Chemicals

Acetaminophen (APAP), atrazine (ATZ), and carbamazepine (CBZ) analytical standards (≥99% purities) were purchased from Sigma-Aldrich Co. (St. Louis, MO) and Honeywell Fluka^TM (Buchs, Switzerland). For dosing the mesocosms, a standard mixture was prepared in methanol at the following concentrations for each chemical: APAP, 5905.5 μg mL⁻¹; ATZ, 226.5 μg mL⁻¹; CBZ, 234.5 μg mL⁻¹. Deuterium-labeled standards of each chemical (APAP-d₄, ATZ-d₅, and CBZ-d₁₀) were purchased from Cerilliant Co. (Round Rock, TX) and Cambridge Isotope Laboratories Inc. (Tewksury, MA), and their mixture as a surrogate standard solution for analysis was prepared in methanol at a concentration of 1 μg mL⁻¹ (for each chemical). Ascorbic acid (> 99% purity) was purchased from ACROS Organics^TM (Mirris Plains, NJ, USA) through Fisher Scientific Inc. (Fair Lawn, NJ), and other reagents such as hydrochloric acid, sodium azide, formic acid, methyl tert-butyl ether (MTBE), and optima high performance liquid chromatography − mass spectrometry (HPLC − MS) grade methanol and water were purchased from Fisher Scientific Inc.

Plant pre-acclimation

Canna flaccida (common name: Canna) plants were donated by Beeman’s Nursery (New Smyrna Beach, FL). Once the C. flaccida seedlings arrived at a laboratory, their fresh weights and shoot heights were measured to facilitate selection of uniform individuals. The most uniform plants were then transplanted into net cups [225 mL (Vol.), 7.6 cm (I.D.) × 10.2 cm (depth), The Accelerator®, Stuewe and Sons, Inc., Tangent, OR, USA] and packed with horticultural-grade perlite. The net cups were then secured into pre-punched holes on buoyant mats (Beemats, LLC, New Smyrna Beach, FL, USA). These floating mats had 7.8 cm diameter holes pre-punched at a spacing of 30 cm (center-to-center) from one another. Based on the surface area of the mesocosm (11,700 cm²), the maximum number of holes available for plants was 20. The Beemats were floated on the water surface within individual 378 L capacity Rubbermaid (Atlanta, GA, USA) cattle watering tanks measuring 90 cm (l) × 130 cm (w) × 66 cm (h) (Supplementary Figure S1a). The mesocosm tanks were located on the University of Florida campus (29°38′21.3″N 82°21′30.7″W; 2401 Memorial Road, Gainesville, FL, USA). Each mesocosm tank was filled with 302 L of water. Approximately 3.8 mL of commercially available fertilizer (Dyna-Gro Liquid Grow 7-9-5 Plant Food, Richmond, CA) and 4.1 g of calcium nitrate (Southern Agricultural Insecticides, Inc., Palmetto, FL) were added to water within each mesocosm to provide nutrition and simulate eutrophic waters. Target concentrations for total nitrogen and total phosphorus were 2.95 and 1.11 mg L⁻¹, respectively. These target concentrations were selected based on environmentally relevant concentrations reported in previous studies (Reddy 1983; Reddy and De Busk 1985; Davis et al. 2001; Healy and Cawley 2002; Mallin et al. 2002; Collins et al. 2010; Wang et al. 2014; Olguín et al. 2017). To combat the development of the fungal pathogen Puccinia thaliae which causes development of the disease Orange Rust on the plants, three commercially available fungicides (myclobutanil, chlorothalonil, and tebuconazole) were applied weekly on rotation throughout the study. Plants were acclimated to the mesocosms for 40 days before adding the target emerging contaminant.

Experiment design

This study evaluated the effect of plant density on removal of contaminants from fortified water within the mesocosms. Three different plant densities were evaluated including: 20 plants per mesocosm (100% plant density treatment; T-100), 15 plants per mesocosm (75% plant density treatment; T-75), and 10 plants per mesocosm (50% plant density treatment; T-50). Triplicate mesocosms were randomly assigned to each treatment level to minimize spatial bias. Two types of controls were also included in triplicate, including mesocosms with contaminants but no plants, and mesocosms with plants but no contaminants. One day before dosing the mesocosms with target contaminants, water used to acclimate the plants was drained from the tanks and the insides of the empty tanks were scrubbed with fresh tap water without the use of detergents. The cleaned tanks were refilled with 302 L of fresh tap water, followed by the addition of the fertilizer (Dyna-Gro Liquid Grow 7-9-5 Plant Food) and calcium nitrate as described earlier. Next the CECs were added to the water to achieve initial nominal concentrations of 130 μg L⁻¹ APAP, 5.0 μg L⁻¹ ATZ, and 5.16 μg L⁻¹ CBZ. These concentrations were 50% of the highest concentrations reported in water from previous monitoring studies (Slobodnik et al. 2012; Douglas et al. 2015; Schaider et al. 2017). Following addition of the nutrients and CECs, the water in each mesocosm tank was stirred using a polyvinyl chloride (PVC) rod, and a sample of the water was collected for analysis of initial CEC concentrations. Lastly, the planted (or not) Beemats were re-floated on surfaces of the water. Weather data (temperature, humidity, and precipitation) during the study were obtained from the Weather Underground meteorological administration database (https://www.wunderground.com/history/). This study began (CECs added) in September 2017, and ran for 12 weeks.

Water sampling

Water samples were collected on the date of treatment and then 1, 2, 3, 4, 6, 8, and 12 weeks after the contaminants were added. Before sampling, the planted mats were removed from the tanks to allow access. The water in tanks was then replenished with tap water up to the 302 L mark (to standardize concentration measurements) and stirred using a PVC rod. After allowing the stirred water to settle for 3 min, 950 mL of water was collected into 1 L amber glass bottles. The pH and electrical conductivity (EC) of water were also recorded at each sampling time using a YSI 650 Multi-parameter Display system with 600XL Sonde (YSI Inc., Yellow Springs, OH, USA). An extra sample was collected from a randomly selected treatment tank each time as a quality control element to evaluate reproducibility in the sampling process. In addition, three samples were collected from a control tank with no CECs added in order to evaluate the performance of the extraction and analytical process (i.e. percent recoveries, %RSD). Soon after sampling, water samples were transported to the laboratory and stored at 4 °C after the addition of sodium azide (1 g L⁻¹) and ascorbic acid (50 mg L⁻¹) for preservation.

Chemical residue analysis

All CECs were extracted from water samples before analysis. Before extraction, water samples were warmed up to ambient temperature. Sample preparation and extraction procedures were based on the methods of Vanderford and Snyder (2006) and Yang et al. (2016), with modifications. The pH of water samples was adjusted to 3 using 1 N hydrochloric acid and 1 N sodium hydroxide, and exactly 200 μL of 1 μg mL⁻¹ surrogate standard solution was added to each sample. For matrix spike and matrix spike duplicate quality control samples, 100 μL of 1 μg mL⁻¹ standard mixture of native chemicals was additionally spiked. The samples were extracted using Oasis hydrophilic-lipophilic balance (HLB) solid phase extraction (SPE) cartridges (6 cm³, 200 mg; Waters, Milford, MA). The HLB cartridges were placed on a vacuum manifold and pre-activated by washing sequentially with 5 mL of methyl tert-butyl ether, 5 mL of methanol, and 5 mL of reagent grade water. The entire volume of the water sample was then passed through the activated SPE cartridge at a flow rate of 10 mL min⁻¹. SPE cartridges were then dried under vacuum for 30 min. CEC residues extracted onto the HLB media were next eluted into a glass tube with 5 mL of methanol, followed by 5 mL of methanol/MTBE (10/90, v/v). The collected eluate volume was reduced to approximately 0.5 mL using a RapidVap system (Model 79000-02, Labconco Co., Kanas City, MO), after which methanol was added to achieve a final extract volume of 1 mL. CEC analytes were analyzed by high pressure liquid chromatography − tandem mass spectrometry (LC − MS/MS; TSQ Quantum Discovery MAX mass spectrometer, Thermo Fisher Scientific Inc., MA). Details regarding analytical conditions for the LC − MS/MS are provided in Supplementary Material.

Statistical analysis

All data were subjected to analysis of variance (ANOVA) using Predictive Analytics Software (PASW) Statistics 18 (International Business Machines Co., Armonk, NY, USA). Duncan’s multiple range test method was used to determine differences between the control and treatments at a significance level (α) of p = 0.05.

Dissipation trend modeling

CEC dissipation trends in water were simulated using a first-order kinetic model (Equation 1). (1) $$C_{\mathrm{w}}\left(t\right)=C_0\times e^{-kt}$$(1) where C₀ and C_w(t) represent the chemical concentration in water at times zero and t, respectively; k is the dissipation constant and t is the time after chemical treatment. The aqueous half-lives (DT₅₀) for each CEC were calculated using Equation 2. (2) $${\text{DT}}_{50}=\hspace{0.17em}\text{ln}\hspace{0.17em}\left(2\right)/k$$(2)

Percentages of CEC dissipation associated with the plant component of the FTWs were calculated for each sampling period using Equation 3. (3) $$R_{\text{FTW}}=\left[C_{\text{C}}\left(t\right)–C_{\text{w}}\left(t\right)\right]/\left[C_0–C_{\text{w}}\left(t\right)\right]\times \text{1}00$$(3) where C_C(t) indicates the chemical concentration in nonplanted controls.

Environmental conditions and plant growth

A summary of environmental conditions (i.e. temperature, precipitation, and humidity) during the study is provided in Supplementary Figure S2. Mean temperatures during the study were between 21.1 − 28.9 °C for the first 5 weeks, decreasing to 10 − 20 °C for the next six weeks, and then decreasing to below 10 °C during the final week. Lower temperatures can slow the growth of cannas (Shi et al. 2004; Hi-spring 2018). Indeed, plant growth was initially rapid, with visually observable decreases in the production of new leaves as the season progressed and temperatures cooled. The possible effects of the cooling temperatures likely reduced transpiration volumes and the activity of other processes responsible for CEC removal. Throughout the entire 84-day experiment period, there were only 16 rainy days, with a total accumulation of 184.2 mm precipitation. This precipitation did not appreciably impact the water levels within the mesocosms. The relative humidity during the study was 78.2 ± 12.0% and pH values ranged from 5.35 to 7.60 in the treatments with plants, and 6.61 to 8.87 in the nonplanted controls (Supplementary Figure S3). The pH of all treatments and controls established with plants was lower than in the nonplanted controls. Measured salinities (0.65 ± 0.03 mS cm⁻¹) were below the level reported to cause salt stress to plants (Karimi et al. 2011).

At the beginning of the experiment, mean fresh weights and shoot heights of seedlings were 140.1 ± 62.9 g and 75.2 ± 7.6 cm, respectively. At the end of the experiment, fresh weights of plants harvested (549.2 ± 67.5 g) were approximately four times greater (Supplementary Figure S4). Interestingly, the majority of new biomass in all of the treatments was associated with root growth (430.1 ± 48.5 g; 78% of total biomass) (Supplementary Figure S4). In addition, plant density on the Beemats affected the total amount of biomass produced during the study. The fresh weight of individual plants in the highest-density treatment (20 plants) was approximately 1.12 and 1.26 times smaller than those in the other treatments with 15 or 10 plants, respectively.

CEC dissipation trends

From a preliminary test and previous study results (Yang et al. 2016), the accuracy of the analytical method used in this study was verified with acceptable recoveries of >99.5%. Instrument detection limits (IDLs) for native and surrogate compounds of APAP, ATZ, and CBZ ranged from 1 to 10 ng mL⁻¹. For the matrix-spiked quality control samples, percent recoveries ranged from 81.6 to 120.1% for CEC residues added at a target concentration of 100 ng mL⁻¹, with relative standard deviations of <20%. No CEC residues were detected in any of the samples collected from the control mesocosm replicates that were not spiked. Initial measured concentrations of the CECs (in the treatment and spiked control tanks) immediately after adding them to the mesocosms (week 0) were 90.6 ng mL⁻¹ APAP, 5.1 ng mL⁻¹ ATZ, and 5.0 ng mL⁻¹ CBZ. These concentrations were close to the nominal target concentrations, except for APAP which was 1.4-fold lower. Measured concentrations were used for all calculations.

The proportions of CECs removed from water within the mesocosm treatments and spiked controls with no plants at weeks 0, 1, 3, 6, and 12 are shown in Figure 1. Dissipation of CECs was generally greater in all planted treatments relative to the nonplanted controls. For APAP, 97.9 − 100% of the residues had dissipated in the planted treatments after 1 week, while only 19.2% of the residues were removed in the controls. However, all APAP residues had dissipated to below method detection limits in all planted treatments and the nonplanted controls after 2 weeks. These results demonstrate that APAP dissipation in water is relatively quick and can be accelerated by FTWs established with C. flaccida at any of the densities evaluated. The rapid dissipation of APAP in water may be associated with its vulnerability to aqueous biodegradation (Lin et al. 2010; Liang et al. 2016). Thus, in addition to uptake by canna plants, the degradation by microflora inhabiting the rhizosphere might enhance the removal of APAP in the tested mesocosm. The adsorption of APAP residues on the surface of FTWs might also contribute to removing the chemical residues from water bodies. However, a previous study reported that the adsorptive removal of APAP by a constructed wetland accounted for only 30% when evaluating adsorption and biodegradation removal co-processes (Vo et al. 2019).

Potential use of floating treatment wetlands established with Canna flaccida for removing organic contaminants from surface water

All authors

Jeong-In Hwang, Zhuona Li, Nick Andreacchio, Francisca Ordonez Hinz & Patrick Christopher Wilson

https://doi.org/10.1080/15226514.2020.1768511

Published online:

05 June 2020

Figure 1. Dissipation (% of original added) of acetaminophen (APAP), atrazine (ATZ), and carbamazepine (CBZ) associated with floating treatment wetlands (FTWs) grown with different densities of C. flaccida plants (T-100 = 20 plants, T-75 = 15 plants, and T-50 = 10 plants). Error bars represent standard deviations, and different lower case letters indicate significant differences between mean values evaluated by Duncan’s multiple range test (p < 0.05).

Display full size

ATZ dissipation within the first week ranged from 39.5 to 50.7%, increasing to 84.1 − 94.5% after 3 weeks for the planted treatments. At the end of the experiment, no ATZ residues were detected in any of the planted treatments. Contrary to the ATZ dissipation trends in the C. flaccida treatments, only 21% of ATZ residues dissipated in the nonplanted controls by the end of the 12-week experiment. These results indicate that any of the plant densities significantly increased ATZ dissipation. Several potential dissipation pathways exist for ATZ, including uptake into the plants, hydrolysis, and sorption to plant root and mesocosm surfaces. ATZ uptake into plants has been reported in some previous studies (Albright et al. 2013; Albright and Coats 2014). Particularly, it is known that ATZ residues within plants can move readily via the xylem vessels (Su and Liang 2011). Hydrolysis of ATZ occurs rapidly in strongly acidic and alkaline environments, and at elevated temperatures (Macbean 2012). However, in this experiment, the measured pH values of the treatment water (6.6 ± 0.9) were near neutral (Supplementary Figure S3), suggesting that dissipation through hydrolysis of the ATZ may have not been significant. Although adsorption of ATZ residues to the roots of C. flaccida plants could result in rapid dissipation under some conditions, this likely did not occur in the current study. Several studies have shown that adsorption of ATZ, which is a weakly basic molecule (pKa = 1.7), increases with decreasing pH (McGlamery and Slife 1966; Clay et al. 1988; Liu et al. 1995). Rhizo-microbial degradation might also be another potential pathway for some of the ATZ dissipation (Lin et al. 2018).

Compared to APAP and ATZ, CBZ dissipation was influenced the least by C. flaccida plants throughout the study. No dissipation occurred after 1 week in the low and medium plant density treatments, and only 0.5% of the residues had dissipated in the highest plant density treatment. After 6 weeks, the CBZ concentrations in the C. flaccida treatments decreased by >50%, while concentrations in the nonplanted controls decreased by 35.1% during the same time period. Unlike the other two CECs, CBZ residues were still detectable in all of the treatments at the end of the experiment. CBZ dissipation at the end of the study ranged from 73.0 − 81.8% in the planted-treatments, and 51.8% in the controls. These results demonstrate that only 21.8 − 30% of the dissipation of CBZ residues was related to the presence of C. flaccida in the planted FTWs during the 12 week study period. In contrast, Macci et al. (2015) reported 42.7% removal of CBZ from contaminated water in a gravel-based system flooded with contaminated water and planted with another Canna species (Canna indica) after 8.5 weeks. The difference in this case may be due to the presence of biofilms associated with the gravel surfaces, which they did not measure.

CBZ is a stable molecule in aquatic environments. Electrostatic potentials in the CBZ molecule can be induced from π-bonds present in the conjugated rings and electron-withdrawing amide group (Krahn and Mielck 1989; Hassan et al. 2013), which might increase its resistance to degradation in the aquatic environment. The pH of the water in the mesocosms (6.6 ± 0.9) was high enough to minimize losses of CBZ by hydrolysis. Wang et al. (2018) reported that under 254-nm UV irradiation, the degradation of CBZ residues in water was insignificant at pH >6, but drastically increased at pH <5. Under the conditions tested, CBZ dissipation due to chemical and photodegradation should have been minimal. C. flaccida plants in the FTW systems contributed some to the removal of CBZ residues, though not as much as observed with APAP and ATZ. CBZ dissipation is likely associated with uptake by C. flaccida plants and adsorption onto surfaces of plants and mesocosms. Chen et al. (2018) reported that plant uptake generally accounts for 22.3 − 51.0% of the removal of CBZ in wetland environments planted with macrophytes like Canna species. Sorption to plant and mesocosm surfaces likely contributed to the balance of the dissipation observed.

Plant densities in the FTWs did not influence dissipation of ATZ and APAP. No studies were available in the literature evaluating the effects of plant densities in FTWs for removing organic contaminants. Dosnon-Olette et al. (2010) evaluated the influence of plant density on removal of the organic fungicide dimethomorph using two species of duckweeds in cultures. They reported that increasing plant densities were positively correlated with removal rates of the fungicide. The difference in plant density relationships between theirs and the current study is likely due to morphological and growth differences between the plants used in the two studies. Duckweeds are single-leaved (up to up to 1 cm) plants that float on the surface of the water and typically have one or two non-branching roots extending a few cm into the water column below; whereas the C. flacida plants used in the current study had extensive, branched root systems. Time-dependent dissipation trends for each CEC in the T-100 treatment are shown in Figure 2. Dissipation trends were described well using a first-order kinetic model as evidenced by coefficients of determination (R²) >0.97. However, CEC dissipation trends in nonplanted controls did not fit the models as well as indicated by lower R² values ranging from 0.53 to 0.87. Particularly, the model fit (R² = 0.53) for ATZ dissipation in the controls was the lowest, and this may be because the trend was more compatible with a zero-order kinetic model (R² = 0.71). Half-lives (DT₅₀) for APAP, ATZ, and CBZ in the T-100 treatment were 3.3 h, 5.6 days, and 27.4 days, respectively, and were much shorter than those observed in nonplanted controls (6.6, 207.9, and 81.8 days, respectively). Likewise, the DT₅₀ estimates in the current study were shorter than those reported in other studies (APAP: 0.7 − 2.1 days, ATZ: 14 − 200 days, and CBZ: 69.7−∞ days, respectively) (Lam et al. 2004; Lin et al. 2010; Macbean 2012; Li et al. 2015; Hamann et al. 2016; International Union of Pure and Applied Chemistry (IUPAC) 2018).

Potential use of floating treatment wetlands established with Canna flaccida for removing organic contaminants from surface water

All authors

Jeong-In Hwang, Zhuona Li, Nick Andreacchio, Francisca Ordonez Hinz & Patrick Christopher Wilson

https://doi.org/10.1080/15226514.2020.1768511

Published online:

05 June 2020

Figure 2. Time-dependent dissipation trends for acetaminophen (APAP), atrazine (ATZ), and carbamazepine (CBZ) in T-100 treatment and control that included no C. flaccida plants. Error bars represent standard deviations.

Display full size

CEC removal efficiency

Differences in CEC concentrations between the nonplanted control and the C. flaccida treatments represent the incremental amounts of CECs removed that were associated with the presence of C. flaccida in the FTW. CEC dissipation percentages (R_FTW) associated with the plant-components of the FTWs at each sampling date are shown in Table 1. The R_FTW values were significantly influenced by the duration of exposure to plants and the type of CECs. The R_FTW values for APAP and ATZ were nearly similar between the treatments regardless of plant densities, while those for CBZ varied with the plant density. At week 1, R_FTW values for APAP were >80%, demonstrating that its removal at the time was accelerated in the presence of C. flaccida plants (likely due to uptake by plants, degradation by microflora within the rhizosphere, and adsorption to root surfaces). Following 2 weeks exposure, R_FTW values for APAP converged to zero due to the complete dissipation of APAP in the C. flaccida treatments and nonplanted controls. At the same time, R_FTW values for ATZ during the entire experimental period ranged from 76.4 to 91.7%, indicating that the plants accelerated the removal of ATZ residues. The R_FTW values did not monotonically decrease during the first 4 weeks, which was associated with the variable concentrations of ATZ measured in the nonplanted controls at the same periods (Figure 2). For CBZ, R_FTW values were null or minor for the first 2 weeks, although the value at week 1 in the T-100 treatment was calculated at 100% because the CBZ concentration in the non-planted control at the same time was measured at a higher concentration than the initial concentration. From week 3 onwards, R_FTW values for CBZ decreased from 66.6 to 81.4% at week 3 to 29.0 − 36.7% by the end of the experiment, depending on the density treatment.

Taken together, these results indicate that FTWs planted with C. flaccida are promising for remediation of surface water contaminated with some CECs. The relatively large leaf transpiration potential of cannas compared to other plant species might be one of major factors contributing to the effective removal of CEC residues (Wang et al. 2016). Moreover, the active uptake by densely developed roots (Supplementary Figure S1b), rhizo-microbial degradation, adsorption to surfaces of plants or FTWs, and natural hydrolysis could also contribute to the removal of CECs (Kadlec and Wallace 2009; Shahid et al. 2018). Dissipation due to photolysis was likely negligible since the water surfaces were covered by the Beemats and black polyethylene sheeting.