Optimization and action mechanism of pollutant removal performance of unsaturated vertical flow constructed wetland (UVFCW) driven by substained-release carbon source

UVFCW

Zhuang, Song & Zhang (2020) used hydrophilic material, i.e., cotton micro-fiber with the total length of 4.5 m was spiral filled in the pores of gravels in the tidal unsaturated area of CW_TU (a tidal unsaturated constructed wetland) to extend the hydraulic retention time and enhance the removal efficiency of ammonia nitrogen (NH₄⁺-N). In light of this, the present study added water-conducting materials (rope-shaped artificial water grass, CW_R) to the unsaturated zone to enhance the removal of NH₄⁺-N. Besides, carbon sources were mixed with the common substrate and filled in different zones of CWs according to the carbon source placement method as follows. As shown in Fig. 1, four types of unsaturated constructed wetlands were designed in this study, namely, CW_R (Fig. 1A), CW_R+Cu (carbon source were mixed with the common substrate uniformly, Fig. 1B), CW_R+Ca (carbon source was concentrated at the depth of three-quarters from the bottom of the saturation zone, Fig. 1C) and CW_R+Cb (carbon source was concentrated in the bottom quarter of the saturation zone, Fig. 1D). The actual UVFCW was shown in Figs. 1E and 1F. In this article, the sustained-release carbon source is made by mixing polybutylene adipate terephthalate (PBAT) and corncob for long-term and high-efficiency application (Zhang et al., 2024). The purpose of using sustained-release carbon sources is to continuously release carbon into constructed wetlands to enhance denitrification. We investigated the characteristics of corncob and PBAT. Although corncob has a good carbon-release effect, its release rate was too fast, and the duration of carbon release is not lasting. In contrast, the release rate of carbon by PBAT was slow, which could encapsulate corncob, slowing down the release of carbon from corncob and increasing the its persistence. Previous studies have shown that the persistence of corncob encapsulated by PBAT was about 85 days longer than that of unencapsulated corncob, and the cost was 7 yuan per kg lower than that of pure PBAT.

The cylindrical UVFCW was made of acrylic material, and the chemical oxygen demand (COD), TP, NH₄⁺-N, and NO₃⁻-N in the influent water were set as 50, 0.5, 5, and 10 mg/L, respectively. According to the previously reported method (Zhuang, Song & Zhang, 2020), the trace elements such as calcium chloride (CaCl₂), magnesium sulfate (MgSO₄), copper sulfate (CuSO₄), molybdic acid (H₂MoO₄), and boric acid (H₃BO₃) were added, and the conventional elements, trace elements, and deionized water were mixed at certain concentrations and placed in large plastic barrels to simulate wastewater in wastewater treatment plant. The influent water flow rate of each UVFCW was 4.17 mL/min; the hydraulic load was 0.34 m³/m²·d; the hydraulic retention time (HRT) was 24 h; and the substrate was 3–5 mm gravel. A pierced polyethylene plastic column was placed in the middle of the substrate to monitor the DO of the water body. Considering the low contribution of plants to the pollutant removal from VFCW, no plants were set in this study (Zhao et al., 2011).

Sample collection and analysis

High-throughput sequencing

After the experiment, the materials in the treatment group were mixed and sampled. The mixed samples were stored in a −80 °C environment in a ziplock bag. The high-throughput sequencing of the collected samples was performed by Shanghai Paisenor Co., Ltd. (Shanghai, China). The primers in the bacterial 16S rRNA V3-V4 region were selected on the Illumina MiSeq system to test and analyze the community structure of the samples. The forward and backward primers were 338 F (sequence: 5′-ACTCCTACGGGAGGCAGCAG-3′) and 806 R (sequence: 5′-GGACTACHVGGGTWTCTAAT-3′) (Claesson et al., 2009). The microbial community in the samples was analyzed on the Paisenor Cloud Platform of Shanghai Paisenor Co., Ltd. (Shanghai, China).

Statistical analysis

The data obtained from the experiment were processed using R (version 4.02). The Origin 9 (version 2019b) was used to plot bar graphs of the pollutant concentration in influent and effluent water and pollutant removal efficiency. The “circlize” package in R (version 4.3.2) was employed to plot the abundance chord diagram at the microbial phylum level. The “corrplot”, “ggplot2”, “reshape2”, “psych”, and “pheatmap” packages in R were used to perform inter-group correlation analysis and plot the correlation between environmental parameters and the relative abundances at phylum and genus levels (p < 0.05). The “pheatmap” package was used for plotting heatmap and analyzing the species differences between samples. The “ggplot2” package was used to draw principal component analysis (PCA) diagrams to quantitatively display the degree of difference in sample species composition.

Adding sustained-release carbon source enhances nitrogen and phosphorus removal

In this study, CWs with different unsaturation rates were investigated. With the changes of unsaturated zone, the DO and pH along water pathway also changed (Supplemental Material). Moreover, the various redox condition led to different TN removal. The results indicated that when the unsaturation rate was 50%, COD and TP removal efficiency were desirable, but the removal effects on TN and NO₃⁻-N were poor, whose removal efficiency were both less than 25%. In addition, the C/N ratio of influent water was low (3.5). Based on these results, we speculated that the lack of carbon source in the denitrification zone was the main factor affecting the removal efficiency of TN and NO₃⁻-N.

In order to increase the removal efficiency of TN and NO₃⁻-N, CW_50% was improved, as shown in Fig. 1. As shown in Fig. 2A, the COD removal efficiency in CW_R control group was higher than that of all the sustained-release carbon source treatment groups. The possible reason was that the amount of organic matter released by the sustained-release carbon source was greater than that from aerobic microbial degradation, resulting in excessive organic matter discharge from the system with water (Huang et al., 2020). The COD removal efficiency in CW_R-Cu group was the lowest. As shown in Fig. 2B, the removal efficiency of NH₄⁺-N in CW_R group was higher than that in the sustained-release carbon source treatment groups. Microbial ammonia oxidation is the first step to remove NH₄⁺-N from TN. The removal of NH₄⁺-N mainly depends on the nitrification reaction under aerobic conditions (Li et al., 2019; Vymazal, 2007). It can be inferred that heterotrophic microorganisms will compete for oxygen during degradation of COD and NH₄⁺-N (Li et al., 2018), resulting in a downward trend in the removal of NH₄⁺-N in the sustained-release carbon source treatment groups. Among them, the removal efficiency of NH₄⁺-N in the CW_R-Cb group was the highest, and the removal efficiency of CW_R-Cu was the lowest. As shown in Fig. 2A, the COD concentration in CW_R-Cu was the highest, indicating that the high concentration of COD inhibited the removal of NH₄⁺-N.

The removal efficiency of NO₃⁻-N and TN in the sustained-release carbon source treatment groups were higher than those in CW_R group (Figs. 2C and 2D), indicating that the addition of sustained-release carbon source significantly increased NO₃⁻-N and TN removal efficiency. Our results were consistent with one previous report that adding sustained-release carbon source (calamus + polyhydroxybutyrate-co-valerate (PHBV)) to CW resulted in the effective removal of NO₃⁻-N and TN (Li et al., 2024). Our data also showed that CW_R-Cu treatment exhibited the best removal effect on NO₃⁻-N and TN, which was better than CW_R-Ca and CW_R-Cb treatments, indicating that the placement of the sustained-release carbon source had a significant impact on the removal of pollutants. Zhou et al. (2022) employed sucrose and reed litter as VFCW carbon sources, and TN removal efficiency in their study was 84.9%; Sun et al. (2022) utilized calamus as the sustained-release carbon source in CW, and the NO₃⁻-N removal efficiency in their study was 76.03%; and the removal efficiency of NO₃⁻-N and TN in these two studies were significantly lower than those in our CW_R-Cu treatment, indicating that the sustained-release carbon source in this study had excellent denitrification function.

The removal efficiency of TP in the sustained-release carbon source treatment groups was significantly higher than that of CW_R control group (Fig. 2E), indicating that the addition of sustained-release carbon source can promote the removal of TP. Since the UVFCW substrate was the same, we speculated that with the increasing organic matter concentrations in UVFCW, the adsorption sites of PO₄⁻-P were increased in the sustained-release carbon source treatment group, thus improving the removal efficiency of TP (Liu & Hu, 2019). The removal efficiency of TP in CW_R-Cu treatment group (81.87%) was better than that in CW_R-Ca (69.74%) and CW_R-Cb (58.06%) groups, indicating that the placement of the sustained-release carbon source had a significant impact on the removal of TP. The uniform placement of sustained-release carbon source exhibited the best effect on TP removal.

In summary, the sustained-release carbon source treatment can significantly improve the removal efficiency of NO₃⁻-N, TN, and TP. Different placement methods of sustained-release carbon sources exhibit different removal effects on pollutants. Next, we further investigated the purification effect of the different sustained-release carbon source placement methods on pollutants.

Effects of carbon source placement on pollutant removal

To understand the mechanism of pollutant removal by carbon source placements, we investigated the water quality changes with water flow in UVFCW. As shown in Fig. 3, the COD removal efficiency in CW_R control group along the reactor was higher than that of the sustained-release carbon source treatment groups (Fig. 3A). The COD removal efficiency in CW_R-Cu group decreased from 36.58% to −3.53% with the increasing CW depth, and that in CW_R-Ca and CW_R-Cb treatment groups were the lowest at the sustained-release carbon source placement point (50 and 70 cm from the top of the column), with a COD removal efficiency of −362.80% and −194.73%, respectively. The possible reason might be that the sustained-release carbon source under CW_R-Cu treatment was more widely distributed in space, and the released carbon source had the highest contact time with bacteria for denitrification, resulting in a downward trend in COD removal efficiency in CW_R-Cu. The sustained-release carbon source under CW_R-Ca and CW_R-Cb treatments released a large amount of organic carbon locally, causing excessively high COD concentration for heterotrophic bacteria, lowering the efficiency for denitrification consumption. The above results jointly indicated that the placement method of sustained-release carbon source affected the carbon release amount, and that the sustained-release effect of uniformly-placed carbon source was better than that of concentratedly-placed carbon source.

The NH₄⁺-N removal efficiency (100%) in CW_R control group along reactor (50, 60, 70, and 80 cm from column top) was higher than that in the groups added with sustained-release carbon source (Fig. 3B). The NH₄⁺-N removal efficiency under CW_R-Ca and CW_R-Cb treatments was the lowest at the sustained-release carbon source placement point (50 and 70 cm), which was 77.42% and 88.42%, respectively. The NH₄⁺-N removal efficiency in CW_R-Cu group showed a decreasing trend (from 99.54% to 87.56%) along the reactor (from 50 to 80 cm), indicating that the corn cobs as sustained-release carbon source might have produced NH₄⁺-N, potentially inducing ammonia pollution (Bao et al., 2016). At the same time, microorganisms will compete for oxygen during the degradation of COD and NH₄⁺-N, further reducing the removal of NH₄⁺-N. The removal efficiency of NH₄⁺-N in CW_R-Cb group was generally higher than that in CW_R-Cu and CW_R-Ca groups. As shown in Fig. 3F, the DO in CW_R-Cu and CW_R-Ca groups was 0–1 mg/L, and that in CW_R-Cb was 1.5–3.5 mg/L, indicating that the placement of the sustained-release carbon source affected the DO removal from the entire system. It has been reported that DO < 0.5 mg/L can inhibit the nitrification process in CW (Tao et al., 2021), based on which, we speculated that the low DO in the uniform placement and upper concentrated placement of the sustained-release carbon source might inhibit the nitrification reaction and reduce the removal efficiency of NH₄⁺-N, and that that in lower concentrated placement of the sustained-release carbon source would show an opposite trend.

The removal efficiency of NO₃⁻-N and TN by CW_R-Cu, CW_R-Ca and CW_R-Cb were higher than those by CW_R (Figs. 3C and 3D), indicating that the sustained-release carbon source at different depths (from 50 to 80 cm from column top) played a role and promoted the denitrification reaction. Zheng et al. (2022) used alkaline reed as an external sustained-release carbon source, and the removal efficiency of NO₃⁻-N and TN increased from −15.97% and 53.13% to 95.5% and 93.7%, which also showed that the addition of sustained-release carbon source significantly promoted denitrification. The removal efficiency of NO₃⁻-N and TN by CW_R-Cu at all depths were higher than those by CW_R-Ca and CW_R-Cb, among which, CW_R-Cb had the poorest removal effect on NO₃⁻-N and TN, indicating that the placement of the sustained-release carbon source affected the denitrification. It may be related to the placement of sustained-release carbon sources affecting DO and carbon source distribution. As shown in Fig. 3F, CW_R-Cb was in an aerobic state, which limited its denitrification reaction, resulting in undesirable removal efficiency of NO₃⁻-N and TN in CW_R-Cb. The anaerobic environment in CW_R-Cu was the best, and its COD concentration in the whole system was high (Fig. 3A), and thus the sufficient carbon source and low DO environment promoted the denitrification reaction (Hong et al., 2017), eventually enhancing the removal of NO₃⁻-N and TN by CW_R-Cu.

The changes of TP in different treatment groups at all four depths were shown in Fig. 3E. As the depth of CWs increased, the removal efficiency of TP in all treatment groups showed an upward trend. Among them, the removal effect on TP by CW_R-Cu and CW_R-Ca at each depth was better than that by CW_R and CW_R-Cb, indicating that the placement methods of the sustained-release carbon source had a significant impact on the removal of TP, namely, the uniform placement of the sustained-release carbon source and the concentrated placement at the top promoted the removal of TP, while the concentrated placement at the bottom inhibited the removal of TP. The high removal efficiency of TP might be due to the fast carbon release rate of the uniformly-distributed and upper-concentrated sustained-release carbon source, providing the carbon source for the growth and reproduction of phosphorus-accumulating bacteria. The high-concentration COD zone is conducive to the growth and reproduction of phosphorus-accumulating bacteria, thus increasing the removal efficiency of TP, which was consistent with our observation in “Effects of sustained-release carbon source placement method on nitrogen- and phosphorus-removing microorganisms” that the higher relative abundance of Pseudomonas was observed in CW_R-Cu and CW_R-Ca.

In summary, the sustained-release carbon source placement method mainly affects the removal of pollutants by affecting the distribution of carbon sources and DO in CWs. Our results reveal that the uniform placement of sustained-release carbon source can significantly improve the removal efficiency of pollutants. Further, we explored sustained-release denitrification mechanism by investigating sustained-release carbon source characterization.

Denitrification performance and characterization analysis of sustained-release carbon source

The denitrification rate of the groups treated with additional sustained-release carbon source was significantly higher than that of CW_R control group (p < 0.05). The CW_R-Cu exhibited the highest denitrification rate (4.05 g/(m³·d)) (Fig. 4A) and the highest COD concentration in its effluent water (Fig. 3A), suggesting that evenly placed sustained-release carbon sources significantly increased the organic carbon compounds in CW and promoted the microbial denitrification process (Gu et al., 2021). As shown in Fig. 4B, the density of the sustained-release carbon sources in CW group decreased, and that in CW_R-Cu group was the lowest, indirectly suggesting that evenly placed sustained-release carbon sources released the most amount of organic matter during the same test period.

The sustained-release carbon sources in different groups before and after use and their SEM and EDS analyses were in Fig. 5. The original sustained-release carbon source (before use) was brownish yellow (Fig. 5A), and the sustained-release carbon source after use in CWs was black or light yellow (Figs. 5D, 5G, and 5J), which is related to the carbon release amount of the sustained-release carbon source. The surface of the original material was coated with PBAT to form a protective layer, exhibiting the stable carbon release effect (Fig. 5B). A great amount of fibrous substances appeared on the surface of the carbon source used in CW_R-Cu and CW_R-Ca, and some fibrous substances appeared in CW_R-Cb, which might be rod-shaped bacteria generated during water treatment (Figs. 5E, 5H, and 5K) (Shao et al., 2020). The C:O ratio on the surface of the original material was 59:41 (Fig. 5C). After use, the C:O ratio on the surface of the sustained-release carbon source decreased (Figs. 5F, 5I and 5L).

As shown in Fig. S1, the intensity of the vibration peak of the carbonyl group (-C=O) at 1,720 cm⁻¹ in the sustained-release carbon source after water treatment was slightly increased, compared with the original material (before use) (Fig. S1B). The carbonyl content ranked in the order of CW_Cu > CW_Ca ≈ CW_Cb > CW_CK. The increase in carbonyl content indicated that the sample was partially degraded. The intensity of the vibration peaks of carboxyl (-COOH) and hydroxyl (-OH) at 3,400 cm⁻¹ of the sustained-release carbon source after use increased slightly (Fig. S1C), suggesting that the hydrophilicity of the sustained-release carbon source was enhanced. The hydrophilic group content ranked in the order of CW_Cu > CW_Ca ≈ CW_Cb > CW_CK. After water treatment, the -COOH, -OH and -C=O groups on the surface of the sustained-release carbon source increased, resulting in an increase in the oxygen content on the surface of the sustained-release carbon source (Kara et al., 2021; Montazer, Habibi Najafi & Levin, 2020). The C:O ratio on the surface of the sustained-release carbon source in CW_R-Cu group exhibited the greatest decrease, indicating that the sustained-release carbon source in CW_R-Cu group was most fully utilized.

Effects of sustained-release carbon source placement method on nitrogen- and phosphorus-removing microorganisms

Figure 6A exhibited the relative abundance chord diagram of microorganisms at the phylum level. Proteobacteria (11.14–59.30%) and Bacteroidota (3.89–39.98%) were the dominant phyla in the microbial communities in all the CWs. Proteobacteria is crucial in driving carbon and nitrogen cycling (Al Ali et al., 2020; Yan et al., 2023) and maintaining organic matter degradation efficiency in each CW (Man et al., 2020). The relative abundance of Proteobacteria in CW_R was generally higher than that in other CWs, indicating that the organic matter released by the sustained-release carbon source inhibited the reproduction of Proteobacteria to a certain extent. Bacteroidota is a common chemically heterotrophic denitrifying bacteria that can decompose macromolecules to enhance biochemical properties (Si et al., 2018). The relative abundance of Bacteroidota in the groups treated with sustained-release carbon source was higher than that in CW_R, indicating that sustained-release carbon source could increase the relative abundance of denitrifying bacteria in CW. The highest abundance (39.98%) was observed in CW_R-Cu, implying that the uniform placement of sustained-release carbon source was more conducive to the reproduction of denitrifying bacteria than the centralized placement. Taken together, the addition and placement of sustained-release carbon source significantly affected the composition of microbial phyla.

Figure 6B showed the microorganism accumulation at the genus level. Denitrifying bacteria can convert NO₃⁻ into N₂ when the carbon source is sufficient (Fig. 6C) (Sun et al., 2022). The total relative abundance of the denitrifying bacteria genera Denitratisoma, Allorhizobium and Thauera was significantly higher in the sustained-release carbon source treatment groups than in CW_R (0.08%) (p < 0.05), indicating that with the addition of sustained-release carbon sources, the relative abundance of denitrifying bacteria in each treatment group was also increased (Fu et al., 2017; Yu et al., 2019). The total relative abundance of denitrifying bacteria in CW_R-Cu group was significantly higher than that in CW_R-Ca and CW_R-Cb groups, indicating that uniform placement of sustained-release carbon source could promote the reproduction of denitrifying bacteria and strengthen denitrification, compared with concentrated placement, which was consistent with the observation of the highest NO₃⁻-N removal efficiency in CW_R-Cu group (Fig. 3C). Pseudomonas belongs to denitrifying phosphorus-accumulating bacteria. Under anaerobic/anoxic conditions, this bacterial genus can use NO₃⁻ as an electron receptor to complete phosphorus uptake and denitrification through its metabolism, and it can also store phosphate present in the environment in the form of poly-P in cells (Zhang et al., 2020). In this study, the added sustained-release carbon source provided sufficient organic carbon sources for phosphorus-accumulating bacteria, allowing them to fully absorb phosphorus (Pietro & Ivanoff, 2015). The highest relative abundance of Pseudomonas (1.17%) was observed in CW_R-Cu, which was in line with the highest TP removal efficiency observed in CW_R-Cu group (Fig. 3E).

The α-diversity and richness of microorganisms in each treatment group were shown in Table 1. Coverage represents the reliability of sequencing. The larger the coverage value, the more reliable the sequencing results. The coverage of the eight samples was greater than 0.99, indicating that the sequencing results could fully reflect the actual situation of microorganisms in CW. Chao 1 represents the richness of species, the larger the Chao 1 value, the higher the richness; Shannon denotes the species evenness, the larger the Shannon value, the better the species evenness; and Simpson indicates the species diversity. The smaller the Simpson value, the higher the species diversity. The Chao 1 and Shannon indexes in CW_R are higher than those in the sustained-release carbon source treatment groups, but the Simpson index in some treatment groups was lower than that in CW_R, indicating that the addition of sustained-release carbon sources inhibited the richness and diversity of some species, but increased the diversity of other species, such as denitrifying bacteria. Chao 1 index value in uniformly placed sustained-release carbon source group was higher than that in concentratedly-placed sustained-release carbon source groups, indicating that uniform placement of sustained-release carbon sources was conducive to increasing microbial species richness.

Figure 7A showed the correlation heatmap between pollutant removal efficiency and the microbial relative abundance at phylum level. The heatmap showed that Bacteroidota had a significant positive correlation with the removal efficiency of TN and NO₃⁻-N. The higher the relative abundance of Bacteroidota, the higher the removal efficiency of TN and NO₃⁻-N, indicating the denitrification function of Bacteroidota (Si et al., 2018). Figure 7B shows the correlation heatmap between pollutant removal efficiency and the microbial relative abundance at genus level. The results showed that denitrifying bacteria such as Denitratisoma, Allorhizobium, and Thauera were significantly negatively correlated with NH₄⁺-N removal efficiency (p < 0.05), but positively correlated with the removal efficiency of TN and NH₄⁺-N, indicating that the higher the relative abundance of denitrifying bacteria, the higher the removal efficiency of TN and NO₃⁻-N, but excessively high NH₄⁺-N concentration had a certain inhibitory effect on denitrifying bacteria.

The heatmap of species composition at the phylum level was shown in Fig. 8A. There were large differences at microbial phylum level composition between CW_R and the sustained-release carbon source treatment groups, indicating that the addition of sustained-release carbon source changed the phylum-level microbial composition in the system. The phylum-level microorganisms in CW_R group were mainly positively correlated with Acidobacteriota, Chloroflexi, and Myxococcota, and those in the sustained-release carbon source treatment groups were mainly positively correlated with Desulfobacterota and Firmicutes. The heatmap of species composition at the genus level was shown in Fig. 8B. Consistent with the phylum-level observation, the addition of sustained-release carbon source changed the types of genus-level microorganisms in the system. The genus-level microorganisms in CW_R-Ca-A and CW_R-Cb had similar compositions and were positively correlated with denitrifying microorganisms (such as Denitratisoma). The two axes of PCA jointly explained 68.7% of the relationship between species information changes and the environment (Fig. 8C), among which CW_R-Ca-A and CW_R-Cb were close on the PCo1 axis (50.1%), which was consistent with the observations on Figs. 8A and 8B, suggesting that CW_R-Ca-A and CW_R-Cb were similar in species composition.