 1:4, 0.08 mg L1 at the shunt ratio of 1:3, 0.07 mg L1 at the shunt ratio of 1:2, and 0.03 mg L1 at the shunt ratio of 1:1. Fig. 2 also shows the temporal variation curve of ORP in each TFCW during a typical cycle. At each shunt ratio, the ORP profile trend was similar to that obtained for DO. From the beginning of Flood phase 1, ORP decreased at each shunt ratio; the ORP values of the water samples increased from 16.5 to 101.0 mV as the shunt ratio was increased from 0:1 to 1:1 at the end of Flood Phase 1. The ORP value decreased over time at each shunt ratio during Flood Phase 2. The results indicated that, a reductive environment gradually came to dominate each TFCW because of DO consumption during the entire flood phase. Furthermore, the reductive environment became increasingly predominant as the shunt ratio was

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

92.79 ± 8.56 96.68 ± 5.68 94.11 ± 9.33 71.62 ± 7.49 83.70 ± 10.48 – – 89.81 ± 6.77 – 4.70 ± 1.39 9.94 ± 2.45 6.94 ± 2.17 12.36 ± 3.58 5.56 ± 1.25 2.08 ± 0.47 0.90 ± 0.04 1.39 ± 0.13 7.72 ± 0.63 93.58 ± 11.24 96.85 ± 3.58 96.23 ± 4.86 80.20 ± 5.67 89.95 ± 9.01 – – 93.41 ± 11.49 – 4.18 ± 1.21 9.43 ± 1.07 4.44 ± 2.06 8.62 ± 2.45 3.43 ± 1.03 2.62 ± 0.56 0.57 ± 0.50 0.90 ± 0.14 7.50 ± 0.72 94.16 ± 2.43 95.90 ± 3.21 97.11 ± 5.96 57.79 ± 1.64 83.23 ± 5.93 – – 94.60 ± 5.24 – 3.80 ± 0.72 12.28 ± 2.16 3.40 ± 1.74 18.39 ± 3.51 5.73 ± 0.84 7.03 ± 1.27 5.63 ± 0.89 0.74 ± 0.08 7.24 ± 0.53 94.05 ± 4.79 95.53 ± 4.81 96.39 ± 7.31 32.79 ± 1.64 84.24 ± 7.42 – – 95.40 ± 5.24 – 3.87 ± 0.78 13.38 ± 3.24 4.25 ± 1.28 29.28 ± 4.58 5.38 ± 1.14 7.42 ± 1.68 16.47 ± 3.40 0.63 ± 0.11 6.90 ± 0.36 94.21 ± 2.36 96.46 ± 3.24 97.23 ± 5.83 4.89 ± 1.56 88.39 ± 7.61 – – 94.60 ± 8.77 –

Removal rate/% Shunt ratio Influent/mg L1

Table 1 Pollutants removal of TFCW at five different shunt ratios.

3.2. Overall performance of the TFCWs

3.77 ± 1.01 10.60 ± 1.23 3.26 ± 2.21 41.43 ± 5.87 3.97 ± 0.23 6.35 ± 1.83 31.11 ± 4.67 0.74 ± 0.15 6.59 ± 0.45

Effluent/mg L1 Effluent/mg L1

increased during Flood Phase 2, which should be conducive to NOx-N reduction (e.g. denitrification, anammox, etc.).

65.12 ± 24.51 299.40 ± 17.72 117.83 ± 12.57 43.56 ± 7.47 34.17 ± 5.11 1.72 ± 0.38 3.37 ± 1.67 13.64 ± 2.39 7.74 ± 0.58

Effluent/mg L1 Effluent/mg L1

1:4 0:1

Fig. 2. Temporal variations of DO and ORP in TFCW during a typical cycle at five different shunt ratios (a: Feeding Phase 1, b: Flood Phase 1, c: Feeding Phase 2, d: Flood Phase 2, e: Drain Phase, f: Rest Phase).

Table 1 shows the characteristics of the TFCW effluents as stable pollutant removal was achieved in each system. Moreover, in order to provide insight into pollutant removal mechanisms, particularly for nitrogen, a cyclic study was performed when the system was considered to be operating at a steady state (Fig. 3). As shown in Table 1, high TSS, BOD, and COD removal rates were achieved when the TFCW was operated at a shunt ratio of 0:1. Increasing the shunt ratio did not reduce the rates at which TSS, BOD, and COD were removed, and the mean removal rates of TSS, BOD, and COD were all greater than 90% at the five different shunt ratios. Regarding the removal rates of COD and BOD, there were no significant differences between the two groups of data at each shunt ratio, indicating that the wastewater could be biodegraded easily due to less refractory organics existed in it. In a CW system, organics can be decomposed by both aerobic and anaerobic microbial processes as well as by sedimentation and filtration of particulate organic matter [21]. Specifically, the reaeration capacity of CWs has a significant impact on organic matter degradation [22]. The oxygen supply in TFCWs can reach 450 g (m2 d)1 [8], resulting in remarkable improvement in organic removal efficiency in

Effluent/mg L1 Removal rate/%

1:3

Removal rate/%

1:2

Removal rate/%

1:1

Removal rate/%

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

TSS COD BOD TN NH+4-N NO 3 -N NO 2 -N TP pH

4

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

5

 Fig. 3. The concentration curves of COD, TN, NH+4-N, NO 3 -N, and NO2 -N in TFCW during a typical cycle at five different shunt ratios (a: Feeding Phase 1, b: Flood Phase 1, c: Feeding Phase 2, d: Flood Phase 2, e: Drain Phase, f: Rest Phase), and the height of plants in each TFCW throughout the experimental period.

comparison with that of traditional CWs. As Fig. 3 shows, most of the COD was consumed within 100 min in all TFCWs after the beginning of Flood Phase 1. During Flood Phase 1, the COD removal rate remained at (91.89 ± 0.80)% as the shunt ratio increased from 0:1 to 1:1. Since the shunt ratio was greater than 0:1, the sudden increase of COD concentration in Flood Phase 2 was monitored due to the adoption of step feeding. And then the immediate consumption might be partly caused by the denitrification process. An ideal effect on TP removal was observed at each shunt ratio because of the significant phosphorus adsorption capacity of the

oyster shell material used in the TFCWs [23]. However, the mean TP removal rate decreased to (89.81 ± 6.77)% at the highest shunt ratio of 1:1, perhaps because of the relatively short hydraulic retention time of wastewater in the TFCW. When the shunt ratio was 1:1, 5 L of wastewater was pumped into the TFCW through the shunt pipe as Flood Phase 2 began, which resulted in an insufficient contact time (CT) between the substratum and wastewater in comparison with the CTs at other shunt ratios; relatively shorter CTs were associated with reduced TP removal rates.

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

6

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

Shunt ratio significantly affected the nitrogen removal capacity of the TFCWs (Table 1 and Fig. 3). A high NH+4-N removal rate (88.39 ± 7.61)% was observed when the shunt ratio was 0:1, but the TN removal rate was only (4.89 ± 1.56)%, while nitrogen in the effluent existed mainly in the form of NO 2 -N. During Flood Phase 1, the NH+4-N concentration decreased significantly from 34.17 to 5.89 mg L1. Subsequently, the NH+4-N concentration continued to decrease to 3.97 mg L1 during Flood Phase 2. Contrary to the variation trend of the NH+4-N concentration, NO 2 -N accumulation occurred at the beginning of Flood Phase 1 and peaked at 30.66 mgL1 at the end of Flood Phase 1. Subsequently, the 1 NO during 2 -N concentration ranged from 29.94 to 31.11 mg L Flood Phase 2. On the other hand, the NO -N concentration 3 increased slightly over the course of the entire Flood Phase; the 1 NO at the end of Flood Phase 3 -N concentration was 6.35 mg L 2. These results demonstrate the efficiency of the TFCW operated with a shunt ratio of 0:1 for removing incoming nitrogen. There are several explanations for these findings. First, a higher nitritation rate [19.24 g (m2 d)1], which is often the limiting step for eliminating nitrogen from wastewater in treatment wetlands [24], was achieved during the entire Flood Phase. This finding was compared with those reported in the similar literatures [13], and the nitritation rate of our TFCW with the shunt ratio of 0:1 was within the range of other wetlands treating different types of wastewater reported in the studies [0.53–32.25 g N(m2d)1], indicating its better nitritation ability. Second, inadequate organic carbon limited denitrification during the entire Flood Phase. Third, significant NO 2 -N accumulation occurred during the entire Flood Phase because of the limited reaeration capacity of the TFCW and the competitive advantages of AOB compared with NOB [25]. Fourth, the reduced TN removal rate might be due largely to the feeble denitrification and anammox. As the shunt ratio was greater than 0:1, the C/N ratio at the beginning of Flood Phase 2 increased along with the increase of the shunt ratio. When the shunt ratio was 0:1, 1:4, 1:3, 1:2, and 1:1, respective values of about 0.58, 1.84, 2.31, 2.97, and 4.04 (C/N) were obtained since Flood Phase 2 began. As we know, organic carbon should be needed in order to achieve denitrification of NOx-N to N2 [20], thus TN removal of the TFCWs could be enhanced partly since the shunt ratio was greater than 0:1. Nevertheless, the enhancement should be severely limited by the following reasons: (a) insufficient organic carbon still hindered the denitrification process even though step-feeding was adopted after Flood Phase 1 ended. Zhi and Ji [12] reveals that a C/N ratio exceeding six is required to achieve complete denitrification without NOx-N accumulation in a TFCW; (b) the competition for the organics occurred between the heterotrophic denitrifiers and other heterotrophic microorganisms; (c) a certain amount of ammonium existed in the secondary feeding, which would be oxidized to NOx-N by utilizing the residual oxygen as the electron acceptors. On the other hand, apart from the classical nitrogen removal route, the autotrophic nitritation-anammox process might also play a certain role in TN removal, since a stricter oxygen-limiting environment was created in the bed after the secondary feeding, as well as the presences of NH+4-N and nitrite. Overall, based on the operational characteristics of our TFCWs, significant NO 2 -N accumulation occurred during Flood Phase 1. Thereafter, the anaerobic/anoxic environment in the bed was further enhanced by the adoption of step-feeding, facilitating denitrification and anammox. Moreover, supplemental electron donors (primarily organics and NH+4-N) were also provided for heterotrophic denitrifiers and anammox bacteria. In general, anammox bacteria are capable of coexisting with heterotrophic denitrifiers in the same reactor [26] and can also grow mixotrophically [27]. Hence, both denitrification and anammox could be improved by utilizing the modified step-feeding mode in the TFCW, and the TN removal rate increased

from (32.79 ± 1.64)% to (80.20 ± 5.67)% since the shunt ratio was increased from 1:4 to 1:2. It was noted that the nitrogen removal rate began to decrease as the shunt ratio was increased to a ratio greater than 1:2. When the shunt ratio was increased to 1:1, the mean TN and NH+4-N removal rates were decreased to (71.62 ± 7.49)% and (83.70 ± 10.48)%, respectively, while the concentration of NH+4-N in the effluent was increased. When the shunt ratio was greater than 1:2, the reduced rate of nitrogen removal should be mainly attributed to the increased size of the secondary feeding, which resulted in the insufficient CT between the substratum and wastewater to ensure significant nitrogen removal during Flood Phase 2. The results also indicated that nitrogen in the effluent of each TFCW existed mainly in the form of NO 2 -N as Flood Phase 1 ended even though a high NH+4-N removal rate could be observed, indicating that the oxygen transfer capacity of each system was relatively poor in comparison with that of other TFCWs [6–8,12]. Generally, reaeration capacity of a TFCW is affected by several key factors, including operating conditions, environmental parameters, and compositions, etc. Changes in reaeration effects may be caused by changes in one or two factors that are closely related to the oxygen transfer rates of the system. So, it could be inferred that the reaeration capacity of our TFCW should be attributed to its unique characteristics. The regulation of the DO concentration in TFCWs will be further investigated in the follow-up study. The growth of reeds during the experiment period was monitored, and the heights are shown in Fig. 3. Nitrogen and phosphorus from the influent acted as nutrients for the development of the plants, and all reeds survived and reproduced well without obvious symptoms of toxicity or nutrient deficiency. The growth curves of reeds were almost coincident in each TFCW, and the heights became almost constant [ (175 ± 3) cm] as the experiments ended. The results indicated that the modified step-feeding mode had no obvious adverse effect on the plant growth. And it could be concluded that the effect of plants on nitrogen removal could be neglected in our study due to the less quantity of plant caused by our lab-scale experimental device. The optimal shunt ratio for wastewater treatment by the TFCWs used in this study was 1:2. The mean removal rates at a shunt ratio of 1:2 were as follows: (93.58 ± 11.24)% for TSS, (96.85 ± 3.58)% for COD, (96.23 ± 4.86)% for BOD, (80.20 ± 5.67)% for TN, (89.95 ± 9.01)% for NH+4-N, and (93.41 ± 11.49)% for TP. 3.3. Absolute abundance of nitrogen functional genes in TFCW Considering the roles of nitrogen transformation genes in nitrogen cycling, there should be a significant correlation between the shunt ratio and variation in the nitrogen transforming microbial population in the TFCWs. Thus, we quantified the absolute abundance of key functional genes involved in nitrogen removal at each shunt ratio. Bacterial 16S rRNA, archaeal 16S rRNA, anammox 16S rRNA, amoA, nxrA, narG, napA, nirK, nirS, qnorB, and nosZ were quantified by real-time PCR. The absolute abundance of bacterial 16S rRNA ranged from 1.14  109 to 1.32  109 copiesg1 as the shunt ratio was increased from 0:1 to 1:1 (Fig. 4). Archaeal 16S rRNA showed low abundance (5.32  103 to 8.32  103 copies g1). Although Archaea were not dominant in the microbial community of each TFCW, these organisms may play an important role in nitrogen transformation and removal [28]. The absolute abundance of amoA and anammox bacterial 16S rRNA, which are involved in NH+4-N transformation, was measured (Fig. 4). amoA encodes a monooxygenase and serves as a marker of aerobic ammonia oxidation, in which NH+4-N is oxidized to NO 2 -N [29]. The anammox bacterial 16S rRNA gene is a marker of anaerobic ammonium oxidation, in which NH+4-N and NO 2 -N are

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

7

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

NO3 N ! NO2 N ! NO ! N2 O ! N2 As both narG and napA are involved in the first step of denitrification, the absolute abundance of each gene was summed (narG + napA) to represent the potential for conversion of NO 3 -N into NO 2 -N [33]. The absolute abundance of narG and napA in the TFCWs all exhibited a continuous increase from 2.18  103 and 8.10  102 copies g1 at a shunt ratio of 0:1–1.02  105 and 3.92  103 copies g1 at a shunt ratio of 1:1, respectively, suggesting enhancement of denitrification during Flood Phase 2. The increase in narG and napA genes expression might enhance  NO 3 -N reduction and eliminate NO3 -N accumulation in the TFCWs. Notably, the narG gene was much more abundant than napA at each shunt ratio, which should be attributed to the species of nitrogen transforming microorganism and the operational conditions of the TFCWs. Membrane-bound nitrate reductase (Nar) and periplasmic nitrate reductase (Nap) are two different types of dissimilative nitrate reductase that may be found in the same or different bacteria. The membrane-bound nitrate reductase encoding gene narG is often used as a marker for anaerobic conversion of NO 3 -N to NO 2 -N [34], whereas the periplasmic nitrate reductase encoding gene napA is often considered as a marker for aerobic conversion   of NO 3 -N to NO2 -N [35]. The reduction of NO2 -N to NO, the second reaction in the denitrification process, is catalyzed by one of two nitrite reductases (Nir): a cytochrome cd1 encoded by nirS or a Cu-containing enzyme encoded by nirK. nirS and nirK are used as markers of nitrite reduction in studies of denitrifying bacterial [36]. The absolute abundance of nirK and nirS increased simultaneously as the shunt ratio was increased. qnorB is involved in the third denitrification step and regarded as a marker of the conversion of NO into N2O [37]. The reduction of N2O to N2 in a reaction catalyzed by nitrous oxide reductase is the final step in the denitrification pathway and can be utilized to control N2O emission [18]. nosZ encodes a nitrous oxide reductase and is used as a marker of complete denitrification [38]. As seen in Fig. 4, the absolute abundance of qnorB and nosZ were all markedly increased from 8.30  102 and 3.65  102 copies g1 at a shunt ratio of 0:1–4.77  104 and 6.01  104 copies g1 at a shunt ratio of 1:1, respectively, indicating reduced N2O emission as a result of denitrification.

Fig. 4. Absolute abundance of nitrogen transformation functional genes in TFCW at five different shunt ratios.

converted to N2 in anoxic environments [30]. The absolute abundance of amoA was greater than that of anammox bacterial 16S rRNA at each shunt ratio, suggesting that aerobic ammonia oxidation was one of the predominant pathways of NH+4-N removal in our TFCWs. Nevertheless, as the shunt ratio was increased from 0:1 to 1:1, the absolute abundance of amoA decreased from 4.13  106 to 3.76  106 copies g1, perhaps because of the decrease in the amount of substrate added at the first feeding and enhancement of the reductive environment in the system during Flood Phase 2. As the shunt ratio was increased from 0:1 to 1:1, the absolute abundance of anammox 16S rRNA increased from 1.07  104 to 7.57  105 copies g1, indicating that anammox was enhanced; therefore, anammox seemed to become increasingly another major NH+4-N elimination pathway in our TFCWs along with the increase of shunt ratio. These findings suggest that increasing the shunt ratio enhanced anammox in the TFCW because it produced anaerobic conditions and increased the abundance of adequate matrixes (NH+4-N and NO 2 -N) during Flood Phase 2, facilitating the reproduction, growth, and activity of anammox bacteria. The nitrite oxidase coding gene nxrA is a marker of oxidation of  NO 2 -N to NO3 -N [31]. The nxrA gene exhibited a variation pattern similar to that of the amoA gene, but it showed decreased abundance (Fig. 4). The similar patterns observed for the amoA and nxrA genes were the result of similar environmental adaptations and ecological interactions by AOB and NOB. Moreover, in consideration of the results described in section 3.2, the lower abundance of nxrA in comparison with that of amoA may have been a result of the nitrification reaction process and the competitive advantages of AOB. Fig. 4 shows the absolute abundance of narG, napA, nirK, nirS, qnorB, and nosZ, which are involved in denitrification. The denitrification pathway from nitrate to nitrogen gas can be divided into four steps as follows [32]:

3.4. Quantitative response relationships between nitrogen transformation rates and nitrogen functional genes In order to identify key functional genes that determine nitrogen transformation rates in the TFCW at different shunt ratios, a series of stepwise regression models were built to offer a linear quantitative measure of the relationships between changes in gene expression and nitrogen transformation rates. Eleven nitrogen functional genes were used as candidate variables in stepwise regression analysis to associate their expression levels with nitrogen transformation rates. Functional gene groups that might illuminate the dynamics of nitrogen transformation processes were also introduced as variables in the stepwise regression analysis (as ratios or summations of the expression levels of different functional genes). Four regression equations describing changes in gene expression

Table 2 Quantitative response relationships between nitrogen transformation rates and functional genes (n = 5). Equations

v(TN) = 345.03 + 15.13[(amoA + anammox 16S rRNA)/bacterial 16S rRNA] + 6.12  10

8

2

bacteria 16S rRNA + 8.52  10 [(nirS + nirK)/bacterial 16S

R2

p-Value

0.999

0.001

0.986 1.000 0.993

0.012 0.004 0.007

rRNA] + 1.52  104(amoA/bacteria 16S rRNA)

v(NH+4-N) = 2.77  107bacterial 16S rRNA + 1.39  105amoA + 8.19[(amoA + anammox 16S rRNA)/bacterial 16S rRNA]  184.90 v(NO3 -N) = 2.09  102nxrA + 59.61[(amoA + nxrA)/(nirS + nirK)] + 19.97[amoA/(narG + napA)] + 4.19 v(NO2 -N) = 9.33  102[(narG + napA)/anammo  16S rRNA] + 1.55  103[amoA/(nirS + nirK)]  7.27  106[(qnorB + nosZ)/bacterial 16S rRNA]  1.62  106(nxrA/bacterial 16S rRNA) + 10.25

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

8

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

 and the transformation rates of TN, NH+4-N, NO 3 -N, and NO2 -N 2 yielded high R values ranging from 0.986 to 1.000 (Table 2). The largest part of the variation in TN transformation rate was explained by four significant variables: [(amoA + anammox 16S rRNA)/bacterial 16S rRNA], bacterial 16S rRNA abundance, (nirS + nirK)/bacterial 16S rRNA, and (amoA/bacteria 16S rRNA). [(amoA + anammox 16S rRNA)/bacterial 16S rRNA] was positively associated with the TN transformation rate, revealing that the CANON process [39] could be one of the main routes for TN removal in our TFCWs. Moreover, [(amoA + anammox 16S rRNA)/ bacterial 16S rRNA] also indicated that changes in the relative abundance, rather than the absolute abundance, of anammox 16S rRNA and amoA were responsible for their relationship with the TN transformation rate. The positive relationship between bacterial 16S rRNA abundance and the TN transformation rate indicated that microbial biomass was strongly associated with TN removal. (amoA/bacteria 16S rRNA) was positively associated with the TN transformation rate. amoA is primarily involved in the nitrification process, thus (amoA/bacteria 16S rRNA) represents the extent of NO 2 -N accumulation, which could be conductive to the subsequent biological processes. Besides, [(nirS + nirK)/bacteria 16S rRNA] was also positively associated with the TN transformation rate, which was the rate-limiting factor for NO 2 -N reduction during the denitrification process. In a CW system, AOB, NOB, and denitrifiers are all involved in the conventional nitrification– denitrification process for nitrogen removal. The complete nitrification and denitrification reactions will be promoted if the percentage of the AOB, NOB, and denitrifiers is almost equal [40]. However, the results indicated that NOB should play a less role in the nitrogen removal of our TFCWs, owing to the occurrence of the significant NO 2 -N accumulation during Flood Phase 1 and the much less abundance of nxrA obtained in the systems in comparison with that of amoA. It seemed only a fraction of the NO 2 -N, produced from the NH+4-N oxidation by AOB, could be further con verted to NO 3 -N by NOB, whereas most of NO2 -N was directly converted to NO, N2O and finally N2, and exhausted from the system by denitrifiers; namely the partial nitrification–denitrification process replaced the conventional nitrification–denitrification process in the nitrogen removal of the TFCWs. Accordingly, the equation indicated that TN removal from the TFCW relied largely on the CANON process and the partial nitrification–denitrification process. The transformation rate of NH+4-N was determined by assessing bacterial 16S rRNA abundance, amoA abundance, and [(amoA + anammox 16S rRNA)/bacterial 16S rRNA]. Bacterial 16S rRNA abundance was positively associated with the NH+4-N transformation rate, indicating that effective NH+4-N removal from the TFCW was dependent to some extent on the presence of an adequate microbial biomass. The abundance of amoA was positively associated with the NH+4-N transformation rate. amoA is involved in NO 2 -N production, thus changes in amoA abundance could reflect  NO 2 -N accumulation caused by nitritation. Increased NO2 -N accu+ mulation is associated with enhanced NH4-N transformation. The [(amoA + anammox 16S rRNA)/bacterial 16S rRNA] ratio was positively associated with the NH+4-N transformation rate, indicating that the CANON process was also responsible for NH+4-N removal, as NH+4-N is directly involved in the CANON process. Therefore, the main NH+4-N transformation pathways included the nitritation process and the coupling of the CANON process with NO 2 -N produced by nitritation. The NO 3 -N accumulation rate was collectively determined by nxrA, [(amoA + nxrA)/(nirS + nirK)], and [amoA/(narG + napA)]. All three variables denoted NO 3 -N accumulation, as amoA and nxrA are involved in nitrification, which is the main process contributing to NO 3 -N production, while narG, napA, nirS, and nirK are involved in NO 3 -N consumption in the denitrification process. The main

pathway for NO 3 -N removal should be denitrification. Although  anammox could remove NO 3 -N when NO3 -N was converted to NO -N by narG/napA, some research has also demonstrated that 2 aerobic ammonium oxidation, rather than nitrate reduction, provides a direct local source of nitrite for anammox in the suboxic zone [41]. The largest part of the variation in the NO 2 -N accumulation rate was jointly determined by [(narG + napA)/anammox 16S rRNA], [amoA/(nirS + nirK)], [(qnorB + nosZ)/bacterial 16S rRNA], and (nxrA/bacterial 16S rRNA). Both [(narG + napA)/anammox 16S rRNA] and [amoA/(nirS + nirK)], which denoted NO 2 -N accumulation, were positively related with the NO 2 -N accumulation rate. [(qnorB + nosZ)/bacterial 16S rRNA] and (nxrA/bacterial 16S rRNA), which represented NO 2 -N consumption in the denitrification and nitrification processes, were negatively related to the NO 2 -N accumulation rate. This equation suggested an ecological and functional interaction between nitrifiers, denitrifiers, and anammox bacteria, and all the three kinds of microorganisms could contribute to NO 2 -N transformation in TFCWs; furthermore, it could be inferred that simultaneous nitrification, anammox, and denitrification (SNAD) processes might occur in the system. In general, the abundance of functional genes that play a significant role in the proper functioning and maintenance of wetland systems is mainly dependent on environmental parameters, wastewater properties, and operating conditions. Since certain nitrogen functional genes acted as key factors regulating the transformation rate of each form of nitrogen, changes in nitrogen removal rates can be caused by changes in one or two factors that are closely related to the abundance of genes involved in nitrogen transformation [42]. Hence, variations in nitrogen transformation rates throughout the experiment could be attributed to fluctuations in the absolute abundance of nitrogen functional genes caused by differences in the microenvironment (e.g. the distribution of DO and organics) of the bed at different shunt ratios. Our findings indicated that nitrifying, anammox, and denitrifying microorganisms, all of which were identified at the molecular level in each TFCW, were involved in nitrogen transformation. Increasing the shunt ratio enriched six functional genes (i.e. anammox bacterial 16S rRNA, narG, nirS, nirK, qnorB, and nosZ) involved in denitrification and anammox, enhancing these processes. As step-feeding was utilized with a suitable shunt ratio, the presence of multiple pathways (mainly the CANON process and the partial nitrification–denitrification process) operating to remove nitrogen should best enhance complete nitrogen removal in the single-stage TFCW, thus achieving high TN removal efficiency.

4. Discussion In our study, a single-stage TFCW with a modified step-feeding mode was constructed with the goal of improving TN removal. The quantities of denitrifiers and anammox bacteria in the system were increased since the shunt ratio was greater than 0:1, facilitating the enhancement of the CANON process and the partial nitrifica tion–denitrification process for nitrogen removal. TFCWs used in this study removed nitrogen most effectively when the shunt ratio was 1:2, and the mean TN removal rate reached up to (80.20 ± 5.67)% correspondingly. Similar results could be found in other studies [43,44], which enhance nitrogen removal in intermittent-aerated subsurface flow constructed wetlands. Intermittent artificial aeration, which is considered as one of the most common operation strategies for the intensifications of nitrogen removal, is an effective method to achieve high TN removal by providing alternate aerobic/anaerobic conditions for the simultaneously occurring nitrification and denitrification. What is more, intermittent aeration is much energy-economic than continuous

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

mode. In comparison with the intermittent-aerated CWs, our TFCWs operated with no need for artificial aeration on the premise of the high TN removal efficiencies, furthermore, designers needn’t consider the fouling of air diffusers within CWs and the provisions for the cleaning or replacement of the diffuser assemblies [13]. The successful development of the TFCW for nitrogen removal demonstrates the potential of such systems, as well as to reduce the costs of aeration and capital construction. However, the modified stepfeeding mode might also have negative effects, such as the complex operational procedures, etc. Overall, still more information is needed to be investigated and explored in our TFCWs. This is the primary report of study, further more studies on improving TN removal in CWs are still going on with full-scale experiments. 5. Conclusions The capacity of a single-stage TFCW with modified step-feeding to treat domestic sewage was assessed at five shunt ratios. The shunt ratio significantly influenced nitrogen removal, but it had little influence on removal of TSS, organics, and TP. A shunt ratio of 1:2 was found to be optimal for pollutant removal; this ratio was particularly effective for TN removal. Molecular biological analyses demonstrated that the abundance of nitrogen transforming bacteria was affected by the shunt ratio; denitrification and anammox were enhanced when the shunt ratio was greater than 0:1. Multiple and complete nitrogen removal pathways were developed in the TFCW at appropriate shunt ratios. The optimal shunt ratio of the TFCW for nitrogen removal was 1:2. Acknowledgments This research was sponsored by the National Natural Science Foundation of China (51508002) and the Natural Science Foundation of Anhui Province (1508085QE99). References [1] J. Vymazal, Constructed wetlands for wastewater treatment: five decades of experience, Environ. Sci. Technol. 45 (2010) 61–69. [2] R. Liu, Y. Zhao, L. Doherty, Y. Hu, X. Hao, A review of incorporation of constructed wetland with other treatment processes, Chem. Eng. J. 279 (2015) 220–230. [3] T. Saeed, G. Sun, A review on nitrogen and organics removal mechanisms in subsurface flow constructed wetlands: Dependency on environmental parameters, operating conditions and supporting media, J. Environ. Manage. 112 (2012) 429–448. [4] G. Sun, Y. Zhu, T. Saeed, G. Zhang, X. Lu, Nitrogen removal and microbial community profiles in six wetland columns receiving high ammonia load, Chem. Eng. J. 203 (2012) 326–332. [5] Y. Ding, X. Song, Y. Wang, D. Yan, Effects of dissolved oxygen and influent COD/ N ratios on nitrogen removal in horizontal subsurface flow constructed wetland, Ecol. Eng. 46 (2012) 107–111. [6] C. Li, S. Wu, R. Dong, Dynamics of organic matter, nitrogen and phosphorus removal and their interactions in a tidal operated constructed wetland, J. Environ. Manage. 151 (2015) 310–316. [7] L. Li, C. He, G. Ji, W. Zhi, L. Sheng, Nitrogen removal pathways in a tidal flow constructed wetland under flooded time constraints, Ecol. Eng. 81 (2015) 266– 271. [8] S. Wu, D. Zhang, D. Austin, R. Dong, C. Pang, Evaluation of a lab-scale tidal flow constructed wetland performance: oxygen transfer capacity, organic matter and ammonium removal, Ecol. Eng. 37 (2011) 1789–1795. [9] X. Ju, S. Wu, Y. Zhang, R. Dong, Intensified nitrogen and phosphorus removal in a novel electrolysis-integrated tidal flow constructed wetland system, Water Res. 59 (2014) 37–45. [10] Y. Zhao, S. Collum, M. Phelan, T. Goodbody, L. Doherty, Y. Hu, Preliminary investigation of constructed wetland incorporating microbial fuel cell: batch and continuous flow trials, Chem. Eng. J. 229 (2013) 364–370. [11] Y. Hu, X. Zhao, Y. Zhao, Achieving high-rate autotrophic nitrogen removal via Canon process in a modified single bed tidal flow constructed wetland, Chem. Eng. J. 237 (2014) 329–335. [12] W. Zhi, G. Ji, Quantitative response relationships between nitrogen transformation rates and nitrogen functional genes in a tidal flow constructed wetland under C/N ratio constraints, Water Res. 64 (2014) 32–41.

9

[13] S. Wu, P. Kuschk, H. Brix, J. Vymazal, R. Dong, Development of constructed wetlands in performance intensifications for wastewater treatment: a nitrogen and organic matter targeted review, Water Res. 57 (2014) 40–55. [14] Z. Wang, C. Liu, J. Liao, L. Liu, Y. Liu, X. Huang, Nitrogen removal and N2O emission in subsurface vertical flow constructed wetland treating swine wastewater: effect of shunt ratio, Ecol. Eng. 73 (2014) 446–453. [15] F. Li, L. Lu, X. Zheng, H.H. Ngo, S. Liang, W. Guo, X. Zhang, Enhanced nitrogen removal in constructed wetlands: effects of dissolved oxygen and stepfeeding, Bioresour. Technol. 169 (2014) 395–402. [16] L. Doherty, Y. Zhao, X. Zhao, W. Wang, Nutrient and organics removal from swine slurry with simultaneous electricity generation in an alum sludge-based constructed wetland incorporating microbial fuel cell technology, Chem. Eng. J. 266 (2015) 74–81. [17] APHA, Standard Methods for the Examination of Water and Wastewater, Amer. Public Health Assoc., Washington, DC, USA, 2002. [18] G. Ji, W. Zhi, Y. Tan, Association of nitrogen micro-cycle functional genes in subsurface wastewater infiltration systems, Ecol. Eng. 44 (2012) 269–277. [19] H. Dong, Z. Qiang, T. Li, H. Jin, W. Chen, Effect of artificial aeration on the performance of vertical-flow constructed wetland treating heavily polluted river water, J. Environ. Sci. 24 (2012) 596–601. [20] J. Vymazal, The use of hybrid constructed wetlands for wastewater treatment with special attention to nitrogen removal: a review of a recent development, Water Res. 47 (2013) 4795–4811. [21] T. Saeed, B. Paul, R. Afrin, A. Al-Muyeed, G. Sun, Floating constructed wetland for the treatment of polluted river water: a pilot scale study on seasonal variation and shock load, Chem. Eng. J. 287 (2016) 62–73. [22] S. Wu, X. Dong, Y. Chang, P.N. Carvalho, C. Pang, L. Chen, R. Dong, Response of a tidal operated constructed wetland to sudden organic and ammonium loading changes in treating high strength artificial wastewater, Ecol. Eng. 82 (2015) 643–648. [23] Z. Wang, J. Dong, L. Liu, G. Zhu, C. Liu, Screening of phosphate-removing substrates for use in constructed wetlands treating swine wastewater, Ecol. Eng. 54 (2013) 57–65. [24] G. Sun, K.R. Gray, A.J. Biddlestone, S.J. Allen, D.J. Cooper, Effect of effluent recirculation on the performance of a reed bed system treating agricultural wastewater, Process Biochem. 39 (2003) 351–357. [25] S. Ge, S. Wang, X. Yang, S. Qiu, B. Li, Y. Peng, Detection of nitrifiers and evaluation of partial nitrification for wastewater treatment: a review, Chemosphere 140 (2015) 85–98. [26] B.K. Pathak, F. Kazama, Y. Saiki, T. Sumino, Presence and activity of anammox and denitrification process in low ammonium-fed bioreactors, Bioresour. Technol. 98 (2007) 2201–2206. [27] S. Sri Shalini, K. Joseph, Nitrogen management in landfill leachate: application of SHARON, ANAMMOX and combined SHARON–ANAMMOX process, Waste Manage. 32 (2012) 2385–2400. [28] G. Angnes, R. Nicoloso, M. Da Silva, P. de Oliveira, M. Higarashi, M. Mezzari, P. Miller, Correlating denitrifying catabolic genes with N2O and N2 emissions from swine slurry composting, Bioresour. Technol. 140 (2013) 368–375. [29] H.M. Dionisi, A.C. Layton, G. Harms, I.R. Gregory, K.G. Robinson, G.S. Sayler, Quantification of Nitrosomonas oligotropha-like ammonia-oxidizing bacteria and Nitrospira spp. from full-scale wastewater treatment plants by competitive PCR, Appl. Environ. Microbiol. 68 (2002) 245–253. [30] A. Mulder, A.A. Van de Graaf, L. Robertson, J. Kuenen, Anaerobic ammonium oxidation discovered in a denitrifying fluidized-bed reactor, FEMS Microbiol. Ecol. 16 (1995) 177–183. [31] F. Poly, S. Wertz, E. Brothier, V. Degrange, First exploration of Nitrobacter diversity in soils by a PCR cloning-sequencing approach targeting functional gene nxrA, FEMS Microbiol. Ecol. 63 (2008) 132–140. [32] X. Zhu, Y. Chen, Reduction of N2O and NO generation in anaerobic-aerobic (low dissolved oxygen) biological wastewater treatment process by using sludge alkaline fermentation liquid, Environ. Sci. Technol. 45 (2011) 2137– 2143. [33] D. Bru, A. Sarr, L. Philippot, Relative abundances of proteobacterial membranebound and periplasmic nitrate reductases in selected environments, Appl. Environ. Microbiol. 73 (2007) 5971–5974. [34] J.C. López-Gutiérrez, S. Henry, S. Hallet, F. Martin-Laurent, G. Catroux, L. Philippot, Quantification of a novel group of nitrate-reducing bacteria in the environment by real-time PCR, J. Microbiol. Methods 57 (2004) 399– 407. [35] K. Deiglmayr, L. Philippot, U.A. Hartwig, E. Kandeler, Structure and activity of the nitrate-reducing community in the rhizosphere of Lolium perenne and Trifolium repens under long-term elevated atmospheric pCO2, FEMS Microbiol. Ecol. 49 (2004) 445–454. [36] E. Kandeler, K. Deiglmayr, D. Tscherko, D. Bru, L. Philippot, Abundance of narG, nirS, nirK, and nosZ genes of denitrifying bacteria during primary successions of a glacier foreland, Appl. Environ. Microbiol. 72 (2006) 5957–5962. [37] G. Braker, J.M. Tiedje, Nitric oxide reductase (norB) genes from pure cultures and environmental samples, Appl. Environ. Microbiol. 69 (2003) 3476–3483. [38] B. Stres, I. Mahne, G. Avguštin, J.M. Tiedje, Nitrous oxide reductase (nosZ) gene fragments differ between native and cultivated Michigan soils, Appl. Environ. Microbiol. 70 (2004) 301–309. [39] A.O. Sliekers, N. Derwort, J.C. Gomez, M. Strous, J. Kuenen, M. Jetten, Completely autotrophic nitrogen removal over nitrite in one single reactor, Water Res. 36 (2002) 2475–2482.

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060

10

Z. Wang et al. / Chemical Engineering Journal xxx (2016) xxx–xxx

[40] G. Ji, C. He, Y. Tan, The spatial distribution of nitrogen removal functional genes in multimedia biofilters for sewage treatment, Ecol. Eng. 55 (2013) 35–42. [41] S. Wang, G. Zhu, Y. Peng, M.S. Jetten, C. Yin, Anammox bacterial abundance, activity, and contribution in riparian sediments of the Pearl River estuary, Environ. Sci. Technol. 46 (2012) 8834–8842. [42] M. Truu, J. Juhanson, J. Truu, Microbial biomass, activity and community composition in constructed wetlands, Sci. Total Environ. 407 (2009) 3958–3971.

[43] J. Fan, S. Liang, B. Zhang, J. Zhang, Enhanced organics and nitrogen removal in batch-operated vertical flow constructed wetlands by combination of intermittent aeration and step feeding strategy, Environ. Sci. Pollut. Res. 20 (2013) 2448–2455. [44] J. Fan, B. Zhang, J. Zhang, H.H. Ngo, W. Guo, F. Liu, Y. Guo, H. Wu, Intermittent aeration strategy to enhance organics and nitrogen removal in subsurface flow constructed wetlands, Bioresour. Technol. 141 (2013) 117–122.

Please cite this article in press as: Z. Wang et al., Enhanced nitrogen removal and associated microbial characteristics in a modified single-stage tidal flow constructed wetland with step-feeding, Chem. Eng. J. (2016), http://dx.doi.org/10.1016/j.cej.2016.11.060