NG25

Effect of multilayer substrate configuration in horizontal subsurface flow constructed wetlands: assessment of treatment performance, biofilm development, and solids accumulation

Abstract This study investigates the influence of multilayer substrate configuration in horizontal subsurface flow con- structed wetlands (HSCWs) on their treatment performance, biofilm development, and solids accumulation. Three pilot- scale HSCWs were built to treat campus sewage and have been operational for 3 years. The HSCWs included monolayer (CW1), three-layer (CW3), and six-layer (CW6) substrate configurations with hydraulic conductivity of the substrate increasing from the surface to bottom in the multilayer CWs. It was demonstrated the pollutant removal performance after a 3-year operation improved in the multilayer HSCWs (49– 80%) compared to the monolayer HSCW (29–41%). Simultaneously, the multilayer HSCWs exhibited significant features that prevented clogging compared to the monolayer configuration. The amount of accumulated solids was notably higher in the monolayer CW compared to multilayer CWs. Further, multilayer HSCWs could delay clogging by provid- ing higher biofilm development for organics removal and con- sequently, lesser solids accumulations. Principal component analysis strongly supported the visualization of the perfor- mance patterns in the present study and showed that multilay- er substrate configuration, season, and sampling locations sig- nificantly influenced biofilm growth and solids accumulation. Finally, the present study provided important information to support the improved multilayer configured HSCW implica- tion in the future.

Introduction
Horizontal subsurface flow constructed wetlands (HSCWs) have been widely used for wastewater treatment as an envi- ronmental friendly technology (Kadlec and Wallace 2008; Vymazal 2009). However, clogging is the most common prob- lem encountered in HSCWs, which can result in a large pro- portion of the influent water passing over the bed as surface runoff. Clogging reduces the possibility of contact between pollutants and biofilm required for biodegradation in the HSCWs, which significantly weakens treatment performance and shortens their life span (Osorio et al. 2007). Thus, solu- tions that prevent clogging and optimize CW utilization have gained considerable attention in the recent decades. Substrate configuration is an important factor that influences clogging as most clogging solids wrap around the substrate pores (Brovelli et al. 2011). Theoretically, substrate configura- tion can be improved by placing finer media with low hydraulic conductivity on the top and coarser media with high hydraulic conductivity at the bottom, which should prevent short circuit of water flow on the surface and decrease dead zones at the bottom. It is believed that this substrate configuration enables a natural and even water flow, which prevents insalubrious clog- ging of the CWs (Knowles et al. 2011; Morales et al. 2013). Furthermore, the uniform water flow pattern would improve pollutant biodegradation due to high biofilm/pollutant contact possibility. The system following this concept was built and recorded higher pollutant removal efficiencies the year follow- ing construction (Bai et al. 2016). However, the performance after a relatively long-term (3-year) operation has not been evaluated after over 3 years of operation.

Clogging is a complex process; however, the exact mech- anisms and consequences of clogging are not completely clear. Blazejewski and Murat-Blazejewska (1997) reported that biofilm growth and solids accumulation are important factors that cause clogging. An investigation of the develop- ment process of biofilm during changing seasons showed that temperature was an important factor in clogging (Platzer and Mauch 1997). Higher temperatures result in higher biological activity, which decreases organic matter accumulation. However, it may also cause higher biofilm growth that pro- motes clogging. In horizontal flow CWs, solids accumulation has been recognized as a major factor that decreases CW lon- gevity (Cooper et al. 2005). Accumulated solids in CWs in- clude both organic and inorganic solids (Knowles et al. 2011). However, comprehensive investigation of biofilm growth and solids accumulation related to spatial and seasonal differences in HSCWs with multilayer substrate configuration has not been conducted.The aim of this study was to determine the influence of multilayer substrate configuration on treatment performance, biofilm development, and solids accumulation after the 3-year operation in HSCWs. Pollutant removal efficiencies were measured in a pilot HSCW treating campus sewage that hadbeen operational for 3 years. Biofilm development and solids accumulation were analyzed in different seasons (April, July, and September) of 2016 to compare the spatial difference between monolayer and multilayer HSCWs. Total viable cell counts (TVC), extracellular polysaccharide (EPS), and extra- cellular protein (EP) were measured to represent biofilm development. Different solid components, including organic and inorganic matters, were investigated to evaluate solids accumulation. Three parallel pilot-scale HSCWs were constructed in December 2012 at Guilin University of Technology, in Southwest China (N 25° 28′,E 110° 31′). A detailed description of the experimental setup can be found in Bai et al. (2016).

Briefly, each system had distinct dimensions of 2-m length, 1.2-m width, 0.7-m height, and water depth of 0.6 m (Fig. 1a). Each HSCW was divided into influent distribution zone (0.2 m long), main reaction bed (1.6 m long), and effluent collection zone (0.2 m long). The main reaction beds of the three HSCWs had different configurations (Fig. 1b). CW1 was monolayer substrate configured, filled with quartz sand (0–6 mm) with hydraulic conductivity (K) of 65 m/d. CW3 consisted of three equal layers with 0.2-m thickness. The K values of the layers were 26, 36, and 64 m/d from the surface to the bottom, and the corresponding sand diameters were 0– 0.4, 0.4–0.6, and 1–3 mm, respectively. CW6 comprised of six equal layers with 0.1-m thickness. From the top to bottom, sand sizes were 0–0.4, 0.4–0.6, 0.6–0.9, 1–2, 2–4, and 4–6 mm, withK values of 26, 36, 43, 55, 75, and 176 m/d, respectively.The laboratory facility was subject to natural temperature variations and light exposure but protected from rain by aglass roof. The influent of this experiment was original cam- pus sewage pre-treated by a septic tank. The three HSCWs shared a 200-L influent tank with separate pump stations func- tioning at identical hydraulic loading rate of 0.3 m3/m2/d, to ensure consistent operation mode. The systems had been con- tinually operational for the past 3 years. The present study was conducted from April 1, 2016, to September 1, 2016.

All HSCWs were planted with Canna indica at the density of 20 individuals per square meter. The average influents chemical oxygen demand (COD), NH +–N, total nitrogen (TN), and total phosphorous (TP) were approximately 145, 23, 32, and 5 mg/L during experimentation period (Table 1).Triplicate water samples were collected from the influent tank and effluent pipes of the HSCWs once a week to monitor water quality and pollutant removal. The pH and oxidation- reduction potential (ORP) were measured in situ using a Hanna Hi9828 multi-parameter portable meter. Air and influ- ent water temperatures were recorded everyday using an au- tomatic temperature logger connected to the sensor. Individual water samples were stored in 100-mL sterile plastic bottles, cooled to 5 °C, and transported to the laboratory. The concen-trations of COD, ammonium (NH +–N), nitrate (NO −–N),TVC, EPS, and EP were tested to evaluate biofilm develop- ment (Ragusa et al. 2004). TVC was measured using the method described by Findlay et al. (1989). Briefly, phospho- lipids were extracted from substrate samples and concentra- tions of phosphate released by phospholipids were determined using the method described by Shatton et al. (1983). TVC was calculated from the phosphate concentration using the conver- sion factor; 1 nmol phosphate is proportional to 3.4 × 107 cells as determined by Findlay et al. (1989). Additionally, bovine serum albumin was used for construction of standard curves. The quantification of biofilm EPS was assayed in phenol sul- furic acid, as detailed by Frolund et al. (1996).

The substrate samples were placed in glass reaction tubes. Subsequently, 1 mL each of distilled water and 5% phenol were added to individual tubes. The tubes were mixed in a vortex mixer for 30 s. The volume of 5 mL concentrated sulfuric acid was added to each tube during the mixing process. Absorbance was measured at 485 nm after cooling. Glucose was used to construct standard curves for polysaccharide concentration. EP was extracted from the substrate samples by adding 1 N NaOH and incubating at 55 °C for 2 h. Equal volumes of 1 NHCl were added to each sample to neutralize the NaOH after cooling to room temperature (Ragusa et al. 2004). The solu-Substrate samples of the attached biofilm and solids accumu- lation analyses were collected on April 1, July 1, and September 1, 2016, using a push core sampler (ø 5 cm, length 100 cm). Substrate samples were taken from the front and back parts of each HSCW at a distance of 0.3 and 1.7 m from the inlet, respectively. Three 50 g samples were taken from each sampled core at depths of − 10 cm (A1), − 30 cm (A2), and − 50 cm (A3) from the surface (Fig. 1a). Visible roots and debris were removed and samples were stored in sterile plastic bags, placed on ice, and transferred immediately to the lab for the following analyses.bilized protein was quantified using PIERCE (Rockford, IL,USA) BCA Protein Assay Reagent Kit (Cat. No. 23227) ac- cording to the method specified by Lowry et al. (1951).

Accumulated solids were detached from 5 g substrate samples using ultrasound equipment for 15 min with 100 mL phospho- rus buffer (K2HPO4, 9.3 g/L; KH2PO4, 1.8 g/L) to form an accumulated solids solution (AS solution). Initially, the AS solution was manually divided into two 50 mL portions. The first 50 mL AS solution was dried at 105 °C for 24 h. Subsequently, the dry residue was burned at 550 °C for 15 min. The weight difference between the final burned cru- cible and original clear crucible was estimated as the total inorganic matter (TIM) content. The weight difference of the crucible before and after burning was estimated as the total organic matter content. The second 50 mL AS solution was filtrated using 0.45 μm filter membrane, and the filtrate solu- tion was analyzed using the same method described above to indicate dissolved organic matter (DOM). The weight differ- ence between total organic matter (the first 50 mL AS solu- tion) and DOM (the second 50 mL AS solution) was calculat- ed as insoluble organic matter (IOM). DOM and IOM were added to calculate the accumulated total organic matter. The accumulation of total solids was the sum of TIM, DOM, and IOM. Additionally, the total organic matter and TIM wereOne-way analysis of variance at 95% confidence level (p < 0.05) was used to evaluate significant differences in pol- lutant removal efficiencies between the three HSCWs. The performance patterns of biofilm growth and solids accumula- tion for the HSCWs were analyzed using principal compo- nents analysis (PCA). XLStat Pro® (XLStat, Paris, France) was used for plotting and data analyses in the present study. Results and discussion The experiment site was located in a typical subtropical envi- ronment with maximum and minimum air temperature with a range of 22–36 and 13–28 °C, respectively, during the whole experiment (Fig. 2). Along with the season changes from April to September, the maximum and minimum air tempera- tures showed clearly improvement from approximately 25 to 30 and 17 to 25 °C, respectively. The influent water tempera- ture also showed a similar tendency which increased from around 20 to 26 °C from April to June and then kept relative stable until September (Fig. 2). Additionally, the ORP values of effluent for the three HSCWs (CW1, CW3, and CW6) were not significantly different with an average of − 100 mV. The three HSCWs also showed similar pH values in a range of 6.9–7.5 along the experiment (data are not shown). Significantly higher pollutant removal abilities were con- tinually observed in the multilayer substrate configured CW6, followed by CW3, and the monolayer CW1 throughout the experiment (Fig. 3). The average removal efficiencies of COD, NH4+–N, TN, and TP were about 70, 78, 62, and 80% for CW6; 58, 60, 49, and 63% for CW3; and 29, 40, 30, and 41% for CW1, respectively. Generally, pollutant re- moval performance before June fluctuated notably and record- ed slightly lower than average values for all the HSCWs than the corresponding values between June and September. Our previous study had reported that multilayer substrate configuration enhanced pollutant removal in the HSCW in the year following system construction (Bai et al. 2016). The higher removal efficiencies were attributed to the multilayer configuration with larger size substrate set at the bottom that improved hydraulic performance by reducing short circuit in the HSCWs. After 3 operational years, the same enhanced performance was observed in the present study, which indi- cated that the optimized multilayer configuration can promote stable and positive effect on pollutant removal. The corre- sponding values were slightly lower than values observed 3 years ago, especially for CW1, which may be due to solids accumulation and clogging (Blazejewski and Murat- Blazejewska 1997; Morales et al. 2013). Additionally, higher pollutant removal abilities after June may be due to higher temperature (Fig. 2). The microorganisms responsible for COD and nitrogen removal function optimally in relatively higher temperature conditions (Akratos and Tsihrintzis 2007; Lv et al. 2016).Multilayer substrate configured HSCWs generally recorded higher concentrations of TVC, EPS, and EP than the mono- layer HSCW in samples acquired both from the front and back parts (Fig. 4). The results indicated that multilayer configura- tion promoted biofilm growth under better hydraulic perfor- mance in the HSCWs. Regarding pollutant removal, organic matters and nitrogen were mostly degraded by substrate- attached biofilms (Lee et al. 2009; Lv et al. 2017). TVC rep- resents the absolute number of bacterial cells that contribute to pollutant degradation. Polymers are mainly composed of polysaccharide (EPS) and protein (EP), and extracellular poly- mers can trap, bind, and concentrate organic materials increas- ing their susceptibility to bacterial biodegradation (Zhao et al. 2009). Thus, the superior biofilm amount in CW3 and CW6 supported improved pollutant removal performance, as shown in Fig. 3. Seasonal differences in the biofilm development were also clearly observed. The concentrations of TVC, EPS, and EP showed distinct increase from April to July and September (Fig. 4a–c). The total TVC values in CW1, CW3, and CW6 showed similar values in July and September with a range of 3.6–3.9 × 105 CFU/g DM. However, in April, CW1 exhibited a significantly lower total TVC value (1.5 × 105 CFU/g DM) compared to CW3 (2.4 × 105 CFU/g DM) and CW6 (2.7 × 105 CFU/g DM). Similar tendencies were also observed for EPS and EP concentrations, which was probably due to warmer temperature as well as active plant growth during July Notably, the seasonal change in biofilm growth also pre- sented a spatial difference, especially for the samples collected at different depths. In April, the values of TVC located in the top layer (A1 at depth of − 10 cm) were clearly lower than the deeper layers. However, A1 TVC increased significantly in July and September, which may attributed to oxygen release from active plants and bacteria growth. The bacteria located in the top layer of the HSCW were mainly aerobic, such as het- erotrophic and nitrifying bacteria (Samsó and García 2013). These results are in agreement with Chazarenc et al. (2009) who measured higher bacterial cell count in aerated CWs with higher oxygen content. The TVC in the deeper positions of A2 and A3 were generally stable over the experimentation period, which may be mainly due to bacterial survival under anaero- bic conditions were not dramatically change. All the HSCWs exhibited significant increase in total solid contents from April to September for both front and back parts of the systems (Fig. 5). It is evident that the increase rate and accumulated amount was higher in CW1, followed by CW3 and CW6. This elevated accumulation could be a consequence of decrease in large void space availability with time. Additionally, the amount of solids accumulated probably in- creased due to biofilm development and solid-trapping ability, which showed an increasing tendency (Fig. 4). A comparison of the total solids accumulation between the different HSCWs revealed that the multilayer HSCWs had notably lower values than the monolayer CW. Optimized hydraulics with even wa- ter flow path may distribute solid trapping possibility to a larger area in each HSCW. Moreover, the total solids accumu- lation was generally always higher in the front part (In) compared to the back part (Out) of the CWs. Chazarenc et al. (2009) also indicated that more solids would be trapped in the front part of CWs when the influent begins flowing into the CWs and majority of the solid would be trapped by the substrate.Figure 6 shows the accumulation of DOM, IOM, and TIM in each CW with time. TIM was the main component of the total solids in all three CWs (72–85%), and the amount in- creased slightly from April to September. Nevertheless, it was assumed that the TIM did not play a crucial role in clogging. As confirmed by previous studies, clogging matter typically consists of highly hydrated gels and sludge with inorganic and organic solids, and the porosity loss due to inorganic solids accumulation is very small (Llorens et al. 2009). This may be due to substrate micro-pore clogging by dead bacteria or growth of organic matter such as active biofilm inside. Organic matter (OM) accumulations were predominantly composed of humic, humin, and fulvic acids, derived from lignocellulosic humic compounds and plant detritus. Nearly 63–96% of these organic matter fractions were relatively refrac- tory (Nguyen 2000). Humic compounds are highly colloidal and amorphous with high hydrophilic potential and physical binding properties. They can form complexes with small quan- tities of biological fraction to form low-density gelatinous sludge with very high water retention capacity. All these prop- erties increase the organic matter’s potential to block big pores, which consequently act as a sieve and restricts flow through of larger particulate solids (Hua et al. 2013). OM contents, includ- ing DOM and IOM, also exert crucial effect on clogging, even though their contribution is less than 20% of the total solids accumulation. As shown in Fig. 6, an increase in accumulated OM was found from April to September for all CWs, while multilayer configuration CWexhibited lower OM content com- pared to the monolayer CW. Lower OM contents in CW6 ex- hibited better characteristics to offset clogging.For the contents of DOM and IOM, a higher proportion of the DOM was found in multilayer configuration CWs com- pared to the monolayer CW, while monolayer CW had higher IOM proportion. IOM was composed of refractory matter, while DOM was mainly composed of adsorbed or entrapped residual pollutants or soluble microbial products. Thus, DOM accumulation was assumed to be better than IOM accumula- tion for clog prevention, because DOM could be flushed out at certain flow rates, while IOM stayed for longer periods. Therefore, multilayer CWs with higher DOM and lower IOM contents seemed to be more beneficial than monolayer CW to prevent clogging.Regarding solids accumulation at different depths, most solids accumulated in the bottom layers (A3 and B3) in CW1. In the multilayer CWs, solids accumulation was distrib- uted more homogeneously among the three layers during all 3 months. In the monolayer CW, preferential paths often existed in the surface layer, which led to higher solids accumulation in the middle and bottom layers, resulting in preferential water flow through the surface layer (Osorio et al. 2007). Thus, the results proved that a rational arrangement of substrates with increasing K values from the surface to bottom was more successful in achieving even dis- tribution and mitigate clogging results with different symbols are shown to compare a systems difference, b seasonal difference, and c spatial difference. The arrows in each plot represent the loading factors in the principal component analysis Solids accumulation (TIM, DOM, and IOM) and biofilm growth (TVC, EPS, and EP) in different layers (A1, A2, and A3) of the three CWs (CW1, CW3, and CW6) under different sampling seasons (April, July, and September) were analyzed using PCA to assess performance patterns (Fig. 7). The first two principal components accounted for 64.94% of the variation. The PCA results were further marked to compare the systems (Fig. 7a), season (Fig. 7b), and spatial differences (Fig. 7c). The performance patterns for different CWs were generally separat- ed, especially between the multilayer and monolayer CWs (Fig. 7a). CW1, located closer to the positive direction of PC2 (up direction), indicated high contribution of solids accumula- tion (IOM and TIM). Figure 7b shows clearly separated groups in different sampling seasons. The samples from July and September were located closer to the positive direction of PC1 compared with April, which was highly contributed by the biofilm development (higher EP, TBC, and EPS concentra- tions). The performance patterns of the samples from different layers (A1, A2, and A3) generally overlapped and did not show clear group differences (Fig. 7c). However, the differences which were not found may be due to the mixing up of different HSCWs and cause the neglect influence of the performance pattern difference. This is supported by Fig. S1, which shows that spatially different patterns were generally found for all the HSCWs when each CW was separately analyzed. Thus, the PCA plots helped visualize the changes in performance patterns under different influencing factors in the present study. It can be concluded that the multilayer substrate configuration, season, and sampling locations could significantly influence the solids accumulation and biofilm growth performance. Conclusions Multilayer substrate configuration HSCWs can promote pollutant removal efficiencies after a relatively long-term (3-year) operation compared to monolayer HSCW due to optimized hydraulics of the water flow path. Lower solids accumulation was found in multilayer HSCWs and was attributed to better use of its available bed space and even flow pattern, which prevented clog- ging in CWs.Multilayer configured HSCWs can delay clogging by pro- viding higher biofilm development for organics removal and consequently lesser solids accumulations. Principal component analysis strongly supported the visualization of performance patterns and showed that the multilayer substrate configuration, season, and sampling locations significantly influence NG25 solids accumulation and biofilm growth.