Removal of perchlorate by a lab-scale constructed wetland using achira (Canna indica L.)

Achira (Canna indica L.) has not yet been tested for its potential in removing perchlorate (ClO4−) from wastewater. In this study, constructed wetlands with and without achira were used to investigate the removal efficiency and removal mechanism of perchlorate. The results showed that more ClO4− was removed by the wetlands with achira relative to those without. Perchlorate removal in the wetlands without achira decreased with time, whereas perchlorate in the wetlands with achira was stably removed. In terms of ClO4− content, the achira tissues were in the descending order of: leaf > aerial stem > flower or rhizome > root. Perchlorate was concentrated in leaves (more than 55.8%) rather than in root (less than 0.67%). Mass balance calculation showed that plant uptake accounted for 5.81–7.34% of initial ClO4− input, while microbial degradation accounted for 29.39–62.48%. The wetlands with achira were favorable for soil microorganism growth and proliferation and in turn ClO4− biodegradation. Furthermore, the effluent pH increased in achira wetland columns and in turn promoting ClO4− removal. The results indicating that the wetlands with achira promote ClO4− removal by improving the rhizosphere environment.


Introduction
Used in such industrial productions as fireworks, rubber, and paint manufacturing, perchlorate has emerged as a contaminant in soil, water, and food. United State Environmental Protection Agency (U.S. EPA) reported that perchlorate concentrations up to 2000 mg/kg, 3700 mg/L, 120 mg/L, and 811 lg/L have been documented in soil, groundwater, surface water, and drinking water, respectively (U.S. EPA 2005), as well as airborne particulate matter ). In addition, perchlorate has also been detected in food stuffs, breast milk, baby formulas, soft drinks, and human body fluids (Her et al. 2010;Abt et al. 2016;Zhang et al. 2010;Kirk et al. 2012;Shelor et al. 2012;Calderón et al. 2017). In 2009, the US EPA identified perchlorate as a drinking water contaminant (U.S. EPA 2019), and the State of California controls the maximum contaminant level of perchlorate at 6 lg/L in drinking water (Xie et al. 2018).This has caused a major concern due to the health effects of perchlorate, which is documented to inhibit iodine uptake by the thyroid gland, decrease thyroid production, and lead to thyroid disease (Abt et al. 2016;Kumarathilaka et al. 2016).
As perchlorate ion (ClO 4 -) is of low reactivity and high mobility, it is very stable and persistent in the environment (Sevda et al. 2018). Physico-chemical and biological methods have been used for ClO 4 remediation (Choe et al. 2013;Sevda et al. 2018;Nair et al. 2020). Bioremediation, e.g., perchlorate reduction by perchlorate-reducing bacteria (PRB) or chlorate-reducing bacteria (CRB) to chloride (Cl -), has been proven to be more environmentally friendly and economical (Bardiya and Bae 2011).
Phytoremediation is a bioremediation technology where perchlorate is reduced via plant uptake, phytodegradation, and rhizo-degradation (Yifru and Nzengung 2008;Bhaskaran et al. 2013). Plant uptake and phyto-degradation are slow processes, whereas rhizodegradation is more rapid (Tan et al. 2004a, b;Tana et al. 2005). Utilizing dissolved organic carbon as a carbon and energy source, PRB can rapidly reduce perchlorate to nontoxic chloride (Dahan et al. 2017;Yifru and Nzengung 2008).
Constructed wetlands are established mainly for wastewater treatment (Sehar and Nasser 2019;Abdel-Mohsein et al. 2020). As constructed wetlands are easy for operation and maintenance, environmentally friendly, and efficient, they have been widely used in treating domestic wastewater (Bilgin et al. 2014;Fang et al. 2018), industrial wastewater (Calheiros et al. 2012;de la Varga et al. 2020), and agricultural wastewater (Vymazal and Březinová 2015;Smith et al. 2019).
In a constructed wetland system, plants play an important role in pollutant removal. Plants can bind pollutants on their roots and remove pollutants from the water. In addition, plants can mediate redox transformation of pollutants in their oxygen-rich rhizosphere. Therefore, plant species is an important factor influencing a wetland's performance. Our previous research has demonstrated that achira (Canna indica L.), a wetland species with beautiful flowers, is tolerant to perchlorate stress (He et al. 2013). Therefore, perchlorate-contaminated waters might be effectively treated by a constructed wetland using achira.
In this study, the main objectives of this work were to explore perchlorate removal potential of achira in vertical flow constructed wetlands. The selected plant species has not yet been tested for use in treatment of wastewaters containing ClO 4 -. Furthermore, achira is a beautiful plant with ornamental value, easily reproductive and has the ability to absorb nutrient contaminants. The present work studied the accumulation of ClO 4 in soil layer, roots, stem and leaves of achira in order to know the removal mechanism and mobility of ClO 4 in constructed wetlands. According to previous reports (Tan et al. 2004b;Seyfferth and Parker 2008;Yifru and Nzengung 2008), microbial and plant degradation may play a more important role than plant uptake in ClO 4 degradation in natural wetland system. Thus, microbial populations, NO 3 in water which proved as a potential competitive electron acceptor with ClO 4 were also discussed. The findings of this study are expected to provide support for high efficient wastewater treatment using constructed wetlands of high ornamental value.

Preparation of constructed wetland columns
Wetland columns were prepared using cylindrical ceramic pots (30 cm height and 26 cm inner diameter) (Fig. 1). For each pot, there was a side opening at the bottom connected to a tube for effluent collecting. Each pot was first filled with a 5-cm layer of cobble (4 kg/pot) and then a 20-cm layer of soil (8 kg/pot). The cobbles, white 1-3 cm diameter, purchased from Goldstone Powder Materials Co., LTD (China), which had been washed with tap water before use. The soil was collected from the surface layer of a paddy field in the Experimental Farm of the South China Agricultural University, air dry and through 2 mm sieve before use. The main properties of the soil are shown in Table 1.

Plant species
The wetland plant species, achira, was selected for its large biomass and high tolerance to perchlorate (He et al. 2013). Plants of the same size at their three-leaf stage were collected from a local nursery. The seedlings height were about 10-15 cm.

Experimental details
Constructed wetland columns with or without plants (achira) were prepared to treat waters containing 0 (C 0 ), 40 (C 40 ), or 100 mg/L (C 100 ) perchlorate. For the three treatments with achira, two achira seedlings at three-leaf stage were transplanted to each wetland column.
Before the experiment started, the columns were first flooded with 6 L tap water for 7 days, drained, and let stand for 1 day. Then, they were fed with 6 L tap water from the top containing 0, 40, or 100 mg/L perchlorate. Water was added every day for 7 days to compensate for evaporation loss. Then, effluents were collected on the 8th day, and the plant and soil samples were collected at the end of the experiment. During the whole experiment (53 days), the 18 wetland columns were sheltered from rain. The properties of the tap water used in the experiment are shown in Table 2.

Sample preparation
The collected effluent samples were stored in plastic bottles and kept in a refrigerator until analysis. The plants including roots were washed with tap water and blotted dry. After plant height, root length, and fresh weights of shoot and root were measured, shoots and roots were dried, ground into powder, and sieved through a 35-mesh (0.5 mm) sieve. Soil samples were freeze-dried, ground, and homogenized with a 35-mesh (0.5 mm) stainless steel sieve.
For ClO 4 concentration determination, the soil and plant samples were processed as described by He et al. (2013). Briefly, 2.0 g of plant powder were added  to remove organic materials. The first 1.0 mL filtrate is discarded and the remaining aliquots are used for analysis. Soil sample (5.0 g) was weighed into a 50-mL conical flask with 20 mL deionized water, shaken at 200 rpm and room temperature for 3 h, and centrifuged at 6000 rpm for 25 min. The supernatant was filtered through a 0.22 lm polyethersulfone membrane filter, an OnGuard H cartridge, and an OnGuard RP successively to remove solid substances, metallic ions, and hydrophobic compounds, respectively.

Perchlorate quantification
Chromatographic column IonPac AG20/AS20 was used at 30°C, the mobile phase was 30 mM KOH at 1.0 mL/min, and the injection volume was 10 lL. After filtration through a 0.22-lm filter, the sample solution was injected into the system. The retention time of ClO 4 was 15.5 min. The concentrations of ClO 4 in the samples were quantified by external calibration. Standard curves were calculated from injection of 50 to 50,000 lg/L calibration standards. A calibration standard was run along with every 10 samples, while a laboratory reagent blank, a laboratory fortified blank, and a laboratory fortified duplicate were run long with every 20 samples. Based on injection of 10 lL injections, the instrument detection limit was 20 lg/L. Quality control samples were prepared as the effluent, plant, or soil samples. The mean recovery of ClO 4 was 100 ± 10%.

Enumeration of soil microbial populations
Populations of culturable bacteria, fungi, and actinomycetes in the rhizosphere samples were estimated using the serial dilution technique. Ten gram soil were mixed with 100 mL sterile water, shaken for 20 min, let stand for 3-5 min, serially diluted, and plated onto 1/10-strength beef-protein agar (BPA), potato dextrose agar (PDA), and starch nitrate agar (SNA) for enumeration of bacteria, fungi, and actinomycetes, respectively.

Statistical analysis
The experiment adopted a completely randomized design in triplicates. Data were analyzed using the analysis of variance (ANOVA) and Duncan's multiple range tests with SPSS-13 statistical package (SPSS Inc., Chicago, IL, USA) at a significant level of P \ 0.05. Results are presented as mean ± standard error. The calculation of perchlorate removal rate was according to Tan et al. (2004b) and modified, computed as follows: where C 0 is the initial perchlorate concentration (mg/ L); V 0 is the initial volume (L); C i is perchlorate concentration at day i (mg/L); and V i is the volume at day i (L).

Results and discussion
Perchlorate removal by the constructed wetland columns Differences in perchlorate removal were observed in the different treatments (Fig. 2). In terms of average ClO 4 removal rate, the treatments were in the order planted For a same perchlorate level in influent, significantly more perchlorate was removed by the planted columns than by the unplanted columns. Furthermore, perchlorate removal in the unplanted columns diminished noticeably with running time. By contrast, perchlorate removal in the planted columns did not show a clear reduction, but increased significantly at the sixth period, being 78.46% and 55.25% in the planted columns of C 40 and C 100, respectively. In comparison, only 41.52% and 20.20% of perchlorate were removed from the influent at day 53 in the unplanted C 40 and C 100 columns, respectively. Tan et al. (2004a, b) also demonstrated that perchlorate removal by wetland columns planted with Bulrush (Scirpus sp.) was higher as compared with columns without plants. Plants are a key component of constructed wetlands and their presence is known to improve pollutant removal (Ballesteros et al. 2016). Plants absorb nutrients, heavy metals, and toxic substances from waters. In addition, they introduce an aerobic root zone and increase aerobic processes as compared with unplanted ''wetlands'' (Faulwetter et al. 2009). Furthermore, wetland plant roots secrete such substances as enzymes, organic acids, and organic carbon, which are favorable for microbial metabolism, growth, and reproduction (Wu et al. 2017). The degradation of perchlorate by rhizosphere microorganisms requires electron donors and carbon sources. Root exudates can provide electron donors and carbon sources for perchlorate-reducing bacteria (PRB) to degrade perchlorate (Tan et al. 2004a, b). In addition, plant-secreted enzymes play a catalytic role in the process of perchlorate biodegradation (Seyfferth et al. 2008a, b;Tan et al. 2004a, b;Hutchison et al. 2013).

Plant uptake
Plant height, and plant biomass in columns C 40 and C 100 were significantly lower as compared with column C 0 (Table 3), indicating that achira growth was inhibited by ClO 4 -. However, the results also demonstrated that achira could tolerate high ClO 4 -(100 mg/L). Perchlorate content in the achira tissues increased with increasing perchlorate level in the wetland system (Table 4). These findings are in line with previous findings for other wetland plants (Seyfferth and Parker 2008).
Perchlorate content in the above-water tissues (aerial stem, leaf, and flower) of achira was much  The data are presented as mean ± standard error (n = 3). Different lowercase letters in a same column indicate significant differences between treatments (Duncan's multiple range tests, P \ 0.05) higher than that in the underwater tissues (root and rhizome). The order of perchlorate content in different tissues of achira was: leaf [ aerial stem [ flower or rhizome [ root. Perchlorate content in achira leaf was 2868.51 and 10,441.06 mg/kg DW (dry weight) in column C 40 and C 100 , respectively. In contrast, lower perchlorate content in root was observed, 141.03 and 703.51 mg/kg DW in column C 40 and C 100 , respectively. It is evident that perchlorate mainly accumulated in leaf (more than 55.80%) rather than in root (less than 0.67%). Similar results have been reported in other plants (Tan et al. 2004b).
On the basis of dry weight, perchlorate bioconcentration factor (BCF) of achira leaf was calculated to be 71.71 and 104.41 in column C 40 and C 100 , respectively ( Table 4). The BCF of achira in the present study is close to that of other aquatic plants, such as smart-weed (Polygonum sp.), watercress (Nasturtium sp.), and bulrush (Scirpus sp.) (Tan et al. 2004a, b).

Mass balance analysis
The contribution of each component to the overall ClO 4 removal by the wetland system was evaluated by mass balance analysis. The non-volatile ClO 4 can be degraded to innocuous Cl -. Assuming steady-state conditions, the following mass balance equation can be derived for perchlorate in the planted wetland columns: where M in is total ClO 4 input to the wetland system (g), M out is ClO 4 output with the effluent from the wetland system (g), M p is the part absorbed by the plants (g), M b is biodegraded, the part degraded by microbial and plant (g), and M s is the part adsorbed by the soil and cobbles.
As shown in Table 5, less than 6.32% of ClO 4 input was adsorbed by the soil or cobbles in any of the treatments, about 5.81-7.34% was uptaked by achira in the planted columns, and more than 29.39% was degraded by biodegraded in any of the treatments. Tan et al. (2004b) reported that 0-14.30% of initial ClO 4 input (32 mg/L) was absorbed by plants, whereas 48.40-99.90% was biodegraded in planted columns. Nzengung et al. (1999) concluded that phytodegradation contributed to approximately 11.00% perchlorate removal in their 26-day experiment. Previous findings in the literature (Tan et al. 2004a, b) and our mass data (Table 5) indicated that microbial perchlorate reduction instead of plant uptake plays a major role in perchlorate removal by planted wetland systems. A significantly higher percentage of the initial ClO 4 input was biodegraded in the planted columns than in the unplanted columns. Of the initial ClO 4 input, 62.48% and 43.27% was biodegraded in the planted C 40 and C 100 treatments, respectively. In comparison, 45.90% and 29.39% was biodegraded in the unplanted C 40 and C 100 columns, respectively. Interestingly, achira greatly increased the proportion of perchlorate being biodegraded, indicating that plants play an important role by creating a more favorable environment for microbial growth and activity. In this study, no external carbon was supplied to the wetland columns. Carbon required by the perchlorate-reducing microbes must be self-supplied in the wetland systems. Therefore, it is expected that plant would play a more important role with time in such wetland systems as it provides both habitat and food for the indigenous perchlorate-reducing microorganisms.

DW dry weight
The data are presented as mean ± standard error (n = 3). In a same column, different lowercase letters indicate significant differences between treatments (Duncan's multiple range tests, P \ 0.05) by perchlorate and the inhibiting effect became more prominent with time. Sha et al. (2020) also showed that microbial community changed when exposed to perchlorate. For a same perchlorate input level, bacterial, fungal, and actinomycete counts were greater in the planted columns than in the unplanted ones, suggesting that achira helped to improve microbial growth in the wetland systems. Correlation analysis showed that there was a significant positive correlation between perchlorate removal and the microbial counts (Table 7). Additionally, bacteria were prevalent over fungi and actinomycetes in the wetland columns no matter with plants or not. Taken together, contaminants are removed in constructed wetlands under the synergistic action of plant-soilmicroorganism, and the symbiont interaction between microorganisms and animals and plants plays a central role (Lee et al. 2009). Microbial species and population in a constructed wetland directly affect the performance of the system.

Effluent pH value in the constructed wetlands
As shown in Table 8, the differences in effluent pH between treatments became bigger with running time, especially after the 44th day, when effluent pH in the planted columns was significantly higher than that in the unplanted columns, suggesting that achira helped to raise pH value in the wetland systems. Furthermore, in the later periods, the effluent pH values of the columns, planted or not, were in the order of: C 100 . Correlation analysis showed that there was a significant positive correlation between perchlorate removal and effluent pH (Table 7). Seyfferth et al. (2008a) demonstrated that increasing pH significantly reduced the amount of ClO 4 taken up by vegetables, for the reason is ClO 4 -/H ? cotransport across the plasma membrane. The pH variation in the constructed wetland directly affected the perchlorate removal efficiency.  Bardiya and Bae 2011). High nitrate level (above 5 mg/L) was found to inhibit ClO 4 biodegradation (Tan et al. 2004b). In the present study, NO 3 concentration in the effluent decreased gradually with the operation continuing, but increased with increasing ClO 4 concentration in both the unplanted and planted columns (Table 9). Effluent NO 3 concentration in the planted columns was significantly lower than that in the unplanted columns after 17 days of operation (Table 9). And as shown in Fig. 2, for a same perchlorate level in influent, significantly more perchlorate was removed by the planted columns than by the unplanted columns. Thus, the results demonstrated that simultaneous ClO 4 and NO 3 removal was achieved in the constructed wetlands, and achira played an important role in it. Correlation analysis showed that there was a negative, though not significant, correlation (r = -0.398, P = 0.054) between ClO 4 and NO 3 concentrations in effluent. This might be that ClO 4 may share an ion carrier or transport pathways with NO 3 - (Seyfferth et al. 2008a). Nitrate is one of the important nutrients for the growth of plants and microorganisms. In higher plants, ClO 4 and NO 3 may share transporters or transport pathways (Seyfferth et al. 2008b). Some findings have also shown that perchlorate-respiring organisms can grow with nitrate as terminal electron acceptor, and vice versa (Ziv-El and Rittmann 2009). Although interference between NO 3 and ClO 4 was observed, simultaneous removal was achieved in our wetland system. Therefore, natural achira wetlands can be used to treat perchlorate-contaminated eutrophic water. However, the design of a constructed wetland system with achira for co-treatment of perchlorate and nitrate should take nitrate concentration into account.

DW dry weight
The data are presented as mean ± standard error (n = 3). In a same column, different lowercase letters indicate significant differences between treatments (Duncan's multiple range tests, P \ 0.05) curation; HH: supervision, conceptualization; GC: writingreviewing and editing, validation.

Declarations
Conflict of interest All authors claim that they do not have any actual or potential conflict of interest including any financial, personal, or other relationships with other people or organizations.