Nutrient removal and microbial mechanisms in constructed wetland microcosms treating high nitrate/nitrite polluted river water

Abstract

In rivers, nitrate/nitrite concentrations often vary with seasons and locations, and excess nitrogen can cause eutrophication. Constructed wetlands (CWs) have been used as a typical and optimal ecological technology to purify river water. In the present study, nitrogen (N) removal and related microbial mechanisms of treating high nitrate/nitrite polluted river water were explored in CW microcosms. Excellent removal performances were simultaneously achieved with low and stable effluent concentrations of NO₃⁻–N (0.29–0.51 mg L⁻¹), NO₂⁻–N (0.65–1.0 mg L⁻¹), NH₄⁺–N (0.18–0.40 mg L⁻¹), and TN (1.24–1.56 mg L⁻¹) in our experimental and control groups. Based on the mass balance approach, plant uptake eliminated 11–14% of the total N input and sediment storage contributed 5–11% of N removal, indicating assimilation into biomass and sediment might be important sections of N removal besides microbial nitrification and denitrification. According to the 16S rRNA gene sequencing results, nitrate had positive effects on microbial community richness and diversity. Proteobacteria were particularly identified to be the dominant bacterial strains involved in N transformation in CWs and accounted for 37.26–52.99%. The relative abundance of Proteobacteria was highest after adding nitrate. Gamma- and beta-Proteobacteria were probably responsible for nitrate biodegradation. Bacillus and Cyanobacteria were speculated to be responsible for N removal and transformation. Overall, the results in this study could provide suggestions for treating high nitrate/nitrite polluted river water.

---

Introduction

Constructed wetlands (CWs) have been constructed as a continual technology for treating wastewater because of their low operation cost, simple technology and favorable treatment efficiency. Surface flow constructed wetlands (SFCWs) have been intensively studied as they are similar to natural environments.^1,2 SFCWs could remove various pollutants and nutrients from wastewater, especially the excess nutrients of agricultural runoff.^3,4

Non-point sources reportedly occupy around two-thirds of contaminant loading to the surface waters, in which nitrate is a main pollutant.⁵ Excess nitrogen (N) can lead to eutrophication, which has attracted more and more attention in many developing countries.^6,7 In China, over three quarters of the 121 major lakes were eutrophic according to an evaluation in 2014.⁸ The N treatment ability greatly depends on influent load, which is not stable in most wetlands.⁹ Because most of the N load in the lake is brought by rivers, including SFCWs, it is important to investigate the N transformation in nitrate polluted rivers.

In CWs, the N removal and transformation processes contain microbial metabolism, plant uptake, soil adsorption and so on.¹⁰ Among these, microbial nitrification and denitrification is the main pathway. Nitrification is that ammonium (NH₄⁺–N) is converted to nitrate (NO₃⁻–N) and nitrite (NO₂⁻–N).¹¹ During the process of denitrification, NO₃⁻ and NO₂⁻ are reduced to nitric oxide (NO), nitrous oxide (N₂O) and nitrogen gas (N₂).¹¹ So far, only a few studies focused on the N removal in constructed wetland microcosms treating high nitrate or nitrite polluted river.⁵ These researches mainly discussed the relationship between N transformation and the factors of N load, temperature and dissolved oxygen (DO). They measured N removal efficiencies and rates and found N removal rates increased as nitrate loading and temperature increased.^4,12–15 However, very little is known about the impacts of high nitrate/nitrite on bacterial communities of CWs.

To get a clear understanding of N transformation process in N polluted river, lab-scale SFCWs were constructed to deal with relatively high concentrations of nitrate/nitrite wastewater. The main purposes are (1) to evaluate the N (NO₃⁻–N, NO₂⁻–N, NH₄⁺–N and TN) removal performances; (2) to quantify the contributions of different removal pathways to N removal during the experimental period; (3) to quantify the key functional genes that contribute to the N transformation; and (4) to assess the differences in bacterial community diversity, as well as bacterial strains involved in N transformation processes.

Materials and methods

Experimental setup and operation

Nine independent SFCW microcosms were constructed under a translucent shed in Shandong University in Jinan, China. The microcosms were polyethylene tubs (50 cm depth, 40 cm diameter) padding with three-layer substrate: at the bottom filled will 5 cm of gravel rubble (diameter about 5 cm), in the middle was 5 cm gravel (diameter about 3 cm) and above was 25 cm sand (diameter < 3 mm) mixed with wetland sediments. Phragmites australis (P. australis) were transplanted from Baiyun Lake in Shandong Province, China, with a plant density of 50 rhizomes per m². The simulated polluted river water was composed of sucrose, KNO₃, (NH₄)₂SO₄, KH₂PO₄ and micronutrients. The key composition of the influents was shown in Table 1. Three parallels were performed in each group. The hydraulic retention time (HRT) of each wetland was 7 days. The experiment was performed for three months after a two-month pre-culture.

Table 1 Characteristics of influents from the CW microcosm units (mean ± standard deviation, n = 12)

Treatment units	COD mg L^−1	TP mg L^−1	TN mg L^−1	NH4^+–N mg L^−1	NO3^−–N mg L^−1	NO2^−–N mg L^−1
Control	61.34 ± 2.53	0.89 ± 0.03	25.00 ± 0.91	7.90 ± 0.38	15.61 ± 0.41	0
W1	67.28 ± 4.90	0.86 ± 0.05	36.06 ± 1.52	7.63 ± 0.50	30.76 ± 0.38	0
W2	64.30 ± 2.91	0.97 ± 0.03	35.39 ± 0.12	7.57 ± 0.73	15.52 ± 0.41	12.47 ± 0.28

---

Water qualities monitoring

Inflow and outflow of each microcosm were sampled every two days to assess the N removal performance. The concentration of NO₃⁻–N, NO₂⁻–N, NH₄⁺–N and TN were analyzed according to standard methods.¹⁶

Mass balance calculations

The mass balance approach was used to analyze different ways of N removal, including plant assimilation, substrate adsorption and precipitation, N₂O emission and other losses involving N₂ emission.¹⁰ TN in plants and sediments were analyzed at the beginning and the end of experiment with a 3-AA3AutoAnalyzer (BRAN-LUEBBE, INC, German) according to Black et al.¹⁷ N₂O emission was measured using a gas chromatography (GC7890B, Agilent). The mass balance was calculated as described by Wu et al.^18,19

DNA extraction

The sediments were sampled to extract DNA using MOBIO PowerSand™ DNA Isolation Kit (USA). A Nanodrops K5500 UV-vis spectrophotometer (Kaiao, Beijing, China) was used to determine DNA yields.

Quantitative real-time PCR

The absolute abundance 16S rRNA and functional genes involved in nitrification (amoA) and denitrification (narG, nirS, nirK, nosZ) were quantified using extracted DNA by quantitative real-time PCR (qPCR). The detailed information and primers were listed in Table S1 of the ESI.† The qPCR was performed using Roche LightCycler480 (USA). Three times were performed to get a mean gene copy number. The qPCR mixture was composed as described by Kang et al.²⁰ Table S2 in the ESI† listed the programs of qPCR. The quantification of 16S rRNA and functional genes were calculated based on our previous research.²⁰

Sequencing and bioinformatic analysis

The Illumina sequencing was performed at the Chinese National Human Genome Sequencing Center (China). The sequencing method and data analysis were operated following our previous work.^2,21 The data obtained were deposited in the NCBI Sequence Read Archive under BioSample Number SAMN05163091.

Data analysis

The statistical program IBM SPSS statistics 20.0 (USA), with a one-way analysis of variance (ANOVA), was used to perform all statistical analyses. The results were considered statistically significant when P < 0.05.

Results and discussion

Nitrogen removal

The concentrations of different forms of N in influent and effluent throughout the study period are shown in Fig. 1. Compared with the influent, obvious decrease was observed in all forms of N of effluents. The units performed stably after 12^th weeks, and the average effluent concentrations of NO₃⁻–N, NO₂⁻–N, NH₄⁺–N and TN were 0.29–0.51, 0.65–1.0, 0.18–0.40 and 1.24–1.56 mg L⁻¹, respectively. Notably, the results showed no significant difference among control, W1 and W2 (P > 0.05). During the initial weeks of experiment, the relative high N removal performance may be attributed to the adsorption on the substrate. During the later operation, high N removal performance could be due to the plants assimilation and microbial metabolism.^4,22–24 As reported in previous researches, the metabolism of microorganisms played important roles in the removal of inorganic nitrogen compounds.²⁵

Fig. 1 NO₃⁻–N, NO₂⁻–N, NH₄⁺–N and TN concentration variations of influent (line + symbol) and effluent (column) in the wetland microcosms throughout the experimental period.

TN concentrations showed a gradual decrease and reached a steady concentration of below 1.5 mg L⁻¹. The NH₄⁺–N concentration, which was markedly decreased during first three weeks and finally reached below 0.4 mg L⁻¹, was almost the same in all groups. NO₂⁻–N concentrations of effluent were reduced from 1.47 mg L⁻¹ to below 1.0 mg L⁻¹ during the experimental period of W2. In W1 and the control group, there was no NO₂⁻–N in the influent. However, NO₂⁻–N was detected in the effluent. So it is speculated that NO₂⁻–N was generated from denitrification and ammonia oxidation.

In experimental groups, especially W1, the NO₃⁻–N concentrations of effluent were higher than control within the first three weeks, because of the higher influent NO₃⁻–N concentration. Finally, the NO₃⁻–N removal efficiency was promoted in the experiment groups. Similar process was also described by previous researches, in which the effectively NO₃⁻–N removal efficiency was owing to microbial denitrification.^26–28 Besides, N were reported used by plants through taking up by roots.²⁹ Previous studies have showed nitrate is the primary form of plant assimilation.³⁰ Therefore, the promoted removal ability of NO₃⁻–N may be owing to the plant uptake and enhanced denitrification of experimental groups.²⁶ To further quantify the N input and output, mass balance was calculated throughout the experiment to evaluate the N transformation and removal accomplished by nitrification and denitrification, plant uptake and absorption.

Quantification of nitrogen transformation and mass balance calculations

Nitrogen balance of this study was investigated throughout the experiment (Table 2). Proportions of N removed by different pathways among different groups were also investigated (Fig. 2). The total N input into wetlands, which represented the amount of N from influent, was 260.76 ± 12.58, 410.23 ± 18.62 and 393.24 ± 15.44 mg N per m² per day in control, W1 and W2, respectively. Though the N concentration of the effluent had no significant difference in all groups, the main N removal pathway in the experimental and control groups varied in the present study. As shown in Table 2, N in effluent from different groups during full experiments were 13.65–17.68 mg N per m² per day, account for 4–8% of N removal. Similarly, sediments accumulated N varied from 18.39 to 23.81 mg N per m² per day, decreased from 11% (control) to 5% (W1) and 6% (W2) of the total N output. Notably, plant assimilated N ranged from 42.89 to 45.76 mg N per m² per day in the experimental groups, which were obviously higher than that of control (31.82 ± 1.22 mg N per m² per day). This result was in accordance with the result of NO₃⁻–N removal efficiency in Fig. 1. The higher nitrate loading promoted the plant uptake and assimilation. Assimilation into biomass and sediment was speculated an important part in N removal.

Table 2 N mass balance in the wetland microcosm units through the experiment (mean ± standard deviation, n = 3)

Parameter	Treatment unit	Input load (mg N m^−2 d^−1) Influent	Output load (mg N m^−2 d^−1)
			Effluent	Plant	Substrate	N2O	Other ^a
a Other involve N2 emission via nitrification–denitrification process, ammonia volatilization.
Nitrogen	Control	260.76 ± 12.58	17.68 ± 0.52	31.82 ± 1.22	23.81 ± 0.91	9.56 ± 0.51	136.91 ± 7.54
	W1	410.23 ± 18.62	13.65 ± 0.48	45.76 ± 1.05	20.83 ± 1.55	5.45 ± 0.12	287.94 ± 15.26
	W2	393.24 ± 15.44	14.82 ± 0.26	42.89 ± 2.88	18.39 ± 2.24	16.43 ± 0.55	281.10 ± 12.59

---

Fig. 2 Proportion of N removed by different pathways among different constructed wetland microcosms during the experimental period.

Besides, the others, such as N₂ emission, ammonia volatilization and measurement errors probably occurring in the experiments, ranged from 136.91 ± 7.54 to 287.94 ± 15.26 mg N per m² per day, which increased from 62% (control) to 77% (W1) and 75% (W2), respectively. It was noteworthy that the addition of nitrate and nitrite increased the N₂ emission. The reason may be the relatively high nitrate and nitrite, which could be treated by CWs and transformed into N₂ by microbial nitrification and denitrification. It could also be speculated that nitrification and denitrification were the key factors involving in wetland N removal, which is in agreement with the previous standpoints reported by Jamieson et al.³¹ Nitrification and denitrification are significant mechanisms for N removal in CWs, which are dominant reactions in N cycle.

The emission of N₂O was detected in CWs, in the present study, about 1–4% of influent N was lost as N₂O emission. Noticeably, compared with control, the N₂O emission of 5.45 ± 0.51 N per m² per day was inhibited in W1, while the N₂O emission of 16.43 ± 0.55 N per m² per day of W2 was strengthened. The results were supported by Beline et al., who mentioned that nitrite accumulation raised the production of N₂O.³² Moreover, Dong et al. also reported that the generation of N₂O could be enhanced by adding nitrite.³³ As described by Jamieson et al., denitrification could not occur if NO₃⁻–N is not in adequate supply.²⁶ Therefore, it could be explained that the addition of nitrate enhanced the transformation from N₂O to N₂ by denitrification, while the accumulation of NO₂⁻–N stimulated N₂O emission or inhibited N₂O reduction.³⁴ To further investigate the N transformation and removal, the expression level of N transformation functional genes of the substrate in CWs were studied using real-time quantitative PCR.³⁵

Nitrogen removal mechanisms

The quantity of nitrogen related genes. The quantification of amoA, narG, nirK, nirS, nosZ, and 16S rRNA genes was performed to evaluate the N transformation process in all groups (Fig. 3). The amounts of 16 s rRNA of the control and experimental groups showed no significant difference (P > 0.05). The amoA, narG, nirS, nirK and nosZ genes were usually used as markers of nitrification and denitrification, which were all related to NH₄⁺–N, NO₂⁻–N and NO₃⁻–N transformation. In general, the amounts of denitrification functional genes of W1 were larger than control, indicating that certain amount of nitrate strengthened the denitrification process.³⁴ The amount of nosZ of W1 (1.16 × 10¹⁰ copies per g soil) was higher than control (5.92 × 10⁹ copies per g soil) (P < 0.05), which was consistent with the results of N₂O emission in the mass balance section (Table 2 and Fig. 2). The nirS in control, W1 and W2 accounted for the majority of all genes with 1.22–3.9 × 10¹⁰ copies per g soil, showed no significant difference (P > 0.05). Besides, the amounts of denitrification functional genes in W1 and control showed no significant difference (P > 0.05), indicated that the addition of nitrite did not change the microbial quantities. However, the N removal performed effectively in W1. Besides the quantities, microbial community structure reportedly played very important factors in N transformation of CWs.¹ The differences in N removal efficiency between experimental and control groups may be attributed to the diversity of microbial community.

Fig. 3 DNA copy numbers for functional genes and 16S rRNA of total bacterial in the substrate.

Microbial community. To further analyze the microbial mechanism, the microbial composition was evaluated through Illumina sequencing analysis of the 16S rRNA gene of the three groups. Each sample contained 30 934–31 819 reads with a read length of 419 bp to 422 bp. Good coverage index of 93.5% to 94.9% suggested that the sequence number was competent to describe the microbial community, with OTUs ranging between 4200 and 4691 (Table 3). According to community richness estimators of Chao and ACE index, the addition of nitrate in W1 could obviously increase the community richness. Shannon and Simpson index were employed to analyze the community diversity, which indicated that the diversity followed the order of W1 > control > W2. Based on these results, addition of nitrate could have a positive impact on bacterial community, which was consistent with the results above (Fig. 1 and 3). The N removal performed well in W2 and the quantities of functional genes in the substrate was larger than control. Notably, Shannon diversity index dropped slightly in W2. The decrease in community diversity revealed the negative effect caused by nitrite addition in CWs. To explain the high N removal efficiency shown in Fig. 1 and Table 2 and qPCR results shown in Fig. 3, community composition was further investigated in the next section.

Table 3 Comparison of phylotype coverage, richness and diversity estimators at a phylogenetic distance of 3%

Sample	OTUs	ACE	Chao	Shannon	Simpson	Coverage
Control	4691	7188	7134	7.017	0.00443	0.938
W1	6822	23 400	15 266	7.489	0.00242	0.935
W2	4200	8730	6737	6.485	0.01653	0.949

---

Bacterial community composition. In addition to bacterial community richness and diversity, community composition is regarded as a significant factor to understand the N removal performance in CWs.¹ Eight different phyla were assigned to analyze bacterial community composition, which is shown in Fig. 4. At the phylum level, Proteobacteria was the dominant and accounted from 37.26 to 52.99% in different groups, with read numbers ranging from 14 268–19 798. Furthermore, the relative abundance of Proteobacteria in W1 was the highest, followed by control and W2. This was in consistent with the N removal efficiency and qPCR results of W1 compared with control, for that many microorganisms related to denitrification belong to Proteobacteria.³⁶ Alpha-, beta- and gamma-Proteobacteria, Firmicutes, and Bacteroidetes were considered as denitrifying representatives, which may contributed to the effective N removal efficiency in experimental groups.

Fig. 4 Bacterial community composition at phylum level in CWs. Sequences that could not be classified into any known group were assigned as unclassified_bacteria.

To further understanding the microbial mechanism of N removal, the response of Proteobacteria through sequencing was further analyzed. In the present study, branches of alpha-, beta-, delta-, and gamma-Proteobacteria were particularly explored (Table 4). The percentage of alpha-Proteobacteria dropped 5.47% in W1 compared with the control, suggesting that alpha-Proteobacteria was sensitive to nitrate. The proportions of gamma- and beta-Proteobacteria in W1 increased 8.17% and 1.60%, respectively, compared to control. Proteobacteria reportedly had a remarkable relationship with nitrogen concentration in sediments, and its beta- and gamma-subdivision played an important role in the nutrient biodegradation.^36,37 The result suggested that nitrate reduction in CWs may be mediated by different gamma- and beta-Proteobacteria, which was coincident with previous research, who reported gamma-Proteobacteria was related in narG diversity in soil environment.³⁸ Therefore, we deduced that gamma- and beta-Proteobacteria were probably responsible for nitrogen biodegradation in W1. In W2, except the percentage of beta-Proteobacteria almost unchanged, the other subdivisions were all decreased by adding nitrite. The results suggested that the growth of Proteobacteria was inhibited by nitrite.³⁹

Table 4 The relative abundance of Proteobacteria detected from different wetland microcosms and the predominant order of Proteobacteria

Class	Control	W1	W2
Alpha	17.40%	11.93%	8.14%
Beta	9.01%	10.61%	9.22%
Delta	9.02%	9.66%	5.85%
Gamma	10.30%	18.47%	7.51%
Unclassified	0.75%	1.55%	6.07%
Total	46.61%	52.99%	37.22%

---

Firmicutes was another important phylum. Obvious increase of the abundance of Firmicutes was observed in experimental groups, especially in W2, compared with the control. Relative abundance of Firmicutes increased from 3.89% to 20.89% in W2. After further analysis, Bacillus, was the most abundant genus under genus level. Some species of Bacillus were reported as anoxic denitrification bacteria.⁴⁰ So we deduced that Bacillus was probably responsible for no significant difference in N removal performance in W2, which had the lowest bacteria richness.

Interestingly, the distinguishing trend was noted in W1 and W2 of the relative abundance of Cyanobacteria. Compared with the control (0.53%), notably increase (15.23%) was obtained in W2, while no significance difference was observed in W1 (0.88%). Cyanobacteria was reported involved in nitrogen cycling.⁴¹ Association of Cyanobacteria and bacteria is more effective than single microorganism in removing nutrients from wastewater.⁴² Therefore, we deduced that Cyanobacteria could be another reason for N removal and transformation in W2.

Relative abundance of Acidobacteria was only second to Proteobacteria in the control group. Dramatically decrease was observed in W1 and W2. Moreover, the percentage of Bacteroide increased from 3.11% to 13.18% and 14.86%, respectively, in W1 and W2. The proportions of unclassified bacteria ranged from 7.13% to 10.32% in all CWs, representing sequences could not be categorized into any known group.

Sequences representative of denitrifying bacteria, nitrite oxidizing bacteria (NOB) and ammonia oxidizing bacteria (AOB) were all detected in CWs as shown in Fig. 5. Among them, denitrifying bacteria were the predominant class, involving approximately 1.9% in control, 8.5% in W1, and 3.6% in W2. The denitrifying bacteria contain Paracoccus, Alcaligenes, Pseudomonas and Bacillus.⁴³ The predominant orders in this study were Pseudomonas and Bacillus. Pseudomonas belongs to gamma-Proteobacteria and Bacillus belongs to Firmicutes.⁴⁰ Compared with the control, great increase of OTUs were observed in W1 (314%) and W2 (93%). Therefore, the dramatically increased relative abundance of denitrifying bacteria could be an evidence of N removal abilities in CWs.

Fig. 5 Relative abundance of different communities.

Conclusion

Lab-scale SFCWs were constructed and operated under variable nitrogen load (260.76 to 410.23 mg N m⁻² d⁻¹). Compared to control, the experimental groups achieved similar and stable N removal efficiency. Microbial diversity and composition variation under nitrate/nitrite addition indicated that nitrate had a positive impact on the microbial community structure of wetlands. High relative abundance of denitrifying bacteria could be responsible for N removal abilities in experimental groups. Gamma- and beta-Proteobacteria were probably the main reasons for nitrate biodegradation, while Bacillus and Cyanobacteria were supposed to contribute to nitrite transformation and removal. This study suggests that high nitrate/nitrite polluted river could be purified by CWs through microbial nitrification and denitrification, assimilation into biomass and sediment.

Acknowledgements

This work was supported by the National Science Foundation of China (51578321, and 21307078), National Water Special Project (2012ZX07203-004), Shandong Provincial Natural Science Foundation, China (ZR2015BM004) and the Independent Innovation Foundation of Shandong University (2014JC023).

References

1. M. Truu, J. Juhanson and J. Truu, Sci. Total Environ., 2009, 407, 3958–3971.
2. J. Liu, J. Wang, C. Zhao, A. G. Hay, H. Xie and J. Zhan, Appl. Microbiol. Biotechnol., 2016, 100, 1459–1467.
3. J. Vymazal, H. Brix, P. Cooper, M. Green and R. Haberl, Constructed wetlands for wastewater treatment in Europe, Backhuys Leiden, 1998.
4. J. Vymazal, Sci. Total Environ., 2007, 380, 48–65.
5. M. W. Beutel, C. D. Newton, E. S. Brouillard and R. J. Watts, Ecol. Eng., 2009, 35, 1538–1546.
6. J. Zhang, X. Huang, C. Liu, H. Shi and H. Hu, Ecol. Eng., 2005, 25, 419–428.
7. P. Nyenje, J. Foppen, S. Uhlenbrook, R. Kulabako and A. Muwanga, Sci. Total Environ., 2010, 408, 447–455.
8. H. Wang, G. Ji and X. Bai, Bioresour. Technol., 2016, 209, 100–107.
9. H. Wu, J. Fan, J. Zhang, H. H. Ngo, W. Guo, Z. Hu and S. Liang, Bioresour. Technol., 2015, 176, 163–168.
10. H. Wu, J. Zhang, R. Wei, S. Liang, C. Li and H. Xie, Environ. Sci. Pollut. Res., 2013, 20, 443–451.
11. T. Saeed and G. Sun, J. Environ. Manage., 2012, 112, 429–448.
12. J. Aber, W. McDowell, K. Nadelhoffer, A. Magill, G. Berntson, M. Kamakea, S. McNulty, W. Currie, L. Rustad and I. Fernandez, BioScience, 1998, 48, 921–934.
13. J. Mitsch, A. J. Horne and R. W. Nairn, Ecol. Eng., 2000, 14, 1–7.
14. T. Sirivedhin and K. A. Gray, Ecol. Eng., 2006, 26, 167–181.
15. G. Phipps and W. G. Crumpton, Ecol. Eng., 1994, 3, 399–408.
16. A. P. H. Association, A. W. W. Association, W. P. C. Federation and W. E. Federation, Standard methods for the examination of water and wastewater, American Public Health Association, 1915.
17. C. A. Black, D. Evans and R. C. Dinauer, Methods of soil analysis, American Society of Agronomy Madison, WI, 1965.
18. J. Wu, J. Zhang, W. Jia, H. Xie, R. R. Gu, C. Li and B. Gao, Bioresour. Technol., 2009, 100, 2910–2917.
19. J. Wu, J. Zhang, W. Jia, H. Xie and B. Zhang, Front. Environ. Sci. Eng. China, 2009, 3, 241–247.
20. Y. Kang, J. Zhang, H. Xie, Z. Guo, P. Li, C. Cheng and L. Lv, RSC Adv., 2016, 6, 34841–34848.
21. C. Zhao, H. Xie, J. Xu, X. Xu, J. Zhang, Z. Hu, C. Liu, S. Liang, Q. Wang and J. Wang, Sci. Total Environ., 2015, 505, 633–639.
22. R. H. Kadlec and S. Wallace, Treatment wetlands, CRC press, 2008.
23. J. Fisher and M. Acreman, Hydrol. Earth. Syst. Sci. Discuss., 2004, 8, 673–685.
24. H. Wu, J. Zhang, C. Li, J. Fan and Y. Zou, Clean: Soil, Air, Water, 2013, 41, 844–850.
25. W. Y. Rick and S. M. Thomas, Curr. Opin. Microbiol., 2001, 4, 307–312.
26. C. C. Tanner and R. H. Kadlec, Ecol. Eng., 2013, 56, 79–88.
27. S. J. Jordan, J. Stoffer and J. A. Nestlerode, Ecosystems, 2011, 14, 144–155.
28. W. J. Mitsch, J. W. Day, L. Zhang and R. R. Lane, Ecol. Eng., 2005, 24, 267–278.
29. T. Kraiser, D. E. Gras, A. G. Gutiérrez, B. González and R. A. Gutiérrez, J. Exp. Bot., 2011, 62, 1455–1466.
30. M. Andrews, J. Raven and P. Lea, Ann. Appl. Biol., 2013, 163, 174–199.
31. T. Jamieson, G. Stratton, R. Gordon and A. Madani, Canadian Biosystems Engineering, 2003, 45, 1.9.
32. F. Beline, J. Martinez, C. Marol and G. Guiraud, Water Res., 2001, 35, 2774–2778.
33. L. F. Dong, D. B. Nedwell, G. J. Underwood, D. C. Thornton and I. Rusmana, Appl. Environ. Microbiol., 2002, 68, 1240–1249.
34. M. Firestone, M. Smith, R. Firestone and J. Tiedje, Soil Sci. Soc. Am. J., 1979, 43, 1140–1144.
35. W. Liang, Z. B. Wu, S. P. Cheng, Q. H. Zhou and H. Y. Hu, Ecol. Eng., 2003, 21, 191–195.
36. K. Heylen, B. Vanparys, L. Wittebolle, W. Verstraete, N. Boon and P. De Vos, Appl. Environ. Microbiol., 2006, 72, 2637–2643.
37. W. Manz, M. Wagner, R. Amann and K.-H. Schleifer, Water Res., 1994, 28, 1715–1723.
38. C. J. Smith, D. B. Nedwell, L. F. Dong and A. M. Osborn, Appl. Environ. Microbiol., 2007, 73, 3612–3622.
39. N. C. Tan, M. J. Kampschreur, W. Wanders, W. L. van der Pol, J. van de Vossenberg, R. Kleerebezem, M. C. van Loosdrecht and M. S. Jetten, Syst. Appl. Microbiol., 2008, 31, 114–125.
40. Z. Shen, Y. Zhou, J. Liu, Y. Xiao, R. Cao and F. Wu, Bioresour. Technol., 2015, 175, 239–244.
41. E. Yergeau, S. Kang, Z. He, J. Zhou and G. A. Kowalchuk, ISME J., 2007, 1, 163–179.
42. S. R. Subashchandrabose, B. Ramakrishnan, M. Megharaj, K. Venkateswarlu and R. Naidu, Biotechnol. Adv., 2011, 29, 896–907.
43. C. A. Carlson and J. L. Ingraham, Appl. Environ. Microbiol., 1983, 45, 1247–1253.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra13929a

---