Microscale Constructed Wetlands with Different Particulate Matters in their Substrates Exhibit Opposite Nitrogen Removal Performances

Abstract

Excess suspended particulate matter (PM) in constructed wetland (CW) substrates may reduce the substrate porosity and thus affect pollutant removal performance. However, it remains unclear how different PMs affect the nitrogen removal performance in CWs. In this study, kaolin and polystyrene (PS) were selected as two model PMs added to CW substrates at a concentration of 100 mg/L. Four CWs were constructed, designated as C-CW without PM addition, K-CW with kaolin addition, M-CW with mixed addition of kaolin and PS, and PS-CW with PS addition. The CWs with or without PM addition showed no significant difference in terms of NH₄⁺-N removal efficiency (p > 0.05), while the removal efficiency of NO₃⁻-N and TN was significantly improved in PS-CW but, in contrast, was considerably inhibited in K-CW and M-CW (p < 0.05). The CWs with PM addition reduced the porosity of the substrates. There was no significant difference in the total solid quality among the CWs with PM addition (p < 0.05), but PS-CW had the highest volatile solid content. The addition of 100 mg/L PS significantly increased the activities of nitrite reductase (NIR) and nitrate reductase (NAR) with a much higher relative abundance of denitrifying bacteria, but it inhibited ammonia monooxygenase (AMO) and nitrite oxidoreductase (NXR) activities (p < 0.05). The activities of the four enzymes were improved to different degrees in K-CW and M-CW, in which the abundance of nitrifying bacteria was higher than that in C-CW. In conclusion, it was noteworthy that the effect of the PMs on the NO₃⁻-N and TN removal performance were qualitatively different (i.e., enhanced vs. inhibited) with different types of PMs. This interesting and important new finding could provide valuable information for a better understanding and evaluation of the role of PMs in the nitrogen removal process during CW operation.

1. Introduction

Due to their economic and social advantages, constructed wetlands (CWs), especially subsurface-flow CWs (SSFCWs), have been widely used to treat secondary effluents, domestic wastewater, etc. [1,2]. However, nearly half of all SSFCWs become clogged to varying degrees within 5 years [3]. Suspended particles are thought to be responsible for the blockage and service life reduction in SSFCWs. They are intercepted by the CW substrates from the influent [4], causing the accumulation of biofilms and the permeability reduction in the CW substrates. Particle accumulation in full-scale SSFCWs reaches as high as 10 kg of matter (dry weight, dw)/m² annually, with the proportion of inorganic accumulated solids being greater than 75% [5] and the particle size ranging from 0.1 to 100 μm [6]. This decreases the porosity of the substrate, and thus affects the treatment performance and service life of CWs [7,8].

Particulate matter (PM) in CW substrates can influence pollutant removal, especially nitrogen removal, by adsorbing pollutants directly and providing habitats for nitrogen-removal-related microbes [9]. The inorganic suspended particles commonly found are mainly composed of iron oxide, aluminosilicate, manganese oxide, alumina and quartz. As a kind of aluminum silicate salt, kaolin is often used in experiments to simulate inorganic PMs because it is stable over the whole pH range [10,11,12]. Liu et al. reported that inorganic PMs with a concentration of 100 mg/L in CW substrates affected the abundance of bacteria related to nitrogen transformation [13], with biofilms being formed by the cooperation of the PMs and extracellular polymeric substances (EPS) [14,15].

Recently, the accumulation of particulate microplastics in the environment has attracted extensive attention, especially in CW substrates, which have become the main gathering place of microplastics with concentrations as high as 1,285,212 pcs/kg·dw (dry weight). As a new type of pollutant, the effect of microplastics on the nitrogen removal performance in CWs has not been fully studied. A few related studies in the literature have suggested that microplastics can affect microbial communities and the nitrogen cycling process. Seeley et al. found that microplastics deposited in estuarine sediments significantly altered the microbial community structure and nitrogen cycling processes [16]. Rong et al. [17] reported that the presence of 7% (w/w) LDPE microplastic particles in soil altered the microbial community niches and nutrient competition in the soil.

Although microplastics are a type of organic matter, they are difficult to degrade in the natural environment, showing stable characteristics such as those of inorganic particles. In addition, they also have large specific surface areas, similar to inorganic particles, and can act as microbial carriers. However, whether microplastics have the same effect on nitrogen removal as inorganic particles has not yet been studied.

The main objective of this study was to explore the roles of typical inorganic particles and microplastics in nitrogen removal in SSFCWs. Two model PMs were added into SSFCW substrates with concentrations similar to those reported in the environment. The physicochemical parameters of the CW substrates and the removal performances of NH₄⁺-N, NO₃⁻-N and total nitrogen (TN) were determined. The changes in particle composition and size, the activities of key enzymes for nitrogen conversion and the microbial community structure caused by the accumulation of PMs were analyzed.

The acquired results could contribute towards a better and more complete understanding of how PMs in substrates affect the nitrogen removal performance in SSFCWs.

2. Materials and Methods

2.1. Particulate Matters

Kaolin and polystyrene (PS), were chosen to simulate inorganic and organic particles, respectively. To investigate the effects of solids accumulation on the treatment performance of CWs, kaolin particles (obtained from Jiangsu Province, China) and polystyrene (obtained from Guangdong Province, China) with particle sizes of 2–3 μm were mixed with CW substrates to a final concentration of 100 mg/L for simulating the rapid blocking process of CWs [13].

2.2. Experimental Setup and Operation

Four lab-scale vertical SSFCWs with the same dimensions of 15 cm × 60 cm (diameter × height) were set up in the laboratory of Shandong University, Qingdao, China. Gravels with Φ = 1.2 cm were arranged in the inlet area to facilitate the uniform distribution of water. Sand (Φ = 3 mm, porosity 36%) mixed with different PMs was filled into the CWs to a height of 50 cm. According to whether the PMs were added and what type of PMs were added, the four SSFCWs were designated as follows: (1) conventional CW (C-CW) without the addition of PMs; (2) kaolin-added CW (K-CW) with the addition of 100 mg/L kaolin; (3) mixed-PM-added CW(M-CW) with the addition of 50 mg/L kaolin and 50 mg/L PS and (4) PS-added CW (PS-CW) with the addition of 100 mg/L PS.

The pore volume was 3.8 L. A perforated PVC pipe (height 55 cm, diameter 3 cm) was inserted into the center of the substrate to monitor the physicochemical parameters in situ. Iris pseudacorus, with a 15 cm height and uniform growth, was selected for transplanting and then cultured with 10% Hoagland nutrient solution for two weeks until the plants grew well. Downwards flow was adopted during the operation of the CWs. Each CW was planted with 3 Iris pseudacorus. Each CW was operated in a continuous-flow mode with a hydraulic retention time (HRT) of 2 days. The hydraulic load rate was 1.3 mL/min. Simulated sewage was prepared according to level 1B effluent of wastewater treatment plants (WWTP), and the characteristics of the influent were as follows: 60.16 ± 1.31 mg/L chemical oxygen demand (COD, sucrose), 8.01 ± 0.02 mg/L NH₄⁺-N (NH₄Cl), 12.19 ± 0.21 mg/L NO₃⁻-N (KNO₃) and 1.01 ± 0.03 mg/L PO₄³⁻-P (K₂HPO₄). The influent with a DO of 5–6 mg/L was pumped into the CWs from a storage tank.

2.3. Water Quality Determination

The influent and effluent were sampled and monitored every 3 d for monitoring the water quality. COD, NH₄⁺-N, NO₃⁻-N and total nitrogen (TN) were determined according to standard methods [18]. Physicochemical parameters, namely dissolved oxygen (DO) and oxidation reduction potential (ORP), were simultaneously detected using a multiparameter digital instrument (Hach HQ40d, Loveland, CO, USA).

2.4. Calculation Method of N Balance in CWs

Nitrogen removal was determined by plant uptake, substrate adsorption, and microbial nitrification–denitrification, and the volatilization of NH₃ was negligible because of the relatively low concentration of NH₄⁺-N and the neutral pH of the influent. The accumulation rate of TN in the substrate was calculated using Equation (1).

$${\eta }_{SR }=\left(C_{SC1}-C_{SC0}\right)M_S{A}^{-1}{t}^{-1}$$

(1)

where η_SR is the accumulation rate of TN in the substrate, mg/(m²·d); C_SC0 and C_SC1 are the TN content of substrates at the beginning and end of experiment, mg/kg; M_S is the mass of the substrate, kg; A is the area of wetland, m² and t is the total running time of the experiment, d.

TN, i.e., the uptake rate of the plants, was calculated using Equation (2).

$${\eta }_{PR}=\left(C_{PC1}{M}_{1,1}-C_{PC0}{M}_{1,0}\right)A^{-1}{t}^{-1}$$

(2)

where η_PR is the absorption rate of TN, mg/(m²·d); C_PC0 and C_PC1 are the TN content of plantsat the beginning and end of the experiment, respectively, whose nitrogen contents were measured using a Kjeldahl Nitrogen Analyzer (Jinan, Shandong, China, Hanon Advanced Technology Group Co., Ltd./, K9840), mg/kg; $M_{1,0}$ and $M_{1,1}$ are the dry weight of the plants at the beginning and end of the experiment, respectively, kg; A is the area of wetland, m² and t is the total running time of the experiment, d.

The removal rate of microbial TN was calculated using Equation (3).

$${\eta }_{MR}{}_{}={\displaystyle \sum }_{i=1}^n\left(C_{i,0}V-C_{i,1}V\right)A^{-1}{t}^{-1}-{\eta }_{SR}-{\eta }_{PR}{}_{}$$

(3)

where${\eta }_{MR}$, ${\eta }_{SR}$ and ${\eta }_{PR}$ are the TN removal rates of microorganisms, substrates and plants, respectively, mg/(m²·d); C_i,0 and C_i,1 are the TN concentrations of each influent and effluent, respectively, mg/L; V is the amount of water in each inlet, L; A is the area of wetland, m² and t is the total running time of experiment; d.

2.5. Scanning Electron Microscope (SEM) Measurement

The substrate samples were washed 3 times with 1× phosphate buffer (pH = 7.2) for 10 min each. Dehydration was successively performed with 30%, 50%, 70%, 85%, 95% and 100% ethanol gradients for 15 min each time. A carbon dioxide critical point dryer (Quorom K850) was used to replace the ethanol with high-purity carbon dioxide and to perform critical point drying. The samples were platinized by a Hitachi MC 1000 ion sputtering instrument with a 10 mA current for 120 s.

2.6. Composition of Accumulated Solids

The size distributions of the PMs were analyzed by a laser particle size analyzer (BT-9300, Dandong, Liaoning, China). At the end of the experiment, 1 L of the substrate samples was taken from the four SSFCWs at different substrate depths (10–20 cm, 20–30 cm, 30–40 cm). The height was measured from the bottom of the devices. Clog matter was sufficiently separated from the gravel substrate by washing and filtration. The substrate samples were weighed before and after the washing process, and the weight loss was obtained as the wet weight of the clog matter. Then, total solids (TS) and volatile solids (VS) were determined by drying the separated clog matter at 105 °C to a constant weight and burning the residue at 550 °C for 40 min, respectively [19].

2.7. Determination of Permeability Coefficient

The change characteristics of the substrate permeability coefficient of the SSFCWs were measured based on Darcy’s Law [9]. When the constant head permeability coefficient method was implemented, a certain head difference was formed in the SSFCWs by controlling the inflow flow, and the head loss height was recorded after the water surface position stabilized [3]. The substrate permeability coefficient can be calculated according to Equation (4):

$$\mathrm{K}=\frac{QL}{A\Delta h}$$

(4)

where K is the permeability coefficient (cm/s); Q is the flow rate (mL/s); L is the filling height of the substrate (cm); A is the cross-sectional area of water (cm²) and Δh is the difference in the water head (cm).

2.8. Microbial Analysis

To determine the effect of PMs on nitrogen transformation at the protein level, the key enzyme activities of ammonia monooxygenase (AMO), nitrite oxidoreductase (NXR), nitrate reductase (NAR) and nitrite reductase (NIR) were measured. The activities of the enzymes were measured in this study according to the method proposed by Ma et al. [20]. the procedure is described in (Text S1).

At the end of the experiment, different types of substrates were collected for DNA extraction. The absolute abundance of the concerned genes (including bacterial 16S rRNA and N-related genes) was characterized by quantitative real-time PCR (qPCR) as described by Xie et al. [21].

Microbial DNA was extracted from the substrate samples using a DNeasy PowerSoil Pro Kit. The universal bacterial primers for 16SrRNA were 5′-ACTCCTACGGGAGGCAGCAGCAG-3′ and 5′-GGACTACHVGGGTWTCTAAT-3′ (V3-V4). The extracted DNA was amplified by PCR. The PCR amplification conditions were as follows: initial denaturation at 98 °C (3 min) followed by 25 cycles of denaturation at 98 °C for 30 s, annealing at 50 °C for 30 s, extension at 72 °C for 30 s and extension at 72 °C for the last 5 min. High-throughput sequencing was based on a sequencing platform from Bio-Pharm Technology Co., Ltd. (Shanghai, China). All sequence readings were clustered into operational taxonomic units (OTUs) (similarity threshold of 97%).

2.9. Data Analysis

The arithmetic mean and standard error of replicates were calculated. The normality of the data was assessed by the Kolmogorov–Smirnov test. A parametric one-way analysis of variance (ANOVA) followed by a post-hoc Tukey HSD test were used to determine the differences in the concentrations of pollutants among the four groups. The same test was employed to assess the effects of PM on TS, VS and microorganisms. All statistical analyses were performed using IBM SPSS Statistics 26.0 and R4.2.1 “http://www.R-project.org (accessed on 12 August 2022)”.

3. Results and Discussion

3.1. Nitrogen Removal Performance

As shown in (Figure 1), during the experimental period, the different CW units showed different nitrogen removal performances. The effluent concentrations of NO₃⁻-N, and TN fluctuated in the initial start-up period and stabilized on the 42nd day in all the CWs. There was no considerable difference in the NH₄⁺-N removal rate among all the CWs, and the effluent concentration of NH₄⁺-N reached below 0.025 mg/L, with the average removal efficiencies reaching 99.6% in all the CWs (Figure 1a). However, the addition of the different PMs to the CWs affected the NO₃⁻-N removal and thus the TN removal. Compared with C-CW, the removal efficiencies of NO₃⁻-N and TN in PS-CW were greatly improved, with the effluent concentrations reaching 3.2 mg/L and 3.25 mg/L, respectively (p < 0.05). On the contrary, the removal efficiencies of NO₃⁻-N and TN in K-CW were lower than those in C-CW. This meant that the addition of kaolin in CW negatively affected the NO₃⁻-N and TN removal performance. The removal efficiencies of NO₃⁻-N and TN in M-CW were higher than those in K-CW but lower than those in PS-CW, which might be related to the different characteristics of the PMs. The experimental results were inconsistent with the inhibitory effect of adding 100 nm of PS on the nitrogen removal in the CWs [20], owing to the fact that the 100 nm particles with smaller particle sizes had stronger microbial inhibitory effects [22]. The different sizes of the same particles led to different results. Therefore, the difference between the two types of PMs needs further discussion.

The different layers of wetland substrates also showed different nitrogen removal performances (Figure 2a–c). For the removal of NH₄⁺-N, K-CW and M-CW showed the same trend, while it was similar in C-CW and PS-CW. In K-CW and M-CW, the concentration of NH₄⁺-N decreased greatly in the 30–40 cm layer, accounting for 80% and 78% of the NH₄⁺-N removal, and the concentration continuously decreased to 0.015 mg/L and 0.026 mg/L in the 10–30 cm layer. However, the 20–30 cm layer contributed the most to the NH₄⁺-N removal in C-CW and PS-CW, with values of 2.45 and 2.25 mg/L, respectively, and the final effluent concentrations were 0.78 mg/L and 0.52 mg/L, respectively. Although the effluent concentrations of NH₄⁺-N in C-CW and PS-CW were slightly higher than those in the previous two groups, there was no significant difference in the effluent NH₄⁺-N concentration among all the groups (Figure 1a).

For the removal of NO₃⁻-N and TN, K-CW and M-CW had the same trend along the flow direction, as did C-CW and PS-CW. In C-CW and PS-CW, NO₃⁻-N and TN were removed stably along the height of the CWs, but the removal rate of PS-CW was higher than that of C-CW, with effluent concentrations of NO₃⁻-N of 3.36 mg/L and 4.18 mg/L in PS-CW and C-CW, respectively. However, in K-CW and M-CW, there was almost no NO₃⁻-N removal in each layer, and accumulation even occurred in the 30–40 cm layer, which might be attributed to the rapid conversion of NH₄⁺-N.

The contribution of different pathways to the nitrogen removal was further explored through a nitrogen mass balance analysis (Figure 2d). In all the nitrogen removal pathways, compared with C-CW, the addition of 100 mg/L PS promoted the contribution of microbial degradation, which was 2.7 times higher than that of K-CW. Although the contributions of plant uptake and media storage to nitrogen removal in PS-CW were higher than those of the other groups, the contribution of microbial degradationwas as high as 54%. Therefore, the contribution of microbial degradationto the removal of TN was improved with more PS addition, proving that PS addition promoted the transformation of nitrogen by microorganisms.

3.2. Particle Size Distribution and Composition of Accumulated Solids in CWs

PMs can be trappedby EPS, plant residues, root exudates, biological mucus and microbial cell bodies [23,24], which can change both the composition and size of the PMs and affect the microenvironment of the CW substrates. The initial particle size of the groups with PM addition was approximately 2.5 ± 0.25 μm (Figure 3a). During the experiment, the average particle size increased to different degrees. The main peak of the size distribution of the PMs in each group moved in the direction of larger particle sizes with the direction of water flow (Figure 3b–d). The average particle size of each layer of PS-CW was the largest among the three groups. In the layer at 10–20 cm, the largest particle size was 37.13 ± 0.22 μm (PS-CW), which was followed by 32.62 ± 0.41 μm (K-CW) and 26.36 ± 0.28 μm (M-CW). Although PMs were not added to C-CW, at the end of the experiment, PMs of a certain size could still be detected in the substrates and showed a trend of a gradual increase with the direction of water flow, which might have been from the substrate fragments and adsorbed biofilm. Based on the above results, different degrees of flocculation occurred in the accumulated solids in the four groups, which might have also been caused by the different adsorption capacities and migration capacities of the particles [9,25].

The SEM micrographs of the CW substrates are shown in (Figure 4). Compared with C-CW, the addition of PMs changed the surface morphology of the CW substrates, with the pores on the gravel surface absorbing much more flocculent substances, most of which were EPS produced by microbial metabolism [22]. Although there were clearly visible adhesive flocculent substances on the substrate surface of K-CW and M-CW, the content was much lower than that of PS-CW. The PMs on the surface of the PS-CW substrates were obviously wrapped in biofilm, and some even had membrane-like structures, which were mostly formed by the accumulation of EPS with 100 mg/L PS addition, which was consistent with the results of the larger particle sizes detected in (Figure 3).

Extracellular polymers are macromolecular substances that can cause the flocculation of inorganic particulate matter, which affects the flocculation of PMs and the microbial structure [9]. Therefore, the total solid (TS) and volatile solid (VS) contents were analyzed to reflect the contents of EPS adsorbed by the PMs (Figure 5). The addition of PMs increased the contents of TS and VS in the CW substrates. For TS, its quality was much higher than that of C-CW due to the addition of PMs in K, M and PS-CW, which could combine with more EPS and other macromolecular substances and resulted in an increase in the ratio of VS. Except for C-CW, the quality of TS in the PM-addition groups showed no difference in each layer (p > 0.05), and all of them increased along the direction of water flow, reaching above 7.0 g/L in the 10–20 cm layer (Figure 5a). For VS, the quality of VS in PS-CW was significantly higher than in K-CW and M-CW, reaching 3.47 mg/g TS in the 10–20 cm layer. The quality of VS in K-CW and M-CW also reached 2.23 and 2.78 mg/L in the 10–20 cm layer (Figure 5b). It has been reported that microplastics can induce the secretion of more EPS [22]. This was also consistent with the result that the 100 mg/L PS sample had more enclosed EPS (Figure 4d). Moreover, EPS has a strong ability to bind and adsorb pollutants [26], which is conducive to the removal of pollutants [27]. The quality of VS in PS-CW was the maximum in each layer, and thus the sample absorbed more EPS, which might have affected the substrate microenvironment and resulted in the different nitrogen removal performance.

3.3. Physiochemical Characteristics of the CW Substrates

The different PMs led to differences in the physicochemical properties and microenvironments of the different CWs (Figure 6). With increasing the operational time of the CWs, the permeability coefficient of all the CWs decreased, which also explained why different degrees of clogging occurred in the SSFCWs with different operation times (Figure 6a). The reduction rate of the permeability coefficient was similar in all the CWs, while the permeability coefficient was the maximum value in C-CW due to the absence of PMs. The difference in the permeability coefficient also led to a difference in DO and ORP, of which PS-CW had the lowest values of DO and ORP, which reached 0.85 mg/L and −280 mV in the 10 cm layer, respectively. The lower value of ORP, the faster the accumulation of EPS, and the more EPS produced by the microorganisms in the pores with a higher blocking potential.

The PMs could adsorb EPS, block part of the filler pores of the substrates, and thus affect the treatment performance of the CWs. Although the substrate of the CWs with PM addition had a rich anaerobic microenvironment with a similar DO with a similar carbon source (represented by COD) consumption (Figure S1), its NO₃⁻-N removal performance was different. Therefore, it was necessary to study the key enzyme activities and microbial structures to further explain the NO₃⁻-N removal of each group.

3.4. Key Enzyme Activities Related to Nitrogen Transformation

The nitrification and denitrification processes mainly rely on nitrifying and denitrifying bacteria to complete their nitrogen metabolic processes with protease catalysis [28,29]. The activities of the key enzymes involved in nitrification and denitrification were measured at the end of the experiment, including ammonia monooxygenase (AMO), nitrite oxidoreductase (NXR), nitrate reductase (NAR) and nitrite reductase (NIR).

As illustrated in Figure 7a–d, compared with C-CW, the addition of kaolin improved the enzyme activities of AMO, which resulted in the rapid transformation of NH₄⁺-N and the accumulation of NO₃⁻-N (Figure 2a,b). However, the addition of 100 mg/L of PS significantly reduced the enzyme activities of AMO and NXR and promoted the activities of NAR and NIR (p < 0.05). The high-quality VS of PS-CW provided part of the carbon source for microbial denitrification, which was consistent with the high NO₃⁻-N removal rate (Figure 1b) under similar carbon source consumption rates (Figure S1). Furthermore, the differences in the activities of the key enzymes could indicate different microbial structures, which will be discussed in the following section.

3.5. Microbial Community Structure in Different PM-Addition CWs

From the above section, it was found that the presence of PMs could affect the microenvironment and the activity of the reactive enzymes, but the community structure of the microorganisms might also affect the transformation of N [30,31,32,33]. The copy number of N-related genes (amoA, nirS and nosZ) in the CW substrates was determined (Figure 8a). The substrates of PS-CW had the highest absolute abundance of 16S rRNA among the four groups, indicating that the abundance of microorganisms was greatly improved with the higher quality of VS. Furthermore, the absolute abundance of amoA, nirS and nirK was consistent with the enzyme activities. The absolute abundances of nirS and nirK in PS-CW were about five orders of magnitude higher than that in the other groups (p < 0.001), which also explained the higher activity of the key denitrification enzymes and the NO₃⁻-N removal rate in PS-CW. Similarly, the absolute abundance of amoA was higher in K-CW and M-CW, which also explained the higher enzyme activity related to nitrifying.

At the genus level, the relative abundance of nitrifying and denitrifying bacteria in each substrate was analyzed (Figure 8b). Compared with C-CW, K-CW and M-CW had a significantly increased relative abundance of nitrifying bacteria, especially Nitrospira (p < 0.05), which was increased by 90.79% and 82.13%, respectively. This was consistent with the results of the high copy number of amoA genes and the nitrification enzyme activity and resulted in a great increase in the NH₄⁺-N removal rate of K-CW and M-CW. Although the relative abundance of nitrifying bacteria, amoA genes and the nitrification enzyme activity of PS-CW were lower than those of C-CW, the removal rate of NH₄⁺-N was not decreased, which may be related to the better reoxygenation effect and stronger nitrification ability of vertical-flow wetlands. Compared with C-CW, the abundance of denitrifying bacteria in PS-CW was significantly increased (p < 0.05), and the relative abundance of Pseudomonas was increased by 21.59%. This was consistent with the high abundance of denitrification genes and enzyme activities, which directly led to a large increase in the NO₃⁻-N removal rate of PS-CW. Similarly, the relative abundance of denitrifying bacteria in K-CW and M-CW was lower than that in C-CW, which directly led to a lower abundance of denitrifying enzyme genes and a lower denitrifying enzyme activity and resulted in the inhibition of the denitrification process.

4. Conclusions

The addition of kaolin and PS showed opposite effects on the removal of NO₃⁻-N and TN in the CWs. The addition of kaolin and/or PS to the CW substrates reduced their permeability coefficients and resulted in a more anaerobic microenvironment. Compared with the CW with kaolin addition, the one with 100 mg/L PS addition had its ORP value decreased more significantly with a larger amount of EPS accumulation. The addition of kaolin in the CWs inhibited the NO₃⁻-N removal due to there being less adsorbed EPS, while it enhanced the NH₄⁺-N removal with a higher abundance of nitrifying bacteria and related enzyme activity. On the contrary, the addition of 100 mg/L of PS in the CWs enhanced the NO₃⁻-N removal. This was attributed to the greatly enhanced absorption of VS, providing significantly more carbon resources for microbial denitrification with a higher denitrifying enzyme activity and microbial abundance.