In this study, a laboratory-scale system combined a vertical flow constructed wetland (VF) with a horizontal flow constructed wetland (HF), which was used to treat the secondary effluent of a wastewater treatment plant. Removal efficiencies of 67.02%, 89.80%, 90.31% and 75.38% were achieved by the system for chemical oxygen demand (COD), ammonium nitrogen (NH4+-N), total nitrogen (TN) and total phosphorus (TP), respectively. The VF showed much higher average loading rates of COD, TP, NH4+-N and TN (7.96 g/m2/d, 0.076 g/m2/d, 0.31 g/m2/d and 0.99 g/m2/d) than in HF (0.65 g/m2/d, 0.016 g/m2/d, 0.25 g/m2/d and 0.50 g/m2/d), during the stable operation period. Biodegradation played a major role in pollutant removal, especially for COD and TN. The results of bacterial community analysis indicated that heterotrophic denitrifying bacteria (Hydrogenophaga and Flavobacterium) were the dominant contributors for nitrogen removal in the VF, while heterotrophic denitrifying bacteria (Rhodobacter, Flavobacterium and Dechloromonas) and the autotrophic denitrifying bacteria Sulfurimonas played the principal roles for nitrogen removal in the HF. Redundancy analyses showed that COD and NH4+-N were the important factors affecting the distribution of nitrogen removal bacteria in the VF, while pH, dissolved oxygen and oxidation-reduction potential were the key factors influencing the distribution of nitrogen removal bacteria in the HF.
The secondary effluents from wastewater treatment plants (WWTPs) with relatively high nutrients can lead to water quality deterioration and eutrophication in those receiving water bodies (Zhou et al. 2011). In particular, the total nitrogen (TN) concentration in the secondary effluent is commonly in the range of 12.0–15.0 mg/L (Gao et al. 2017a), which is much higher than the surface-water criteria of the Chinese National Surface Water Environmental Quality Standard (2.0 mg/L for grade V, GB3838-2002, 2002). Moreover, surface-water contamination is regarded as a serious problem in China, with 11.5% severely polluted freshwater and 26.2% slightly polluted freshwater in major lakes and reservoirs (Gao et al. 2017a). Hence, advanced treatment of secondary effluent is of great importance, especially for TN removal.
Advanced tertiary treatments for secondary effluents from WWTPs have been developed, such as activated carbon absorption (Pramanik et al. 2015) and membrane-based technologies (Yang et al. 2015). However, these advanced tertiary treatment technologies require a substantial outlay and high operating and maintenance costs. By comparison, constructed wetlands (CWs) as an ecologically efficient solution seem to be more feasible for secondary effluent treatment, with advantages in terms of low cost, easy operation and maintenance as well as limited secondary pollution. Two types of CWs, classified according to water flow–vertical flow constructed wetland (VF) and horizontal flow constructed wetland (HF) – are extensively used. A combined vertical flow and horizontal flow constructed wetland (VF-HF) system includes the advantages of two types of CWs to provide appropriate aerobic and anaerobic conditions for organic matter and TN removal (Vymazal & Kröpfelová 2011, 2015). Secondary effluent is characterized by high nutrients, high refractory organic matter and varied water quality and quantity, which are difficult to treat effectively in single CWs (Xu et al. 2016). Therefore, the multi-stage system of VF and HF was developed for the purpose of pollutant removal from secondary effluent to produce better quality effluent (Xu et al. 2016).
Previous investigations have shown that pollutants in secondary effluent, such as chemical oxygen demand (COD), TN, ammonia nitrogen (NH4+-N) and total phosphorus (TP), can be removed by physical, biochemical and the combination of physical and biochemical processes. Microbial activities in biochemical processes in the CWs are recognized as a major contributor to the removal of pollutants in wastewater, especially for nitrogen (Wu et al. 2016). However, microbial community structure and their metabolic activities can be easily affected by changing operation conditions such as pH, dissolved oxygen (DO), temperature, carbon source and nitrogen oxides, etc. (Wu et al. 2016; Xu et al. 2016).
Therefore, a laboratory-scale combined VF-HF system was developed for secondary effluent treatment. The main objectives of this investigation were to (1) evaluate the contaminant removal efficiencies of the VF-HF system for secondary effluent treatment of WWTPs, (2) explore the microbial community distribution in the system, and (3) gain insight into the relationships between the active microbial populations involved in nitrogen removal with the associated environmental variables.
MATERIAL AND METHODS
Experimental system and conditions
The laboratory-scale combined VF-HF system was located at Jiangnan University (120°27′39.20″E, 31°49′02.51″N) in Wuxi City (China), where a subtropical monsoon climate predominates. The system consisted of a regulating tank, followed by a VF unit and an HF unit in series, as is shown in Figure S1 (available with the online version of this paper). The regulating tank (500 L) was used for the homogenization of the influent, which was the secondary effluent of a WWTP in Wuxi City. The VF was a cylinder (Φ 60 cm × 100 cm) made of stainless steel, with an effective volume of 280 L. From bottom to top, the VF was filled with 30 cm gravel with particle size 10–30 mm, 30 cm active carbon with particle size 5–10 mm, and 30 cm zeolite with particle size 4–5 mm. The HF was a rectangular tank (100 cm × 50 cm × 50 cm) also made of stainless steel, with an effective volume of 250 L. The HF was filled with 20 cm zeolite with particle size 4–5 mm as the bottom layer, and 20 cm quartz sand with particle size 2–4 mm as the upper layer.
The secondary effluent of a nearby WWTP with inverted A2/O process was pumped into the regulating tank, then was further continuously guided into the VF-HF system, with a daily treatment capacity of 60 L. In this process, the hydraulic loading rates of the VF and HF were calculated as 21.23 cm/day and 12.00 cm/day, respectively. The water quality of the secondary effluent is shown in Table S1 (available online). The average concentrations of COD, NH4+-N, TN and TP were 62.85 mg/L, 3.86 mg/L, 9.70 mg/L and 0.63 mg/L, respectively, with the COD/TN ratio of 6.48.
Sampling strategy and monitoring
Water samples from each treatment unit such as the effluents of the regulating tank, the effluents of the VF and the effluents of the HF were collected every two days. Water quality parameters such as pH value, DO concentration and oxidation-reduction potential (ORP) value were determined onsite at the time of sample collection. Water samples were transferred to the laboratory for immediate measurement of the following parameters: COD, NH4+-N, nitrate nitrogen (NO3−-N), TN and TP. All the water samples were collected in triplicate.
Additionally, in order to investigate the complete microbial community structures, microbial samples were collected from the water and biofilm on substrates according to Adrados' method (Adrados et al. 2014), at the end of the system operation. Five microbial samples were collected from the water of secondary effluent (WSE) in the regulating tank, the top layer of the VF (VF1-1, 0–30 cm), the bottom layer of the VF (VF1-2, 60–90 cm), the top layer of the HF (HF1-1, 0–20 cm) and the bottom layer of the HF (HF1-2, 20–40 cm), respectively. All the microbial samples were collected in triplicate. The sampling areas are shown in Figure S1.
Water sample analysis
The pH, DO and ORP values of the water samples were immediately measured with a portable multi-parameter water quality analyzer (PRO1020, YSI, USA). After the corresponding standard pretreatment and reagent addition, the concentrations of COD, NH4+-N, NO3−-N, TN and TP were determined according to Standard Methods (APHA 2005).
Microbial community analysis
Total bacterial DNA was extracted from a 250 mg pellet of the above-mentioned samples using MP FastDNA® SPIN kit (MP Biomedicals, USA), according to the manufacturer's protocol. The concentration and purity of DNA were quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher, USA), and the integrity of DNA was evaluated via 1% agarose gel electrophoresis (5 V/cm, 20 min) before it was used for amplification. The V3-V4 region of bacterial 16S rRNA genes was amplified using the universal primers 338 F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806 R (5′-GGACTACHVGGGTWTCTAAT-3′) according to Xu's method (Xu et al. 2016). A mixture of the amplicons was then used for sequencing on the Illumina PE300 at the Genomic Research Center at Shanghai Majorbio Bio-pharm Biotechnology Co. Ltd, Shanghai, China (Wang et al. 2016).
The rarefaction curve and clustering analyses were plotted with R tools. The relationships between microbial community structures and the environmental variables were analyzed using redundancy analysis based on R vegan package. The Monte Carlo permutation test was carried out to ascertain the significance of the relationships between microbial community structures and the selected environmental variables.
RESULTS AND DISCUSSION
Overall performance of organic matter and nutrients removal
The advanced treatment of secondary effluent from a WWTP was investigated in the VF-HF system in this investigation, for the purpose of organic matter and nutrient removal. After ∼44 days of operation, the system became stable with the COD, NH4+-N, TN and TP concentrations of the system effluents below 30.00 mg/L, 1.00 mg/L, 1.00 mg/L and 0.30 mg/L, respectively (Figures 1 and 2). This meant that the effluents of VF-HF system could reach the ‘Surface Water Environmental Quality Standard’ (GB3838-2002) grade IV, in most cases. Specifically, removal characteristics of COD, TP and TN in the system were explained as follows.
Figure 1(a) shows COD concentrations of the influents, VF effluents and HF effluents respectively, during the whole operation period. In the first 10 days of operation, the VF-HF system showed a high average COD removal efficiency of 73.11%, and COD was observed to be mainly removed in the VF. This is most likely due to the adsorption and retention by the substrates such as zeolite and active carbon in the VF (Zhou et al. 2017). Then, COD removal efficiency fluctuated during the prolonged operation time from the 12th day to the 32nd day, which ranged from 26.89% to 90.84%. The decrease of COD removal efficiency may be due to the saturation of substrates' adsorption capacity (Zhou et al. 2017). After 44 days of operation, the treatment performance of the VF-HF system was observed to be stable and biodegradation was presumed to play a dominant role within the system. The system exhibited an overall COD removal efficiency of 67.02% after 44 days, which was higher than Xu et al. (2016) (up to about 55%) under similar climatic conditions. Specifically, 85.58% COD removal occurred in the VF. For example, during this period, the average COD concentrations of the influents, VF effluents and HF effluents were 63.95 mg/L, 26.46 mg/L and 21.09 mg/L, respectively. As shown in Table S2 (available online), the average COD loading rate in the VF (7.96 g/m2/d) was higher than in the HF (0.65 g/m2/d) during the stable operation period. This indicated that the substrates in the VF provided an appropriate place for microbial growth, and thus strengthened COD degradation (Zhou et al. 2017).
The results presented in Figure 1(b) indicate that both the VF and the HF contributed to phosphorus removal. After 44 days of operation, the treatment performance of the VF-HF system was observed to be stable. Mean TP concentrations of influents, and effluents in the VF and the HF were 0.65 mg/L, 0.29 mg/L and 0.16 mg/L, respectively. The VF contributed 55.38% TP removal efficiency individually, and the VF-HF system revealed a 75.38% TP removal efficiency overall. The removal efficiency of TP was similar to Xu et al. (2016) (77.8%). As shown in Table S2, the average TP loading rate in the VF (0.076 g/m2/d) was higher than in the HF (0.016 g/m2/d) during the stable operation period. This suggested that most TP was intercepted in the VF, by the removal pathway of substrates (e.g. active carbon, zeolite) adsorption (Dong et al. 2012).
Figure 2 illustrates the nitrogen removal in the VF-HF system during the operation period. As shown in Figure 2(a), TN removal efficiency increased with the increase of operation time. The overall removal efficiency of TN rose from 48.39% to 83.83% until the 44th day, then further increased to 90.31% at the end of this process (after 44 days), which was higher than that previously reported by Ong et al. (2009) (67%) and Xu et al. (2016) (69.8%). In addition, NO3−-N as the denitrification electron acceptor also showed a similar removal tendency to TN removal, which is shown in Figure 2(b). Results indicated that the microorganisms of nitrification and denitrification played important roles in the VF-HF system for TN removal (Xu et al. 2016). The enrichment and evolution of the bacterial communities continuously developed with the prolonged operation time.
Comparing TN removal in the VF with those in the HF after 44 days, the HF (81.59%) showed much a higher TN removal efficiency than the VF (47.35%). However, from the perspective of loading rates, the average TN loading rate in the VF (0.99 g/m2/d) was higher than in the HF (0.50 g/m2/d) during the stable operation period, especially after 44 days (Table S2). This result was consistent with Tao et al. (2017), in which NO3−-N concentrations in the effluents of the VF and the HF were relatively low with the mean value of 0.24 mg/L and 0.10 mg/L respectively (Figure 2(b)), demonstrating sufficient denitrification in both units.
The NH4+-N removal in the VF-HF system is shown in Figure 2(c). At the beginning of the operation, NH4+-N concentrations of the VF effluents and HF effluents increased with the increase of operation time. The NH4+-N concentrations in the VF and HF effluents reached the maximum values of 4.12 mg/L and 2.17 mg/L on the 18th day and the 24th day, respectively. At this point, the microorganisms of nitrogen removal were not well evolved, and NH4+-N was presumed to be chiefly removed by substrate adsorption (Lu et al. 2016; Zhou et al. 2017). The decrease of NH4+-N removal efficiency might be explained by the saturation of the substrates with the prolonged operation time (Lu et al. 2016; Zhou et al. 2017). With the accumulation and evolution of nitrogen removal microorganisms, NH4+-N removal efficiency increased, especially in the HF. During the stable period (after 44 days), the average NH4+-N concentrations of VF effluents and HF effluents were 2.51 mg/L and 0.40 mg/L, respectively, showing a 84.06% NH4+-N removal rate in the HF and 89.80% in the whole system. The NH4+-N removal efficiency in this system was slightly higher than previously reported by Xu et al. (2016) (82.3%). Moreover, as shown in Table S2, the NH4+-N loading rate (0.31 g/m2/d) in the VF was higher than in the HF (0.25 g/m2/d) during the operation period. This result was consistent with Tao et al. (2017). The VF was principally operated in anaerobic-anoxic conditions, with the average DO concentrations in the top layer of 0.74 mg/L and in the bottom layer of 0.13 mg/L, respectively (Figure S2, available with the online version of this paper). Low NH4+-N removal efficiency in the VF (Figure 2(c)) might be caused by nitrification inhibition with the low DO concentration and ammonification from the organic nitrogen (Vymazal & Kröpfelová 2011; Xia et al. 2013). The HF provided a favorable aerobic environment for NH4+-N nitrification, especially in the upper layer of the HF (DO concentration 3.79 mg/L). Moreover, the anaerobic-anoxic conditions in the bottom of the HF favored denitrification (DO concentration 0.60 mg/L), which led to the high TN removal efficiency. In addition, relatively high organic nitrogen (∼3.22 mg/L) in the influents also could be transformed into NH4+-N through anaerobic ammonification by Pseudomonas (Li et al. 2008). This also might be the main reason for the increase of NH4+-N concentrations in the effluents (Xia et al. 2013) of VF from the 62nd day to the 66th day.
Previous studies have shown that nitrogen is mainly removed by microorganisms in CWs by nitrification-denitrification (Zhong et al. 2015; Wu et al. 2016). Results in the ‘Nitrogen removal’ section also demonstrated that biodegradation played an important role in the VF-HF system. Hence, bacterial communities in the VF and HF were characterized by using 16S rRNA high-through sequencing. Rarefaction analysis was used to standardize and compare the reflected microbial diversity among the samples, recognizing the rationality of samples sequencing data. For the Sobs index (Figure S3(a)), the rarefaction curves approached a plateau, suggesting that further sequencing will result in few additional operational taxonomic units (OTUs). In addition, rarefaction curves in the Shannon index (Figure S3(b)) reached a plateau, indicating that sequences reflect the majority of microbial diversity information among the samples. Therefore, results in Figure S3 verified that the sequencing data are reasonable. (Figure S3 is available online.)
Diversity and clustering analysis of the microbial community
As is presented in Table S3 (available online), 20,432–26,932 OTUs were observed in the five samples, demonstrating that the microbial communities were highly complex in each area of the VF-HF system. Community diversity (Shannon, Simpson), richness (Sobs, ACE, Chao 1) and calculated coverage percentage (Good's coverage) of the samples are also listed in Table S3. Good's coverage indices clustered from the OTUs of the samples were all higher than 0.99, further suggesting that the collected gene sequences could represent the bacterial OTUs in each sample. Results of the indices indicated that the microbial community in the WSE had lower richness and diversity than in the VF-HF system. In addition, the samples from the bottom layers (Sobs 1,451–1,502, ACE 1,767–1,775, Chao 1 1,823–1,826, Shannon 5.77–5.86 and Simpson 0.01–0.01) had obviously higher richness and diversity than those from the top layers (Sobs 819–821, ACE 981–1,051, Chao 1 993–1,044, Shannon 4.87–5.01 and Simpson 0.02–0.03), which showed no obvious difference in the different units.
Hierarchical clustering analysis (Figure S4, available online) suggested WSE, VF1-1 and HF1-1 were in one group, and VF1-2 and HF1-2 consisted of the other group. This meant a significant difference in the microbial community structure between the top and bottom layers in the same units. Results were similar to the result from indices analysis. Environmental factors such as DO distribution (Figure S2) in the units might have contributed to shaping the bacterial community (Wu et al. 2016; Xu et al. 2016).
Classification and identification of the microbial community
In addition to the richness and diversity of the microbial communities, the community composition in the CWs system played critical roles in pollutant removal (Xu et al. 2016). In order to compare differences in the compositions of the microbial community, relative community abundance of the five samples in VF-HF system were characterized at the phylum, class, family and genus levels.
Fourteen major bacteria (relative abundance >1%) at the phylum level are shown in Figure 3(a). Proteobacteria and Bacteroidetes were the top two predominant phyla in the area of the WSE, VF1-1 and HF1-1, accounting for 82.49%, 90.94% and 82.08% of the total microbial communities, respectively. The relative abundance of Firmicutes and Spirochaetae in the WSE were 5.04% and 4.16%, respectively. Actinobacteria and Cyanobacteria also contributed the relatively low proportion of 3.71% and 3.54% to the total microbial communities respectively, in HF1-1. As for those in VF1-2 and HF1-2, Proteobacteria, Bacteroidetes and Parcubacteria were the top three predominant phyla, in total accounting for 64.99% and 68.84% of the microbial communities, respectively. Other bacteria such as Actinobacteria (2.61–5.17%), Cyanobacteria (2.33–5.61%), Gracilibacteria (3.99–5.61%), Firmicutes (1.26–5.37%) and Chloroflexi (1.93–2.66%) showed much lower relative abundance in the area of VF1-2 and HF1-2. As should be pointed out, most of the bacteria involved in nitrification and denitrification belong to Proteobacteria (Gao et al. 2017b). Bacteroidetes and Chloroflexi are mostly denitrifying bacteria (Jin et al. 2014). This meant a great potential for nitrogen removal bacteria in the VF-HF system. In addition, substantial differences in relative abundance were observed in the phylum distribution of the five samples. From the secondary effluent (WSE) to the VF-HF system (VF1-1), the relative abundance of Proteobacteria and Bacteroidetes decreased slightly, and the relative abundance of Firmicutes and Spirochaetae decreased significantly. Parcubacteria, Gracilibacteria and Firmicutes, with relative abundances of 13.32–15.86%, 3.99–5.61% and 1.26–5.37% respectively were more abundant in VF1-2 and HF1-2 than in VF1-1 and HF1-1 (less than 0.5%). Nitrospirae, proven to have the ability to perform nitrification by Zhong et al. (2015), showed a much higher relative abundance of 1.59% in HF1-1 than the other samples. According to DO concentrations in the VF-HF system (Figure S2), HF1-1 was primarily under aerobic conditions, while VF1-1, VF1-2 and HF1-2 were mainly under anaerobic and anoxic conditions. Microorganisms such as Parcubacteria and Gracilibacteria, which are suitable for growth in anaerobic and anoxic environments, might be involved in degradation of organic compounds and denitrification. Nitrospirae suggested high nitrification activities in HF1-1, and led to the high NH4+-N removal efficiency (84.27% in Figure 2(c)) in HF.
At the class level, 25 major bacteria (relative abundance >1%) are shown in Figure 3(b). Betaproteobacteria (32.73%), Gammaproteobacteria (17.46%) and Flavobacteriia (16.41%) were the top three predominant classes in the WSE. Betaproteobacteria (32.73%), Sphingobactriia (14.46%), and Deltaproteobacteria (14.05%) were the top three predominant classes in VF1-1. From the secondary effluent (WSE) to the VF-HF system (VF1-1), the relative abundance of Betaproteobacteria, Sphingobacteriia and Deltaproteobacteria increased, and the relative abundance of Gammaproteobacteria and Flavobacteriia decreased. Sphingobacteriia (16.40%), Betaproteobacteria (15.03%) and Parcubacteria (12.82%) were the top three predominant classes in VF1-2. This indicated that different areas in the VF had obviously different distribution of microbial classes. Besides, Alphaproteobacteria (29.77%), Gammaproteobacteria (21.40%) and Sphingobacteriia (13.74%) were the top three predominant classes in HF1-1. Betaproteobacteria (21.79%), Parcubacteria (11.03%) and Sphingobaceteria (8.50%) were the top three predominant classes in HF1-2. This further showed that there were obvious bacterial community differences in the different units and different areas in the HF. In particular, the denitrifying bacteria in the two samples of the HF, such as Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria and Epsilonproteobacteria with the total relative abundance of 95.30%, were much higher than in the VF (70.68%). Betaproteobacteria, a typical heterotrophic denitrifying bacteria, was observed to be accumulated in the areas of VF1-1 and HF1-2, and showed less abundance in the areas of VF1-2 and HF1-1. Gammaproteobacteria, a typical autotrophic denitrifying bacteria, had mostly accumulated in HF1-1 (21.40%), Epsilonproteobacteria, another typical autotrophic denitrifying bacteria, had mostly accumulated in the areas of HF1-2. Results were well consistent with those in Figure 2(a). High TN removal efficiency in the HF of 81.59% might be due to the high relative abundance of denitrifying bacteria in the HF. Nitrospira, a typical nitrifying bacteria, was clearly detected in HF1-1, with the highest relative abundance of 1.59%, demonstrating higher nitrification in this area. This also could explain the fact that the HF had a higher NH4+-N removal efficiency than the VF (Figure 2(c)).
At the family level, the major bacteria (relative abundance >2%) are shown in Figure 3(c). Flavobacteriaceae (15.78%), Comamonadaceae (12.54%) and Neisseriaceae (8.56%) were the top three predominant families in the WSE. Comamonadaceae (25.54%), Xanthomonadaceae (8.47%) and Desulfomicrobiaceae (8.32%) were the top three predominant families in VF1-1. From the secondary effluent (WSE) to the VF-HF system (VF1-1), the relative abundance of the major bacteria (Comamonadaceae, Desulfomicrobiaceae and Xanthomonadaceae) increased, and the relative abundance of the major bacteria (Flavobacteriaceae and Neisseriaceae) decreased. Comamonadaceae, a possible heterotrophic denitrifying bacteria (Zhong et al. 2015), was the top predominant family in VF1-1 (25.54%), and was less abundant in HF1-2 (6.78%), VF1-2 (5.89%) and HF1-1 (3.65%). Another possible heterotrophic denitrifying bacteria, Rhodocyclaceae, was also observed in all areas of the VF-HF system, with the highest accumulation in HF1-2 (11.84%). The Xanthomonadaceae family (belonging to Gammaproteobacteria) were possible facultative autotrophic denitrifying bacteria (Gao et al. 2017b). The relative abundance of Xanthomonadaceae was 18.89% in HF1-1 and 8.47% in VF1-1, respectively. In all, TN removal processes appeared to occur in the whole VF-HF system. Comamonadaceae was presumed to be the main microbial family for denitrification in the VF, and Xanthomonadaceae as well as Rhodocyclaceae were the dominant microbial families for denitrification in the HF.
At the genus level, the major bacteria (relative abundance >2%) are shown in Figure 3(d). Flavobacterium (15.76%), Deefgea (6.43%) and Aeromonas (6.16%) were the major bacteria in the WSE. From the secondary effluent (WSE) to the VF-HF system (VF1-1), the relative abundance of the major bacteria (Flavobacterium, Deefgea and Aeromonas) decreased, and the Desulfomicrobium, Hydrogenophaga, Saprospiraceae, etc. were enriched in VF1-1. Members of Parcubacteria, Silanimonas, Saprospiraceae, Flavobacterium, Porphyrobacter and Desulfomicrobium were most frequently detected. Parcubacteria showed higher abundance in both VF1-2 (12.82%) and HF1-2 (11.03%), and much lower abundance in VF1-1 (0.02%) and HF1-1 (0.33%) respectively. Silanimonas was the most abundant in HF1-1 (11.99%), and Desulfomicrobium was the most abundant in VF1-1 (8.32%). Overall, different dominant bacteria were distributed in the different CWs and functional areas, showing various mechanisms for pollution removal.
Major function genera for nitrogen removal
In previous studies, the genera of Dechloromonas, Comamonas, Thiobacillus, Nitrosospira, Flavobacterium, Hydrogenophaga and Azospira were considered to play a vital function in the removal of organic matter and nitrogen (Zhong et al. 2015; Wang et al. 2017). The major possible function genera involved in nitrogen removal in the four samples of the VF-HF system (VF1-1, VF1-2, HF1-1 and HF1-2) are summarized in Table 1. Nitrification consisted of two processes: the conversion of ammonium to nitrite by ammonium oxidizing bacteria (AOB) and the conversion of nitrite to nitrate by nitrite oxidizing bacteria (NOB). Nitrosomonas, belonging to AOB, had relatively high affinity constants for ammonia and relatively high maximum activity, which was detected in all the areas in the system (Domingos et al. 2011). The sequences belonging to Nitrosomonas in the HF were higher than in the VF. Hence, the diversity of Nitrosomonas in the HF was higher than in the VF. This indicated that nitrification resistance against perturbation in the HF was higher than that in the VF (Domingos et al. 2011). Nitrospira, belonging to NOB, showed much higher sequences in the HF than in the VF. This also indicated that nitrification resistance against perturbation in the HF was higher than that in the VF (Domingos et al. 2011). The results further demonstrated the high NH4+-N removal efficiency in the HF.
|Nitrification (Zhong et al. 2015)||Nitrosomonas||5||6||3||17|
|Denitrification (Zhong et al. 2015; Gao et al. 2017b; Wang et al. 2017)||Hyphomicrobium||1||0||37||2|
|Nitrification (Zhong et al. 2015)||Nitrosomonas||5||6||3||17|
|Denitrification (Zhong et al. 2015; Gao et al. 2017b; Wang et al. 2017)||Hyphomicrobium||1||0||37||2|
The majority of denitrifying bacteria were enriched in the facultative environment, and utilized the organic compounds as electron donors for nitrogen removal. Table 1 lists the dominant genera which had been characterized as denitrifying bacteria (Zhong et al. 2015; Gao et al. 2017b; Wang et al. 2017). Although the sequences of the genera cannot represent all the denitrifying bacteria, the data were still helpful for revealing denitrifying bacteria distribution in the four sampling areas. It is well known that a large proportion of possible heterotrophic denitrifying bacteria belongs to Betaproteobacteria, such as Comamonadaceae family (the genera of Hydrogenophaga and Acidovorax) and Rhodocyclaceae family (the genera of Azospira and Dechloromonas) (Gao et al. 2017b; Zhong et al. 2015). Alphaproteobacteria, including Rhodobacter, Rhizobium and Hyphomicrobium, etc. were also characterized as the important heterotrophic denitrifying bacteria. In addition, Sulfurimonas, as the most commonly reported autotrophic denitrifying bacteria (Zhong et al. 2015), were also detected in this investigation. In the VF, Hydrogenophaga and Flavobacterium, with sequences of 1,349 and 1,312 respectively, were detected as the dominant denitrifying genera. Hydrogenophaga and Flavobacterium, as the main heterotrophic denitrifying bacteria mainly accumulated in the area of VF1-1. This might be due to the relatively high COD concentration (abundant electron donors) and low DO concentration (anaerobic-anoxic environmental conditions) in the area of VF1-1 (Wu et al. 2016; Xu et al. 2016). In the HF, Rhodobacter, Flavobacterium, Dechloromonas and Sulfurimonas, with sequences of 949, 1,364, 1,359 and 1,233 respectively, were also detected as the dominant denitrifying genera. Rhodobacter and Flavobacterium were the predominant genera in HF1-1, while Dechloromonas and Sulfurimonas were the dominant genera in HF1-2. Heterotrophic and autotrophic denitrifying bacteria showed higher diversity and richness in the HF than in the VF.
Relative influences of water physicochemical properties on microbial community
Microbial community structure, especially nitrifying and denitrifying bacteria, could be easily influenced by wastewater variables such as COD, TN, pH, DO, etc. (Mietto et al. 2015). In this investigation, redundancy analysis was used to reveal the relationship between the dominant bacteria in the four samples of the VF-HF system (VF1-1, VF1-2, HF1-1 and HF1-2) with environmental factors. Dominant bacteria involved in nitrogen removal at family (Comamonadaceae, Xanthomonadaceae and Rhodocyclaceae) and major function genera level (Hydrogenophaga, Dechloromonas, Flavobacterium and Sulfurimonas) were selected as target microorganisms. Environmental factors included the main wastewater indicators concerning nitrogen transformation (COD, NH4+-N and TN) and the environmental physicochemical parameters (pH, DO and ORP) (Table S4, available online). Due to the concentrations of NO3−-N being very low and the fact that the number of environmental factors should be less than the number of samples in redundancy analysis, we selected the NH4+-N and TN for redundancy analysis.
The redundancy analyses (RDA) between water physicochemical properties and the top 15 microbial families from 16S rRNA MiSeq sequencing in the different functional areas within the VF-HF system is shown in Figure 4. As was discussed above, different functional microorganisms accumulated in the different areas of VF-HF system. Moreover, Figure 4(a) suggests that Gemmatimonadaceae showed a positive correlation with TN. Desulfomicrobiaceae and Comamonadaceae showed a positive correlation with COD, while Lentimicrobiaceae showed a positive correlation with NH4+-N. Figure 4(b) suggests that Xanthomonadaceae and Saprospiraceae showed a positive correlation with DO. Xanthomonadaceae, Rhodobacteraceae, Erythrobacteraceae and Chitinophagaceae showed a positive correlation with pH, while Parcubacteria, Gracilibacteria, env.OPS_17 and Anaerolineaceae showed a positive correlation with ORP. For the dominant denitrifying bacteria such as Comamonadaceae, Xanthomonadaceae and Rhodocyclaceae, Figure 4(a) reveals that Comamonadaceae was more positively related to COD and NH4+-N. This indicated that the effects of COD and NH4+-N on nitrogen removal bacteria in the VF was higher than in the HF. Figure 4(b) reveals that Xanthomonadaceae was positively related to pH and DO, and Rhodocyclaceae was positively related to ORP. In addition, DO (p = 0.042) was most correlated to the community variances in the VF-HF system. The significance of these relationships was confirmed by the Monte Carlo permutation test. In conclusion, the effects of pH, DO and ORP on nitrogen removal bacteria in the HF were higher than in the VF.
Figure 5 further shows the interrelation between 14 genera (Table 1) involved in nitrogen removal with the physicochemical properties of the VF and the HF. Results suggested that high sequences of heterotrophic denitrifying bacteria were enriched in the VF, and high sequences of both heterotrophic and autotrophic denitrifying bacteria were enriched in the HF. In addition, higher nitrifying bacteria sequences were also observed to accumulate in the HF. The dominant genera of Hydrogenophaga, Acidovorax, Thauera and Rhizobium accumulated in the VF (Figure 5(a)) were more influenced by COD and NH4+-N than other genera, and Flavobacterium was positively correlated with TN. These results demonstrated that the effects of COD and NH4+-N on nitrogen removal bacteria in the VF were higher than in the HF. Figure 5(b) shows that pH and DO have the greater positive effect on Flavobacterium, Nitrospira, Rhodobacter and Hyphomicrobium than other genera, and ORP exerted an obvious influence on Thiobacillus, Nitrosomonas, Arcobacter and Sulfurimonas. The nitrifying bacteria were principally accumulated in the HF, indicating that the effects of pH, DO and ORP on nitrogen removal bacteria in the HF were higher than in the VF. In brief, the differences in COD, NH4+-N and TN removal within the VF and HF were mainly due to the distribution of pollutants removal microorganisms in the system. Water physicochemical properties such as COD, NH4+-N, pH, DO, ORP, etc. also had a great influence upon the distribution of bacteria, especially nitrogen removal bacteria.
The VF-HF system exhibited high contaminant removal efficiency during treatment of the second effluent from the WWTP. Average removal efficiencies of 67.02% COD, 89.80% NH4+-N, 90.31% TN and 75.38% TP were achieved during the stable operation period. The VF showed much higher average loading rates of COD, TP, NH4+-N and TN (7.96 g/m2/d, 0.076 g/m2/d, 0.31 g/m2/d and 0.99 g/m2/d) than in the HF (0.65 g/m2/d, 0.016 g/m2/d, 0.25 g/m2/d and 0.50 g/m2/d), during the stable operation period. The nitrification-denitrification process was the major removal pathway for nitrogen. Microbial community analysis demonstrated that heterotrophic denitrifying bacteria, such as Hydrogenophaga and Flavobacterium, were the dominant contributors to nitrogen removal in the VF. The heterotrophic denitrifying bacteria, such as Rhodobacter, Flavobacterium and Dechloromonas, as well as the autotrophic denitrifying bacteria Sulfurimonas, played the important roles for nitrogen removal in the HF. Redundancy analysis showed that COD and NH4+-N were the important factors affecting the distribution of nitrogen removal bacteria in VF, while pH, DO and ORP were the key factors influencing the distribution of nitrogen removal bacteria in HF.
This work was supported by the National Water Pollution Control and Treatment Science and Technology Major Project of China (No. 2017ZX07203-003 and No. 2017ZX07204-002) and the Fundamental Research Funds for the Central Universities (No. JUSRP1703XNC).