Four rectangular mesocosm tanks with horizontal subsurface flow CWs were constructed and labeled as A, B, C, and D (Fig. 1). A schematic of the experimental layout is shown in Fig. 1(c). The polyacrylic rectangular tanks had length, width, and depth of 1 m, 0.33 m, and 0.3 m, respectively. To exclude the parameter length-to-width ratio from the study and have common conditions for comparison, the tanks were constructed of the same shape and dimensions, even though in real operating CW systems the length-to-width ratio is determined from the hydraulics of the system and depends on the substrate material size. The relatively small width of the CW was chosen in order to ensure plug flow conditions in the units and the relatively large length was chosen in order to examine the pollutant removal along the units. Each tank has inlet and outlet connected to a valve for controlling the inflow and outflow. The wetlands substrate held 32 L of water therefore a batch volume of 30 L was passed through each wetland with a constant flow rate of 1.2 × 10⁻⁷ m³/sec, achieving an overall hydraulic retention time (HRT) of 3 days (Table 1). The wastewater was passed through a perforated acrylic pipe (diffuser, r = 0.75 cm) and placed across the entire width at the upstream side of the tank. These diffusers were placed horizontally, so that the wastewater flow has a uniform distribution across the tank. Similarly, an acrylic pipe was placed at the downstream end to collect the effluent sample for analysis (r = 0.25 cm). Three different porous media with various radius (D₅₀) were used (Fig. 1(d)), e.g., Cobbles (D₅₀ = 90 mm, range 30–180 mm) placed at a bottom depth of 5 cm, medium gravel (D₅₀ = 15.0 mm, range 4–40 mm) placed at a height of 17 cm, and pea gravels (D₅₀ = 10.0 mm, range 4–25 mm) placed at a height of 5 cm (Fig. 1(c)). Wetlands A and B consist of gravels as media, whereas biochar was used in Wetland C (33.15 Kg) and Wetland D (16.57 Kg) (see Table 1). The biochar used in this work were obtained from woody materials of oak tree (Quercus sp). The dried woody biomass was charred for 10 h at temperature (~600°C) at a heating rate of 10°C/min. Each Wetland was planted with Canna seedlings at equal distance from each other (Fig. 1(a) and 1(b)). Wetland A was the control and left unplanted, whereas wetlands B, C, and D were planted with the Canna seedlings. The experiments were performed after successful acclimatization and establishment of the plants and microbial community (2 months). Studies were conducted in various batches of 3 days over a period of 7 weeks. Samples were collected on every 3^rd day. A modified OCED synthetic wastewater containing peptone (160 mg/L), meat extract (110 mg/L), kaolin clay (100 mg/L), urea (30 mg/L), NaCl (7 mg/L), CaCl₂ (4 mg/L), MgSO₄ (2 mg/L), K₂HPO₄ (28 mg/L), and potassium hydrogen phthalate (425 mg/L), at pH 6.7 was used to simulate the best characteristics of domestic wastewater [22]. Although synthetic wastewater does not fully simulate the domestic wastewater but use of synthetic wastewater for such experimental studies has advantages of minimizing variations of influent characteristics and is also safer for people in the laboratory.

Synthetic wastewater was passed within wetland beds with hydraulic retention time (HRT) of 3 days in the temperature range 25 ± 2°C with average flow rate of 1.2 × 10⁻⁷ m³/sec in each wetland unit. Water samples were collected at the end of third day from the inlet and outlet over a period of seven weeks. The water samples were analyzed immediately in the laboratory for pH, total suspended solids, turbidity (2100N Hatch Turbidimeter, USA), COD, total nitrogen, NH₃, NO₃-N, and total phosphorus, PO₄-P. Water quality analysis was performed according to the respective APHA standard methods [23].

Average percentage removal was calculated from the data obtained for a period of seven weeks. The datasets were analyzed using one-way analysis of variance (ANOVA) using Microsoft Office Excel 2007 (Microsoft, USA). A significant difference was considered at the level of p < 0.05. Subsequent pair-wise comparisons were performed using Tukey post hoc tests (see supplementary information).

Biochar used in this experiment was purchased from a local market. Scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM/EDX) was used to characterize the biochar (main elements: C:90%; O₂:8%; P:0.54%; S:0.1%; K:0.38%; and Ca:0.38%). The SEM images (Fig. 2.) clearly show that the biochar was composed of irregular forms and varying particle sizes with very coarse and heterogeneous surfaces. Moreover, the SEM image also shows shallow channels of various diameters, originating from tracheid and vessel cells. The average pore size calculated from the SEM image ranged from 1 to 10 μm. These structures may be important for the high internal surface area and adsorption ability as an excellent absorbent. The pH_pzc (point of zero charge) was determined using the known method as described in [24]. The pH_pzc of biochar was 9.55. This inferred that below pH 9.55 the surface of these solids carried a net positive charge and beyond which it contained with net negative charge. The biochar had 9% moisture and 6% ash content which was determined by gravimetric method. The mineral content of the gravel was characterized using Xray fluorescence (XRF) the main elements were CaO 78.21%; SiO₂ 9.95%; MgO 5.46%; Al₂O₃ 3.51%; Fe₂O₃ 1.6%.

The average concentration and percentage removal of suspended solid are shown in Table 2 and Fig. 3. The average suspended solid removal was 98.62%. However, there was no significant difference in the removal percentage of suspended solids and turbidity among all the four wetlands. Moreover, there was no noticeable difference observed between the performance of the planted and unplanted wetlands. Other gravel-bed wetland studies using unplanted controls have reported no significant differences in their suspended solids removals compared to the planted and unplanted systems, indicating that the removal is primarily because of physical processes which are hardly affected by the plant growth and gravel type [25, 26]. Moreover the removal of suspended solids can be attributed as merely a mechanical filtering effect. Microbial breakdown of the organic portion of suspended solid however is an important slower mechanism of removal in operational wetlands. Turbidity reduction was more efficient than the suspended solid removal. The average turbidity removal was 99.33% among all the wetlands (Fig. 3) and observed within 3 days. The wetland-treated effluent was highly clarified with turbidity level < 0.5 NTU (Table 2).

The pH is an important parameter in nutrient uptake from the aqueous solutions as it determines the surface charge of the adsorbent, degree of ionization, and speciation of the adsorbate in the aqueous media [27]. The initial influent pH was 7.28. After passing the wastewater in wetland beds, pH increased to 8.0 and 7.7 in wetlands C and D, respectively (Table 2). The increase in the pH in wetland C can be explained due to high pH_pzc (~9.55) and highly alkaline nature of biochar contributed by the ash content (~6%). The observed pH ranges among all the four wetlands were within the recommended range (4.0 < pH < 9.5) for the existence of many treatment bacteria, denitrifiers, and nitrifiers. Moreover, the variation in pH occurred, can also be attributed to high concentrations of hydroxyls in the environment, carbonates, bicarbonates, sulfates, ammonia, and other ions from the biodigestion of organic matter [3].

The mechanisms involved in the nitrogen removal in CWs include volatilization, ammonification, nitrification/denitrification, plant uptake, and matrix adsorption [3, 28]. Moreover several studies also have shown the ability of biochar to retain nitrogen [29]. Wetland C containing biochar exhibited a higher removal percentage of TN of 58.2%, which was significantly greater (p < 0.05) than those of wetlands B (40.10%) and A (34.31%) (see Fig. 3 and supplementary information). The values of the TN removal were lower as compared to organic matter. This particular result was observed in most of the wetland systems, probably because total nitrogen and total phosphorus removal require longer HRTs. Furthermore, poor nitrogen removal in wetlands is generally due to limited availability of oxygen, as the removal of organic matter consumes most of the oxygen released by the plant roots [30]. Moreover, organic nitrogen (amino acids, peptides, urea, etc.) removal mechanisms also depend upon the microbial degradation process, which also requires longer retention time and extent of microbial activity. Furthermore, the conditions in the vegetated beds of HSF CWs are usually anoxic and/or anaerobic, so that the major obstacle for higher removal of nitrogen is a low rate of nitrification [28]. There was no significance variance (p > 0.05) observed between the unplanted wetland A (34.31%) and planted wetland B (40.10%) (see supplementary information). Nevertheless, several other studies confirmed the effect of vegetation on the removal of total nitrogen [2, 3].

Denitrification is believed to be the dominant and long-term mechanism, particularly when nitrate loading rates are high [3]. Several studies have demonstrated that biochar facilitates denitrification and enhances microbial activity in soil [31]. Wetland denitrification occurs in the anoxic zones of the sediments beneath an aerobic water surface layer or in anoxic microsites of a biofilm attached to plant tissue or substrata. The highly porous biochar contains a high surface area, providing anoxic condition for the microbial biofilm formation and could be the possible reasons for the significant reductions (p < 0.05) in nitrate in the case of wetland C (92.08%) as compared to wetland B (82.8%) (Fig. 3). No significance variance (p > 0.05) observed between the unplanted wetland A (76.7%) and planted wetland B (82.8%) (see supplementary information). However, several other studies confirmed the effect of vegetation on the removal of nitrates.

In CWs ammonia nitrogen removal is due to adsorption, nitrification, plant uptake and volatilization. The adsorption of ammonia on to biochar is well known [12, 32]. Moreover, ammonia adsorbed onto biochar is bioavailable and provides a source of nitrogen for plants when the biochar-NH₃ complex is placed in the soil-plant matrix [32]. Considering the net positive charge on biochar surface at the prevailing pH due to high pH_pzc (~8.55), ammonium adsorption is less evident but cannot be ruled out completely. It can be assumed that ammonia removal could be due to combine effect of adsorption, nitrification-denitrification, and volatilization (to marginal extent). The biochar containing wetland demonstrated 58.3% ammonia removal in wetland C as compared to 50.01% in wetland B (Fig. 3). In addition, the pores inside biochar and biochar particles may provide microenvironment conditions and favor the growth of microbial communities such as ammonia-oxidizing bacteria [33]. Other studies demonstrated the role of plant in the removal of ammonia; however, in this study, no significant results (p > 0.05) were obtained between the planted and unplanted beds, and this was attributed due to the difference in influent characteristics, size and shape of the mesocosm unit, and operating conditions as compared to the other studies.

Biochar has the potential to adsorb phosphorus or affect the precipitation of phosphorus [34]. The phosphorus removal is strictly associated with the physiochemical and hydrological properties of the filter material, whereas phosphate is mainly adsorbed or precipitated in filter media [3]. The efficiency of HSF CWs for phosphorus removal is shown in Fig. 3. There was a significant removal (p < 0.05) of total phosphorus in wetland C (79.5%) as compared to wetland B (71.8%). This phenomenon can be explained by point of zero charge (pH_pzc) of biochar. The pH_pzc was found to be ~9.5; therefore, the biochar surface was expected to be positively charged in most natural aqueous conditions and electrostatically attract the negatively charged phosphate species. Although the initial solution pH values in this study were ~7, the reductions of aqueous phosphate during the experiments affected the dynamics of solution pH [34]. Moreover, the vegetation plays an important role as they provide temporary storage of phosphorus. The assimilation by plants and microorganisms mainly supports the phosphorus removal in CWs [35]. The lack of vegetation in wetland A (65.6%) could be also responsible for the lower phosphorus removal as compared to the vegetated wetland B (71.8%) (p < 0.05).

The orthophosphate removal was significantly similar (p < 0.05) to the total phosphorus removal with the highest removal in wetland C (67.7%) (Fig. 3). As the biochar surface is expected to be positively charged at low pH owing to high pH_pzc value and significantly able to adsorb the free PO₄ ions. In addition, significant orthophosphate removal was also observed in wetland B (56.5%), because of the presence of high calcium oxides in gravels, causing sorption of orthophosphate ions onto the surface of gravels and/or by the precipitation of orthophosphate ions with calcium ions [36]. The presence of aluminum and iron oxalates present in gravels also affects the reduction of the orthophosphate [34]. With regard to the PO₄ removal, this study shows a significant variation (p < 0.05) in the percentage removal between wetlands A (44.6%) and B (56.5%) for the unplanted and planted units, respectively. Several studies reported the significant variation in the planted and unplanted wetlands with respect to the phosphate removal, demonstrating that plants play a role in the removal of inorganic phosphorus [37, 38].

Higher removal efficiency among the various constituents for all the units was observed for COD. Higher removal rates were observed in wetlands C (91.3%) comprising biochar as media as compared to wetland B (81.5%) (Fig. 3). The π–π bond interactions between molecules and biochar, direct molecular-biochar electrostatic attraction/repulsion, and intermolecular hydrogen bonding are some of the proposed combined mechanisms for the COD removal. Apart from the adsorption, mechanism such as precipitation, oxidation, and anaerobic digestion are also responsible for the reduction of COD [3].With regard to the organic matter (COD), several studies show a difference in the removal activity between the planted or unplanted wetlands [2]. This study also shows significant removal difference between wetlands A (78.07%) and B (81.47%), the results show that the plants play a role in the organic matter removal [3].

The effect of Hydraulic Retention time (HRT) was studied at interval of 24 hours up to 5 days. The effect of HRT variation was observed on nutrient removal among all the four wetlands. Turbidity removal (99%) was achieved among all the wetlands within the 24 hours after introducing the wastewater. The turbidity was reduced from 60.1 NTU to 1.4 NTU within 24 hours in all the wetlands (Fig. 4(a)).

Moreover, turbidity level below 0.5 NTU was achieved within 3 days. pH variation was observed in all wetlands. The pH was observed to increase from initial value of 7.7 to final 7.9 and 7.4 in wetland C and D respectively. While a gradual decrease in pH was observed in wetland A and B to 7.1 and 6.9 respectively as retention time was increased up to 5 days (Fig. 4(b)). The increase in pH as observed in wetland C and D is likely due to alkaline nature of biochar contributed by carbonates, bicarbonates. A significant TN removal was achieved within 3 day among all wetlands (Fig. 5(a)). TN removal was observed to slow down on 4 and 5^th day. Moreover, TN removal was 1.4 times higher in wetland C and D than those observed in wetland A and B on 3^rd day. In case of TP, the best TP removal was achieved in wetland C and D on 4^th Day (Fig. 5(b)). However, the TP removal slowed down after 4 days HRT, possibly suggesting that TP removal in wetland was relatively slower compared as compared to TN removal. Similar observation was seen in case of PO₄-P, where PO₄-P concentration was reduced from 16.3 mg/L to 5.9 mg/L in wetland C and 6.8 mg/L in wetland D on 4^th day (Fig. 6(a)). While, PO₄-P concentration was 9.5 mg/L in wetland A and 9.3 mg/L in wetland B on 4^th day. The COD concentration was observed to reduce drastically from 421.3 mg/L to 56.9 mg/L in wetland C and 77.5 mg/L in wetland D on 4th day (Fig. 6 (b)). However, wetland A and B, exhibited slow removal of COD and the lowest concentration was observed to be 113.5 mg/L and 103.3 mg/L in spite of increasing HRT up to 5^th Day. The prime reason for high COD removal is due to filtering, settling and adsorption process contributed due to biochar. Although, most of the nutrients were reduced up to HRT of 3 days, but for better removal efficiency it is recommended to either reduce the flow rate or increase the HRT up to 4 days.

The macrophytes growing in constructed treatment wetlands have several properties in relation to the treatment processes that make them an essential component of the design [39]. All the planted wetlands (B, C and D) showed positive growth in the nutrient rich synthetic wastewater, without any obvious symptoms of toxicity or nutrient deficiency. After a period of 3 months, mean plant height ranged from 0.8 m to 1.53 m. Root and rhizome growth accounted to be 40 % of total plant growth with maximum depth of 15–23 cm below the ground. The biomass growth was varied among the mesocosm units and no significant effect was observed with respect to biochar addition on plant growth.