Effects of selection and compiling strategy of substrates in column-type vertical-flow constructed wetlands on the treatment of synthetic landfill leachate containing bisphenol A

Landfilling is the most popular method for treating solid wastes globally, considering the environmental, financial, social, and urban development factors. Leachate generated from landfills comprises highly concentrated organic and inorganic contaminants. Although the chemical composition of the landfill leachate varies depending on the solid waste composition, the duration after landfilling, and the climatic conditions at the landfill site (Renou et al. 2008), high concentrations of recalcitrant phenolic compounds such as bisphenol A (BPA) and alkylphenols are frequently detected (Boonnorat et al. 2014). Poor construction and inexperienced operation and maintenance of landfills leads to the discharge of leachate into the surrounding soil, groundwater, and air, which is toxic to the environment (Haslina et al. 2021). Owing to the potential danger to humans and ecosystems, landfill leachates require appropriate treatment before being discharged into the environment to minimize potential risks.

Constructed wetlands (CWs) are low-cost, environmentally friendly, and widely applicable technologies with minimum operation and management requirements. Thus, they are regarded as a promising solution for the treatment of landfill leachate, especially in developing countries (Vymazal 2007). CWs comprise substrates, plants, and microorganisms, and various physical, chemical, and biological mechanisms, including adsorption, biodegradation, plant uptake, and volatilization, contribute to the pollutant removal (Vymazal 2007). Previous studies have demonstrated the efficient removal of not only the biochemical oxygen demand, ammonium-nitrogen (NH₄-N), and phosphorus (Renou et al. 2008) but also specific pollutants such as heavy metals and recalcitrant organic compounds from landfill leachate (A et al. 2017a, 2017b; Yi et al. 2017).

The efficiency of pollutant removal in CWs varies depending on the system design. In particular, the selection and combination of the CW type and enclosing substrate are vital determinants of the treatment performance of CWs. Many types of CWs have been developed, including the free-water surface CW, horizontal flow CW, and vertical flow CW (Wu et al. 2015). In addition to the single use of certain CW types implemented traditionally, hybrid CWs that combine different CW types based on hydrologic processes have been applied successfully (Wu et al. 2015). Some studies have demonstrated that two-stage hybrid CWs can improve the overall wastewater treatment efficiency while simultaneously removing different types of pollutants compared to single-stage CWs (Li et al. 2014; Silvestrini et al. 2019).

The selection of substrates also affects the overall pollutant removal ability of CWs as they account for most of the CWs. They have multiple functions, including direct effects as the adsorbent and indirect effects as the supporting media for plants and biofilms to facilitate the biological removal of pollutants (Wu et al. 2015). While certain substrates are laid homogeneously in most CWs (Wu et al. 2015), the combined use of different substrates, such as sandwiching an accessory substrate with a primary one, has proven to be useful for expanding their pollutant removal capacity (Wirasnita et al. 2018).

Despite extensive studies on the CW system design, including the CW type, substrate selection, and compiling strategy, only few studies have directly compared the effects of distinct CW design on the leachate treatment performance (Silvestrini et al. 2019; Saeed et al. 2020). To the best of our knowledge, the effects of the removal of specific recalcitrant organic pollutants have rarely been explored. Therefore, this study aimed to compare the effects of CW type and substrate selection, combination, and compiling strategies on the treatment capacity of synthetic leachate containing BPA as the representative recalcitrant phenolic pollutant. We constructed five different laboratory-scale column-type CWs, including single planting CWs consisting of gravel with and without sandwiching with other substrates (activated carbon or brick crumbs) and two-stage hybrid CWs combining planting gravel CW and unplanting CW by stuffing other substrates in series. Moreover, their leachate treatment performance was evaluated in terms of general organic and nitrogenous compounds along with BPA.

The synthetic landfill leachate comprised sodium acetate (120 mg), sodium propionate (60 mg), and sodium humate (60 mg) as the organic components; KH₂PO₄ (14 mg), NH₄Cl (107 mg), CaCl₂·2H₂O (132 mg), and MgSO₄·7H₂O (40 mg) as nutrients (Soda et al. 2016); and BPA (5 mg) as a recalcitrant pollutant. This was prepared in 1 L of tap water. The concentration of BPA in the synthetic leachate applied here was relevant to BPA concentrations found in the leachates from solid waste disposal sites (Urase & Miyashita 2003). BPA of the highest available grade was purchased from Tokyo Chemical Industry (Tokyo, Japan). The concentrations of chemical oxygen demand (COD) and NH₄-N in the synthetic leachate were 200 and 26.7 mg/L, respectively.

Ten lab-scale CWs (duplicate for each of the five types; Figure 1) were set up in a greenhouse at Osaka University, Osaka Prefecture, Japan (34° 49′ 17″ N, 135° 31′ 17″ E). Three substrates with different properties were used: gravel (10 mm diameter), a widely used substrate in CWs (A et al. 2017a), activated carbon (2–4 mm diameter) with a high adsorption property (Korotta-Gamage & Sathasivan 2017), and brick crumbs (2–4 mm diameter) with low cost and good hydraulic performance and adsorption properties (Ren et al. 2007). The five types of CWs were labeled as CW1–CW5. CW1–CW3 consisted of a single plastic column (18 cm diameter × 70 cm height), wherein CW1 was filled with 45 cm of gravel, whereas CW2 and CW3 were filled with 25 cm of gravel at the bottom and 10 cm of activated carbon (1 kg) and brick crumbs (3.3 kg) above the gravel layer, respectively, and subsequently with gravel up to 45 cm from the bottom. CW4 and CW5 were two-stage hybrid CWs with two plastic columns (15 cm diameter × 60 cm height and 12.5 cm diameter × 50 cm height). The larger column was filled with 40 cm of gravel, whereas the smaller one was filled with 1 kg of activated carbon and 3.3 kg of brick crumbs for CW4 and CW5, respectively. The common reeds (Phragmites australis) purchased from Morikawa gardening (Osaka, Japan) were pre-cultivated in a greenhouse for approximately one month to a height ranging from 50–78 cm above the ground, and planted in CW1–CW3 and in the larger column in CW4 and CW5. They were not planted in smaller columns in CW4 and CW5. All CWs were equipped with pumps for water circulation. Sampling ports were installed at the bottom of the column for CW1–CW3, whereas at the bottom of the smaller column in CW4 and CW5 (Figure 1).

All lateral CWs were covered with aluminum foil from the outside to prevent the substrates' direct exposure to sunlight. In addition to sunlight, over 1,300 lux of light was provided with fluorescent lamps for 12 h per day (6:00–18:00). The air temperature in the greenhouse was maintained at 27 °C or higher to simulate moderate conditions for plant growth and bioactivity in CWs.

All CWs simultaneously started operating on 13 May, 2020. The CWs were filled with 5 L of synthetic leachate without BPA and operated with water circulation at a vertical flow rate of 3.5 mL/min for 7 days. Subsequently, the water in the CWs was discharged, and fresh synthetic leachate was poured into the CWs for the next batch operation. After an acclimation period of 4 weeks, the influent was replaced with a synthetic leachate containing BPA, and a 7-d batch treatment was conducted five times. The hydraulic loading rate (HLR) under the operational conditions of this study was approximately 2.8 cm/d. This HLR was lower than 4.9–9.8 cm/d in the previous CW studies treating BPA and other recalcitrant organic pollutants (A et al. 2017a; Wirasnita et al. 2018). In the previous study, BPA removal was not necessarily high even after a long acclimation period (3 months), which would allow sufficient biofilm formation (A et al. 2017a). In contrast, the acclimation period was relatively short in this study, and biofilm formation was likely insufficient. Thus, this study applied a low HLR with a long hydraulic retention time.

Fifteen milliliters of the influent (synthetic leachate) and effluent (treated water) samples were collected from each CW. Effluent sample collection was conducted every day during the first run of the 7-d cycle period for the treatment of BPA-containing synthetic leachate, whereas samples were collected on the last day during the second and third runs, and on the second, fifth, and seventh days during the fourth and fifth runs. The collected samples were analyzed for the pH, temperature, electrical conductivity (EC), dissolved oxygen (DO), COD, nitrate-nitrogen (NO₃-N), NH₄-N, and BPA.

The pH, temperature, DO, and EC were measured using a portable multimeter (HQ40d; Hach, CO, USA). For chemical analysis, the samples were filtered using a 0.45 μm pore size filter paper (Advantec, Tokyo, Japan). The concentrations of NO₃-N and NH₄-N were determined by analyzing each ion using ion chromatography (HIC-20A system; Shimadzu, Kyoto, Japan), as described previously (A et al. 2017a). The COD was measured using a COD_Cr cell test and a mobile colorimeter (Spectroquant^® Move 100; Merck Millipore, Darmstadt, Germany). The concentration of BPA was measured using high-performance liquid chromatography (LC-10A_vp system, Shimadzu), as described previously (Ghaju Shrestha et al. 2021).

After the leachate treatment experiments, the CWs were maintained by feeding synthetic leachate without BPA for 7 days. Subsequently, all the CWs were destroyed for sampling the substrates and plant roots for microbial analysis. The gravel samples were collected from three different layers of CW1 and from the top (40–45 cm) and bottom (5–10 cm) layers of CW2 and CW3. Activated carbon and brick crumbs were also collected from the middle layer (25–30 cm) in CW2 and CW3. The gravel samples of CW4 and CW5 were collected from the top (35–40 cm), middle (20–25 cm), and bottom (5–10 cm) layers of the large column, whereas the activated carbon and brick crumbs were collected from the top (20–30 cm), middle (10–15 cm), and bottom (5–10 cm) layers of the small column. Primary roots and root hairs of the plants were collected from each CW. Five grams (wet weight) of substrates and 2 g of roots were suspended in 45 and 18 mL of sterilized 0.9% saline solution, respectively. The mixture was treated in an ultrasonicator (1510 J-DTH; Yamato Scientific, Tokyo, Japan) for approximately 1 min to disperse the biofilm attached to the substrates and roots.

One milliliter of the prepared sample solution was serially diluted and spread onto the R2A medium (Merck Millipore) to count the cultivable heterotrophic bacteria. Additionally, the diluted sample was spread onto a basal salt agar medium (Toyama et al. 2009) containing BPA (5 mg/L) as the sole carbon source to detect the cultivable BPA-degrading bacteria. The plates were incubated at 28 °C for 24 h and one week for the cultivation of heterotrophic and BPA-degrading bacteria, respectively.

Bacterial DNA was extracted from the prepared sample solution using the CicaGeneus DNA extraction kit (Kanto Chemical, Tokyo, Japan). The extracted DNA was subjected to 16S rRNA amplicon sequencing. Amplification of the V3/V4 region of the 16S rRNA gene was conducted using a two-step PCR procedure. The first PCR was conducted using the universal primers 515F and 806R (Takai & Horikoshi 2000; Baker et al. 2003). The amplified products were purified and the second PCR conducted the barcoding, after which the DNA library was subjected to sequencing on an Illumina MiSeq platform (2 × 300 bp paired-end sequencing; Illumina, CA, USA) at the Bioengineering Lab Co. Ltd (Kanagawa, Japan). The obtained sequences were aligned to the operational taxonomic units (OTUs) obtained based on a threshold of 97% similarity and classified into phylogenetic groups.

All statistical analyses were conducted using Microsoft Excel 2016 software. The Mann-Whitney U test was performed to test the differences in related parameters between CWs with different configurations.

The general parameters of the effluent in all the CWs during the five-week experiment (five 1-week batch cycles) for treating the BPA-containing synthetic leachate are shown in Figure 2. The temporal trends of pH and temperature were similar for all five CWs. The pH was between 7.9 and 8.8 and showed a slightly decreasing trend during the five-week operation. The effluent temperature was in the range of 26.3–32.1 °C. The DO concentration was 3.0–5.1 mg/L, indicating aerobic conditions in all the CWs. The EC was 51–79 μS/m during the experimental period. DO concentration seemed higher and the EC seemed lower in two-stage CWs (CW4 and CW5) compared to single-column CWs (CW1, CW2, and CW3), although the differences were not statistically significant.

The average number of shoots of common reed in each CW was 17–35 at the beginning of the experiment, which increased to 26–50 after 5 weeks. The height of the plants also increased from 58–73 cm to 80–115 cm during the experiments. The plants in all CWs were healthy throughout the experiment, and all the CWs functioned well without clogging.

NH₄-N in the influent was removed with 70% or higher efficiency, irrespective of the CW and time period (Figure 3). Nearly complete NH₄-N removal (≥99%) was achieved over five weeks in CW2 and CW4. Additionally, in CW5, although the NH₄-N removal efficiency was 94% in the first week, nearly complete removal (>97%) was observed after the second week. In CW3, NH₄-N removal improved from 76% in the first week to nearly 99% in the fifth week. Conversely, in CW1, NH₄-N was completely removed during the initial two weeks, after which the removal efficiency declined to approximately 85% in the third and fourth weeks, and subsequently improved to 94% in the fifth week. NO₃-N was not detected in effluent samples from any CW.

COD removal was also evaluated in the second week (Figure 3). The removal efficiency of COD was 90% or higher in all CWs, except in CW1 and CW5 in the second week. The removal efficiency of COD improved in the third and fourth weeks, with efficiencies of 98–100%. The removal efficiency declined slightly in the final week, except for CW2. The overall COD removal performance was slightly higher in CW2 and CW4 compared to the other three CWs.

Figure 4 shows the time courses for the removal of BPA by the five different CWs. BPA was completely removed in all five batch cycles in CW2 and CW4. In CW5, complete BPA removal was observed, except for a temporal decline in the fourth week (89%). In both CW1 and CW3, BPA removal was <60% and <70% in the first and second weeks, respectively, and increased to >90% after the third week. Among the three single-column CWs, the 5-week removal efficiencies were significantly higher in CW2 than in CW1 (p < 0.05), whereas no significant difference was observed between CW1 and CW3. Temporal variations within a batch cycle during the first, fourth, and fifth weeks also revealed that BPA removal occurred rapidly after feeding the influent in CW2 and CW4, whereas this was relatively slow in the other three CWs.

The abundances of cultivable heterotrophic and BPA-degrading bacteria were respectively present at 6.8–7.6 and 5.8–6.4 log CFU/g on the substrates and at 8.2–8.3 and 7.5–7.6 log CFU/g on the plant roots (Figure 5). BPA-degrading bacteria accounted for 6–20% of heterotrophic bacteria. Although the abundances of heterotrophic and BPA-degrading bacteria did not vary noticeably among the five CWs, their abundances were slightly higher on activated carbon (7.3–7.5 and 6.1–6.3 log CFU/g, respectively) and brick crumbs (7.3–7.6 and 6.1–6.4 log CFU/g, respectively) compared to gravel (6.8–7.2 and 5.8–6.2 log CFU/g, respectively) in CW2–CW5.

The bacterial communities on the substrates and plant roots after the experiment are shown in Figure 6. From the 26 samples analyzed, 32,139–71,794 reads were obtained. The phylum Proteobacteria was predominant in all CWs (33.7–94.8%). Bacteroidetes was also abundant (1.1–35.6%). Its abundance appeared high on plant roots (9.6–35.6%) as compared to substrates (1.1–28.0%). Additionally, Firmicutes was dominant in plant roots in CW1 (33.0%) and at the top (60.5%) and middle (29.0%) of the gravel filter (the larger column) in CW5.

Among Proteobacteria, Gammaproteobacteria (13.5–57.6%) and Betaproteobacteria (5.4–64.7%) were the two dominant classes (Figure 6). Pseudomonas was the dominant genus in Gammaproteobacteria with an abundance of 0.79–41.4%. Betaproteobacteria appeared to be dominated on activated carbon in CW2 and CW4. Nitrosomonadaceae, which is classified as Betaproteobacteria and is responsible for ammonia oxidation, was present only at 0–1.63%. Its abundance was especially low (0.10% at the highest) in the planted columns (CW1–CW3 and larger columns in CW4 and CW5), except in the top part of the gravel in CW2 (0.81%), compared to the unplanted columns in CW4 and CW5 (0.46–1.63%). Additionally, the detected genera included known BPA-degrading bacteria, such as Cupriavidus (Fischer et al. 2010), Novosphingobium (Toyama et al. 2009), Sphingobium (Zhou et al. 2013), and Sphingomonas (Sasaki et al. 2005) in addition to Pseudomonas (Telke et al. 2009) and Nitrosomonas (Roh et al. 2009), at >0.1% of the total bacterial communities in at least one of the used substrates.

In this study, five different CWs were applied to understand the effects of the CW type, substrate selection, and compiling strategy on the treatment capacity of synthetic leachate containing BPA. Based on the results of plant growth and general water quality parameters such as the pH, temperature, EC, and DO (Figure 2), moderate conditions to support plant growth and bioactivity were maintained, irrespective of the CW configuration. Higher DO concentrations and lower EC values in the two-stage CWs seemed to provide more aerobic conditions and ion exchange capability than single-column CWs.

Considering the pollutant removal from the synthetic leachate, 70% or higher NH₄-N was removed in all CWs throughout the five repeated batch treatments (Figure 3). Substrate adsorption, plant uptake, and microbial nitrification are potential mechanisms for NH₄-N removal in CWs. As NO₃-N was never detected in the effluent and the abundance of ammonia-oxidizing bacteria was very low, the contribution of nitrification to NH₄-N removal was minor. Therefore, the substrate adsorption and plant uptake are the primary mechanisms for NH₄-N removal in CWs, as reported previously (Vymazal 2007). Additionally, the good growth of common reed in all CWs suggested that differences in NH₄-N removal between the five CWs may be caused by distinct substrate inclusion. The complete removal of NH₄-N in CW2 and CW4 could be attributed to the high adsorption capacity of activated carbon used in these CWs. Moreover, higher NH₄-N removal efficiencies in CW5 compared to CW3 during the initial 3 weeks despite the use of the same substrates (gravel and brick crumbs) suggested that two-stage CWs were more effective in NH₄-N removal than a single CW. This also indicates the importance of the CW configuration for improved NH₄-N removal. Although the specific reason for the improved NH₄-N removal in the two-stage CW (i.e., CW5) compared to the single CW (i.e., CW3) was not revealed, the abundance of ammonia-oxidizing bacteria in the second unplanted column of CW5 might have affected it.

Considering the removal of COD, the single-column CW using activated carbon (CW2) and two-stage CWs (CW4 and CW5) had higher efficiencies than single-column CWs using gravel (CW1) and brick crumbs (CW3) (Figure 3). The importance of the substrate can be observed in CWs using activated carbon as they effectively removed various organic compounds (Korotta-Gamage & Sathasivan 2017). The removal efficiencies of COD in synthetic leachate using CWs obtained in this study were higher than those using the ozone combined with hydrogen peroxide and microwave-activated persulfate processes (Chen et al. 2019). This corroborates the usefulness of CWs in treatment of organic pollutants with minimum operation and management.

The mass production and widespread use of BPA and its detection in landfill leachate and wastewater have quested for its effective treatment (Sánchez-Avila et al. 2009; Yi et al. 2017). Therefore, BPA has been applied as the representative recalcitrant phenolic pollutant in previous CW studies (A et al. 2017a; Wirasnita et al. 2018) and was also selected in this study. Consequently, our study revealed that BPA removal abilities in all CWs of this study after the third week was similar to those obtained in previous CW studies (A et al. 2017a; Wirasnita et al. 2018) and superior to those achieved using a membrane bioreactor combined with external ultrafiltration (74%) (Fudala-Kszaizek et al. 2018), suggesting the usefulness of CWs even in the removal of recalcitrant organic pollutants, like BPA.

Daily monitoring during the first week showed a large difference in the initial BPA removal among the applied CWs (Figure 4). Complete removal of BPA on the first day in CW2 and CW4 indicated a strong impact of activated carbon as the adsorbent. Adsorption by activated carbon has been used for the removal of BPA from wastewater (García-Araya et al. 2003). Conversely, although the inclusion of brick crumbs in a single-column CW (i.e., CW3) did not improve the BPA removal ability compared to a single-column CW using only gravel as the substrate, its use in two-stage CWs (i.e., CW5) improved the BPA removal ability. Therefore, it was suggested that the adsorption performance of the substrates varies depending not only on the adsorption property of the substrate but also on its application strategy in the CW (Ren et al. 2007). This highlights the importance of the substrates' inclusion strategy and the selection of substrate materials for the effective removal of recalcitrant pollutants in CWs, such as BPA.

In contrast, the physical adsorption of pollutants onto the substrate occurs for a relatively short period, and its adsorption capacity does not vary or declines through repeated loadings. The daily increase in the BPA removal efficiency during a batch cycle and improvement in the BPA removal efficiency over the five repeated batch cycles (Figure 4) were mostly caused by biodegradation. This was corroborated by the detection of cultivable BPA-degrading bacteria (Figure 5) as well as diverse bacterial genera, including known BPA-degrading bacteria (Figure 6) on the substrate and the roots of P. australis. BPA biodegradation occurs preferentially under aerobic conditions (Ike et al. 2006; Zhang et al. 2013; Im & Löffler 2016). Thus, higher BPA removal in two-stage CWs than in single-column CWs (Figure 5) is partly attributed to the higher activity of BPA biodegradation in two-stage CWs, which could create more aerobic conditions (Figure 2). Furthermore, the degradation of BPA by BPA-degrading bacteria facilitates the maintenance of the BPA adsorption ability of the substrates.

This study investigated the performance of CWs with different CW types, substrates and substrate compiling strategies on the treatment of synthetic landfill leachate containing BPA. Based on the results, it was clarified that the selection of the substrate largely affects the pollutant removal performance. Among the substrates applied in this study, activated carbon was superior in effectively and quickly removing the pollutants (NH₄-N: 99–100%, COD: 93–100%, BPA: 100%). Additionally, the compiling strategy of the substrate in CWs showed a distinct difference in the removal efficiencies of pollutants. Between the single-column (NH₄-N: 76–99%, COD: 90–100%, BPA: 63–94%) and two-stage CWs (NH₄-N: 94–99%, COD: 89–98%, BPA: 89–100%) using brick crumbs, the latter showed better pollutant removal ability. This indicated the possibility of enhancing the pollutant removal ability by improving the substrate compiling strategy regardless of the applied substrate. Further studies are required to understand the differences in the pollutant removal mechanism and efficiency with different compiling strategies using the same substrate. In addition, this was a model study treating synthetic landfill leachate containing limited and known pollutants. Thus, further studies using real landfill leachate which may contain diverse pollutants affecting the activities of plants and microorganisms in CWs are needed to clarify the practical usefulness of CWs on the treatment of recalcitrant organic pollutants such as BPA.

This study was supported by the Environment Research and Technology Development Fund (JPMEERF18S11715) of the Environmental Restoration and Conservation Agency of Japan. The authors would like to thank Mr Kazuki Nishihata (Osaka University) for his help with the CW experiments. The authors would also like to thank Editage (www.editage.com) for English language editing.

The authors declare no conflicts of interest.

All relevant data are included in the paper or its Supplementary Information.