Top Navigation Bar

Article Request Page ASABE Journal Article

Nitrate Removal by Floating Treatment Wetlands Amended with Spent Coffee: A Mesocosm-Scale Evaluation

M. G. Keilhauer, T. L. Messer, A. R. Mittelstet, T. G. Franti, J. R. Corman

Published in Transactions of the ASABE 62(6): 1619-1630 (doi: 10.13031/trans.13431). Copyright 2019 American Society of Agricultural and Biological Engineers.

Submitted for review in March 2019 as manuscript number NRES 13431; approved for publication as part of the Wetland Ecosystem Resilience Collection by the Natural Resources & Environmental Systems Community of ASABE in September 2019.

The authors are Mary G. Keilhauer, Graduate Student, School of Natural Resources, Tiffany L. Messer, Assistant Professor, Department of Biological Systems Engineering and School of Natural Resources, Aaron R. Mittelstet, Assistant Professor, Department of Biological Systems Engineering, Thomas G. Franti, Associate Professor, Department of Biological Systems Engineering, and Jessica R. Corman, Assistant Professor, School of Natural Resources, University of Nebraska, Lincoln, Nebraska. Corresponding author: Tiffany Messer, 217 L.W. Chase Hall, University of Nebraska, Lincoln, NE 68583-0730;phone:402-472-2232; e-mail:


A floating treatment wetland design was evaluated for water quality improvements.

Nitrate-N removal rates were quantified using spent coffee grounds as a carbon source.

Nitrate-N removal rates increased throughout the growing season

Abstract. The Midwestern U.S. is vulnerable to eutrophic conditions from high nutrient concentrations. Floating treatment wetlands (FTWs) are an innovative wetland design for nutrient removal from nonpoint sources and provide a unique treatment. The objectives of this project were to quantify nitrate removal in traditional and carbon-amended FTWs planted with Midwestern plant species during the establishment year. Three greenhouse experiments were conducted throughout the growing season using 18 mesocosms. Two vegetation designs were evaluated: rush species ( and ) and diverse species (, , , , , and ). Spent coffee grounds were applied to 9 of the 18 mesocosms as a carbon amendment. Nitrate-N removal increased during the establishment growing season in the FTW systems (Spring: 15.0% to 17.3%, Summer 1: 82.8% to 92.6%, Summer 2: 86.4% to 94.7%). Nitrate-N removal was also impacted by carbon amendments (FTW without amendment: 82.8% to 94.7%, FTW with amendment: 88.4% to 96.1%). Carbon additions were found to enhance denitrifying conditions even in the absence of FTWs (decreased dissolved oxygen, increased available organic carbon). Significant differences in nitrate-N removal were not observed between FTW vegetation designs. This study provides new insight on the impacts of the growing season, plant species, and carbon amendments on FTW nitrate-N removal performance during the establishment year.

Keywords. Best management practices, Carbon amendment, Floating treatment wetlands, Nitrogen removal, Spent coffee grounds

Nitrate-N (NO3-N) concentration increases in Midwestern surface waters are strongly influenced by land use, urban point-source loadings, baseflow index, and rainfall (Davies and Neal, 2007; Meynendonckx et al., 2006; Mittelstet et al., 2019). Eutrophication from nitrogen (N) and phosphorus (P) results in water quality impairments in approximately 13% (223,496 km) of rivers and 21% (15,958 km2) of reservoirs and lakes in the U.S. (USEPA, 2016). These impairments have significant repercussions for ecosystem function, hu-man health, and the economy. In an ecosystem, nutrients, specifically N and P, lead to rapid growth and die-off of algae following nutrient depletion or seasonal changes (Heisler et al., 2008; Howarth et al., 2000). Algal blooms have potential to result in anoxic conditions, fish kills, loss of biodiversity and habitat, and, in the case of toxic or harmful algal blooms, water bodies unsafe for human interaction and consumption (Carpenter et al., 1998; Bittencourt-Oliveira et al., 2016; Efting et al., 2011; Hilborn and Beasley, 2015). Further, significant economic losses are due to nutrient pollution and eutrophication from both N and P contributions (Bingham et al., 2015; Dodds et al., 2009; Iowa, 2017). There are multiple sources for the nutrients and contaminants that enter surface waters; however, nonpoint sources (NPS; i.e., agriculture, urban practices, stormwater) are reported as the dominant sources for nutrient water quality impairments (Howarth et al., 2000). NPS pollution creates a unique management challenge because nutrients are difficult to track, quantify, and regulate.

Regional nutrient challenges in the Mississippi River basin have continued to grow, with one of the largest indicators being the 15,540 km2 dead zone in the Gulf of Mexico, which has increased with increasing nitrate fluxes (Rabalais et al., 2002). One of the contributors to this impairment is NPS pollution from urban and agriculturally intensive land use in the Midwest (NRCS, 2011). Additionally, the Midwest has water resources that are important drinking and recreation sources, but surface water and groundwater in this region continue to have NO3-N levels that exceed the chronic and toxic concentration standards for human health (>10 mg NO3-N L-1). More concerning, NO3-N ingested at concentrations below 10 mg L-1 was recently reported to increase the risk of adverse birth outcomes and cancer (Temkin et al., 2019). In Nebraska, more than 80% of monitored lakes have been designated as impaired due to E. coli, mercury, or harmful algal blooms, and 90% of monitored wells exceed NO3-N concentration limits (NDEQ, 2018). Consequently, the Midwest has a significant need for cost-effective, passive water treatment systems to address the growing issue of NO3-N pollution.

Figure 1. Diagram of major nitrogen pathways in a floating treatment wetland (FTW).

To combat nutrient enrichment conditions, a collection of best management practices (BMPs) has been developed. Specifically, for NO3-N management in Nebraska, riparian buffers, wetlands, and urban BMPs (e.g., bioretention cells, permeable pavement) have been recommended for agriculturally dominated or highly developed watersheds (Mittelstet et al., 2019). Often, the greatest constraint for BMPs in reducing surface and groundwater contamination is land requirement (Frankowski, 2002). However, floating treatment wetland (FTW) systems, buoyant mats that suspend plant crowns above the water while allowing the roots to reside in the water column, have the potential to provide water quality treatment without impacting adjacent land use (Keizer-Vlek et al., 2014). FTWs provide water treatment for NO3-N, total N (TN), ammonium-N (NH4-N), total microcystin-LR, E. coli, and total P (TP) (Chauzat and Faucon, 2007; Jones et al., 2017; Keizer-Vlek et al., 2014; Mercado-Borrayo et al., 2015; Saeed et al., 2014, 2016; Tanner and Headley, 2011; Winston et al., 2013; Xian et al., 2010). Specifically, FTWs reduce the TN, TP, and total suspended solids (TSS) in influent by 4% to 88%, 8% to 53%, and 5% to 80%, respectively (Borne et al., 2013; Lane et al., 2016; Nichols et al., 2016; White and Cousins, 2013). FTWs are also cost-effective compared to traditional surface flow treatment wetlands (Koustas and Selvakumar, 2003; Tyndall and Bowman, 2016), which allows this BMP to be more affordable for low-income communities. This is due to no land requirements, minimal time and labor for installation, and implementation within the existing infrastructure, such as retention ponds (Lane et al., 2016). While FTWs have had great success in nutrient removal, nutrient reductions have been highly variable among studies. Those differences are attributed to the variability of plant type, maturity, study length, FTW placement, hydraulic retention time (HRT), surface coverage, and loading concentration (Borne et al., 2013; Lane et al., 2016; McAndrew et al., 2016; Nichols et al., 2016; White and Cousins, 2013). Therefore, improved understanding of the NO3-N removal processes in FTWs is critical for improving the design and efficiency of this BMP.

FTWs are unique due to their higher surface area interaction with NO3-N-rich water. Two major NO3-N removal processes in FTW systems include denitrification and plant uptake (fig. 1). Specifically, because NO3-N is negatively charged and not subject to soil immobilization, it moves freely in the water column (Mitsch and Gosselink, 2015). Further, FTWs have the potential for more plant uptake with NO3-N in the water column compared to traditional constructed wetlands because of the exposed roots and the large surface area contact in the water column (Vymazal, 2007). Denitrification, a microbial process carried out by facultative bacteria in anaerobic conditions in which microbial communities transform NO3- to nitrogen gas (N2), is the ideal NO3- removal process because it permanently removes N from the system (Mitsch and Gosselink, 2015). The five major environmental conditions that promote denitrification include moderate pH (6 to 8), moderate to warm temperature (18C to 24C), NO3-N presence, available organic matter, and anaerobic conditions (Messer et al., 2017a; Vymazal, 2007). Plant uptake is considered only a temporary removal mech-anism because macrophytes die and decay, and their stored N can be released back into the water column (Lane et al., 2016). The impact of plant species on NO3-N removal remains unclear, as macrophytes vary in structure, growth rate, gas exchange rate, uptake potential, and ideal nutrient removal conditions, all of which contribute to the efficiency of NO3-N removal in wetlands (Stottmeister et al., 2003).

Figure 2. (a) Greenhouse mesocosm setup, (b) the mesoLAB in June 2018 at the University of Nebraska-Lincoln, and (c) a floating treatment wetland mat (61 cm 61 cm, Beemats LLC, New Smyrna Beach, Fla.) with ten aerator pots.

Lack of carbon is a primary denitrification inhibitor in wetland systems (Burchell et al., 2007; Pulou et al., 2012). Each year, an estimated six million tons of spent coffee grounds are generated in the world as a waste product, which could be an affordable, available carbon source (Tokimoto et al., 2005). Spent coffee grounds, coffee that has already been roasted, ground, and brewed, has been identified as a sustainable waste product that has potential to enhance denitrifying conditions due to its high organic matter content (90.5 g per 100 g dried spent coffee grounds), slight acidity (pH 5 to 7), and communities of key microbes with denitrifying genes (Adi and Noor, 2009; Ballesteros et al., 2014; Mussatto et al., 2011). Furthermore, spent coffee grounds enhance water treatment in wastewater applications through adsorption of pharmaceuticals, dyes, and heavy metals (Franca et al., 2009; Ma and Ouyang, 2013; Macch et al., 1986; Sulyman et al., 2017). Therefore, to test the ability of spent coffee grounds to enhance denitrification, we amended FTW mats with this carbon source.

As FTW designs have continued to develop, research is still needed to evaluate the effects of macrophyte species and carbon amendments on NO3-N removal. The objective of this study was to quantify the removal capacity of NO3-N in FTWs based on native Midwestern vegetation types (rush species vs. diverse species), time in the growing season, and the addition of a carbon amendment (spent coffee grounds vs. no amendment). To do this, four FTW designs were evaluated throughout the growing season: (1) diverse macrophyte species, (2) diverse macrophyte species with carbon amendment, (3) rush macrophyte species, and (4) rush macrophyte species with carbon amendment.

Materials and Methods

A series of three experiments were conducted in the Messer Ecological Systems Observation Laboratory (meso-LAB), a wetland mesocosm laboratory located in a climate-controlled greenhouse at the University of Nebraska-Lincoln (UNL) (fig. 2). Atmospheric temperatures in the greenhouse remained between 23C and 26C during the day and between 17C and 19C at night, which were representative of average temperatures during the growing season (162 days between late April and early October; Dewey, 2017). The FTWs were placed in twelve 378.5 L black plastic Rubbermaid stock tanks (fig. 2b). Due to limited space in the greenhouse, six 56.8 L tanks containing only water were placed in the mesoLAB as experimental controls. Previous mesocosm studies have successfully used smaller control tanks to compare wetland treatments and open-water systems (Messer et al., 2017a, 2017b, 2019). Each tank was outfitted with a stage gauge, and the 18 tanks were randomly dispersed using a random number generator for the 18 positions available along the sides of the greenhouse (fig. 2a). The FTW mats were composed of a single 1 cm thick, 61 cm 61 cm buoyant foam mat (Beemats LLC, New Smyrna Beach, Fla.), ten 7.5 cm biodegradable Mirel plastic aerator pots, and ten macrophytes. Each aerator pot contained one macrophyte, which allowed plant roots to grow into the water column (fig. 2c).

The experiments included two distinct macrophyte mat designs: diverse species and rush species. Plant species were chosen based on their native status in Nebraska. The diverse species mats consisted of swamp milkweed (Asclepias incarnata), longhair sedge (Carex comosa), fox sedge (Carex vulpinoidea), rush (Juncus effusus, Juncus torreyi), and southern blue flag iris (Iris virginica). Two of each macrophyte were placed in the diverse species mats. Swamp milkweed and southern blue flag iris, both flowering macrophytes, were chosen for additional aesthetic and pollinator value (NRCS, 2018). The rush species mats were composed of four common rush (Juncus effusus) and six Torreys rush (Juncus torreyi) plants. All macrophytes were established in substrate for three months (January to March) in a climate-controlled greenhouse (Nebraska Statewide Arboretum, Inc., Lincoln, Neb.) prior to being planted in the FTW mats. The FTW mats covered 55% of the available surface water of the tanks, which was slightly higher than the practitioner-recommended surface water coverage of 5% to 50% for full-scale lakes (KCI, 2015; Lane et al., 2016). The control treatments contained no FTW mats or macrophytes to mimic open water in a lacustrine system. In this study, tap water was used as the source water for each experiment. Water quality grab samples were collected and analyzed at an analytical laboratory to determine background concentrations. The average concentrations were 0.78 0.50 mg L-1 of NO3-N, 0.18 0.12 mg L-1 of NH4-N, and 2.38 0.18 mg L-1 of dissolved organic carbon (DOC).

FTW Establishment Experiment (Spring)

The first experiment was designed to assess the potential of FTWs during early life stages of the macrophytes following initial planting in early spring and prior to adding carbon amendments. This experiment is referred to as Spring because it represents the FTWs at the beginning of the growing season. The experiment was performed from 9 to 19 March 2018 and was designed as a single pulse load of NO3-N with a target concentration of 10 mg L-1. This concentration was achieved by adding technical-grade KNO3 (Fisher Scientific, Pittsburgh, Pa.) directly to the stock tanks and then mixing the water vigorously with a stirring mechanism for 5 min (Avantor VWR, Radnor, Pa.).

The Spring experiment was composed of three treatments: (1) FTW with diverse species (Diverse), (2) FTW with rush species (Rush), and (3) no FTW with open water (Control). On day 0 of the experiment, the FTW and Control tanks were filled with tap water, followed by the addition of technical-grade KNO3 to achieve the target NO3-N concentration. When the water depth reached 25 cm, laboratory-grade KNO3 was added to the tanks, and the tanks were filled to a final depth of 51 cm, or approximately 295 L. The Control mesocosms were filled with tap water to a depth of 36cm, or approximately 52 L. The water volume added to each tank was recorded using a flowmeter attached to the hose (P3 International, New York, N.Y.).

Sampling occurred on days 0, 1, 3, 4, 7, and 10. Water quality grab samples were collected 0.15 cm from the air-water interface, filtered, and acidified as specified for each analysis method. Samples were analyzed for NO3-N (EPA 127-A) and NH4-N (EPA 103-A) using an AQ2 discrete analyzer (Seal Analytical, Mequon, Wisc.). DOC was analyzed using a 1010 TOC analyzer (Oceanography International, College Station, Tex.) with Standard Method 5301B (Rice et al., 2005). Water temperature, dissolved oxygen (DO), and pH were measured with a handheld YSI Pro Plus (YSI Inc., Yellow Springs, Ohio). Evapotranspiration was accounted for by measuring water depth during each sampling event with a stage gauge.

FTW Carbon Amendment Experiments(Summer 1 and Summer 2)

Two FTW mesocosm experiments were conducted to assess the impact of carbon amendment and a traditional FTW design during the summer season. NO3-N removal and water quality parameters throughout both experiments were evaluated similar to the Spring experiment. However, these experiments lasted 21 days, in contrast to 10 days, and were conducted from 14 June to 4 July 2018 and from 10 to 31 July 2018. These two experiments are referred to as Summer 1 and Summer 2, respectively. The mesocosms comprised six experimental treatments with three replicates of each treatment, for a total of 18 mesocosms: (1) FTW with diverse species (Diverse), (2) FTW with diverse species and spent coffee amendment (Diverse Coffee), (3) FTW with rush species (Rush), (4) FTW with rush species and spent coffee amendment (Rush Coffee), (5) no FTW with open water (Control), and (6) no FTW with open water and spent coffee amendment (Control Coffee). On day 0 of each experiment, all FTW mesocosms were filled with tap water to a depth of 52 cm, or approximately 303 L, and all Control treatments were filled with tap water to a depth of 27 cm, or approximately 38 L. Similar to the Spring experiment, the water volume of each tank was recorded using a flowmeter in line with a hose (P3 International, New York, N.Y.).

The experiments were designed as a single pulse load of NO3-N with a target concentration of 6 mg L-1. This value was chosen to represent Nebraska lake concentrations that were observed during the growing season in Nebraska (NDEQ, 2018). Similar to the Spring experiment, the mesocosms were dosed with a single pulse of N using technical-grade KNO3 (Fisher Scientific, Pittsburgh, Pa.). Water quality grab samples were taken before and after spiking the mesocosms to determine background NO3-N concentrations from the source water and account for differences in initial NO3-N concentrations.

Spent coffee grounds were collected from the UNL dining services. Immediately following the single NO3-N pulse load to the mesocosms, spent coffee grounds were applied as carbon amendments to the surfaces of six FTW mats and directly into the water of three of the Controls. Spent coffee ground amended mats comprised 9 of the 18 sampling units (3 Diverse Coffee, 3 Rush Coffee, and 3 Control Coffee). For the FTW mesocosms with coffee (Diverse Coffee and Rush Coffee), 430.2 g of dried spent coffee grounds were added to the surfaces of the mats. For the Control mesocosms with coffee (Control Coffee), 53.8 g of dried spent coffee grounds were added directly into the water of the three Control tanks. Coffee mass values were chosen to maintain a uniform mass of coffee per volume of water (1.4g L-1). Additional spent coffee grounds were collected and tested for extractable carbon using Standard Method 5310 and for caffeine using the SPE/LC/MS/MS multi-residue method, respectively (Kasprzyk-Hordern et al., 2007).

NO3-N, NH4-N, DO, pH, temperature, and chlorophyll a were measured in the water column on days 0, 1, 2, 3, 7, 8, 9, 10, 14, and 21 of the experiments using an EXO2 sonde (YSI Inc., Yellow Springs, Ohio). Water quality grab samples were collected throughout the experiments to validate the EXO2 sonde and account for drift throughout the experiments. Water quality grab samples were compared to EXO2 sonde readings. In the event that EXO2 sonde readings drifted, a linear regression equation was used to adjust the EXO2 sonde readings to match water quality grab sample results. All EXO2 sonde readings and water quality grab samples had coefficients of determination (R2) of >0.97. In the Summer 1 and Summer 2 experiments, a change in laboratories for water sample analyses resulted in a change in analytical methods. Water quality grab samples were analyzed for NO3-N, NH4-N, TN, and TOC using methods EPA 353.2, SM 4500-NH3, SM 4500P, and SM 5310, respectively. Water depths were measured on sampling days using permanently installed depth gauges in each mesocosm, and the depth values were used to adjust for evapotranspiration, which ranged from 0.5 to 2 cm d-1 depending on treatment and experiment. After completion of the study, one plant of each species was harvested from each treatment. Crowns and roots were separated, dried, weighed, composited by treatment, and analyzed for TN using the Dumas combustion method (Plank, 1991; Sweeney, 1989).

NO3-N Removal

NO3-N first-order removal rates were calculated for all mesocosms following each experiment using the first-order removal rate equation, which has been found to represent pulse-flow wetland systems as assessed in this study (Benjamin, 2010; Brezonik and Arnold, 2011; Messer et al., 2017c):



CT = final NO3-N concentration (mg L-1)

Co = initial NO3-N concentration (mg L-1)

t = time (d)

k = removal rate (d-1).

NO3-N concentrations were considered background values once reaching 0.2 mg NO3-N L-1, at which point the NO3-N removal rates were not evaluated further. The concentrations were predetermined based on the precision limits of the EXO2 sonde.

NO3-N percent removals were calculated in each mesocosm using the day 1 concentrations and NO3-N concentrations (Co) on the last day prior to reaching the minimum detection limit of 0.2 mg L-1 (CT) (Benjamin, 2010; Brezonik and Arnold, 2011):


Statistical Analyses

Statistical analyses were performed to determine if the NO3-N, NH4-N, and DO concentrations were significantly different over time and between design treatments. A linear mixed effects model was applied using the GLIMMIX procedure in SAS (SAS Institute, Cary, N.C.):



= overall mean response

ai = ?xed effects

j = random effects

aij = random effects due to interactions

eijk = error unaccounted for by the effects.

In this equation, i, j, and k represent the treatment (Rush, Rush Coffee, Diverse, Diverse Coffee, Control, and Control Coffee), time (day of experiment), and experiment (Spring, Summer 1, and Summer 2), respectively. The Tukey-Kramer method was used to adjust for variability in starting concentrations between treatments to support comparisons between replicates with slightly varying initial concentrations between mesocosms:



y'(t) = ratio of NO3-N concentrations at time t

y(t) = NO3-N concentration at time t

y(0) = initial NO3-N concentration.

Finally, Tukeys honest significance test was used to determine the significance of the effects of mesocosm treatment, experiment, vegetation design, and carbon amendment. All statistical comparisons were held to a level of significance of a = 0.05. All analyses were performed using SAS (SAS Institute Inc., Cary, N.C.).

Results and Discussion

NO3-N removal varied over the course of the growing season, with increasing removal rates observed once the plants were established (fig. 3 and table 1). During the Spring experiment, low NO3-N removal rates (0.02 0.01 d-1) were ob-served in comparison to the Summer 1 and Summer 2 experiments (0.10 0.01 d-1 to 0.28 0.08 d-1). The limited reduction was likely due to the small size of the macrophytes and/or macrophytes being initially rooted in substrate. While NO3-N removal rates can be higher when plants are not rooted in substrate (Pavlineri et al., 2017), the root system development in the water column would have minimized this factor in the Summer 1 and Summer 2 experiments. Additionally, during the Spring experiment, the FTW mesocosm NO3-N concentrations at the completion of the experiments were not significantly different compared to the Control (a = 0.05), while in the Summer 1 and Summer 2 experiments the NO3-N concentrations in all treatments were significantly lower than the Control at the completion of the experiment (a = 0.05). NO3-N concentrations were significantly lower in the rush (Rush and Rush Coffee) and diverse (Diverse and Diverse Coffee) FTWs compared to the Control by day 10 during the Summer1 and Summer 2 experiments. However, no significant differences were observed in NO3-N removal rates between FTW vegetation designs (a = 0.05).

Figure 3. Average NO3-N concentrations in mesocosms over time during the (a) Spring, (b) Summer 1, and (c) Summer 2 experiments. Filled symbols represent lab-tested data, and open symbols represent EXO2 sonde data. Error bars represent standard deviations.

During the Summer 1 and Summer 2 experiments, the nine mesocosms amended with coffee, including the Coffee Control, and their non-amended counterparts had significantly greater NO3-N removal (83% to 98%) compared to the Control (3% to 10%) by day 10 (a = 0.05; table 1). Significant differences were not observed between the FTW coffee-amended and Control Coffee mesocosms for the entire 21-day Summer 1 experiment and the first ten days of the Summer 2 experiment (a = 0.05). Further, within the first ten days of the Summer 1 and Summer 2 experiments, Control Coffee had percentage removals (82% to 98%) similar to the FTW coffee-amended designs (70% to 97%).

Figure 4. Photos of biomass changes during the floating treatment wetland study: (a) diverse species and (b) rush species.

Given that NH4 has the ability to transform to NO3 in aerobic conditions, it was important to quantify the NH4 in the system. Overall, NH4-N presence and transformation were low, with respect to NO3-N, with average starting concentrations of 0.49 0.02 mg L-1, 0.23 0.08 mg L-1, and 0.10 0.08 mg L-1 in the Spring, Summer 1, and Summer 2 experiments, respectively. Therefore, the contribution of NH4-N nitrification to the NO3-N concentrations was considered minimal.

Table 1. Average removal rates (k, d-1) and standard deviations in mesocosms during establishment experiment (Spring), first coffee amendment experiment (Summer 1), and second coffee amendment experiment (Summer 2). Removal rates were determined using initial NO3-N concentrations and NO3-N concentrations prior to reaching minimal NO3-N detection limits (0.20 mg L-1).
TreatmentSpringSummer 1Summer 2
k (d-1)Removal (%)k (d-1)Removal (%)k (d-1)Removal (%)
Control0.02 0.0114.7 1.30.00 0.012.6 10.10.01 0.0110.2 3.7
Control Coffee--0.19 0.0789.3 3.90.50 0.1598.0 2.3
Diverse0.02 0.0117.3 2.30.15 0.0488.3 12.70.28 0.0894.7 4.8
Diverse Coffee--0.21 0.0592.6 4.60.54 0.2696.1 6.4
Rush0.02 0.0115.0 2.40.12 0.0882.8 11.20.10 0.0186.4 4.2
Rush Coffee--0.23 0.1090.4 2.90.30 0.2988.4 10.7

Biomass Designs

While significant differences in NO3-N removal were not observed between designs (a = 0.05), implications of biomass design for NO3-N removal were assessed. Throughout the establishment growing season, there was an increase in biomass (fig. 4). Plant uptake is a major NO3-N removal pathway in wetland systems; therefore, understanding biological uptake by plants is critical for providing management recommendations. The TN content in plant roots and crowns was assessed at the completion of the experiment to provide insight into the TN stored in plant tissue at the end of the growing season (fig. 5).

Ideally, the TN content within the plants would have been determined at the beginning and end of the experiment. However, the plants were established as seedlings with limited plant tissue (as shown in fig. 4), and the TN content would have been minimal based on the weight of the plants as compared to the weight of the plants at the end of the growing season. Therefore, biomass samples were only taken at the end of the experiment. The TN contents in the plants at the end of the experiment were 43.8 11.0 g m-2 in the FTWs with rush species and 26.5 7.7 g m-2 in the FTWs with diverse species, which were comparable to TN uptake values for wetland species reported in other studies (18.6 to 29.4 g N m-2) (Keizer-Vlek et al., 2014; Xu and Shen, 2011). While the rush species had on average 60% more TN accumulation than the diverse species, a statistically significant difference between the two treatments could not be determined due to the limited sampling replications (fig. 5). The impact of the carbon amendment (spent coffee grounds) on TN content in the plant biomass at the end of the experiment was also evaluated. However, no significant differences were observed in plant biomass TN content between the amended and non-amended treatments (data not shown).

Figure 5. Average total nitrogen (TN) composition per mat at the end of the growing season. Error bars are standard deviations.

Comparison to Past Studies

Table 2. Denitrification parameters, including pH range, DO range, temperature range, and mean DOC (with standard deviations), during the Spring, Summer 1, and Summer 2 experiments. NA indicates that data were not available or collected during the study.
SpringSummer 1Summer 2
CoffeeNo CoffeeControlCoffeeNo CoffeeControlCoffeeNo CoffeeControl
to 7.89
to 8.34
to 7.80
to 8.08
to 8.67
(mg O L-1)
to 8.88
to 8.22
to 5.25
to 10.66
to 8.38
to 8.76
to 5.71
to 9.33
to 21.70
to 20.70
to 24.95
to 25.61
to 23.33
to 22.65
to 22.48
to 20.99
Mean DOC
(mg C L-1)

In our study, we evaluated the impact of plant species and carbon amendment during the growing season of the establishment year of four FTW designs. During the Summer 1 and Summer 2 experiments, NO3-N removal rates ranged from 82.8% to 96.1% for the FTW treatments (Rush, Rush Coffee, Diverse, and Diverse Coffee; table 1). Recent studies have reported variations in implementation success based on factors including NO3-N loading, season, and carbon amendment. A field-scale FTW study evaluated the removal of NO3-N following 17 storms throughout an establishment year and reported a median reduction of approximately 50%, which was significantly more NO3-N removal than the control pond (Borne et al., 2013). The NO3-N removal rates were likely less in that study compared to our study due to the lower NO3-N concentrations (~1 mg L-1), which have been found to reduce NO3-N removal rates (Messer et al., 2017c). In contrast, Winston et al. (2013) found that NO3-N concentrations were significantly greater after implementa-tion of FTWs. However, their results were likely due to the initially lower concentrations of NO3-N in the stormwater pond (0.35 mg L-1). Zhang et al. (2018) used biofilm carriers to enhance denitrifying microbial communities on FTWs, which resulted in NO3-N removals of 82.8% to 98.1%, which were similar to our results. However, we did not observe biofilm formation in our study.

Indicators of Denitrification

A focus of this study was to determine if denitrifying conditions were present in the FTWs. The five environmental conditions that promote denitrification include moderate pH (6 to 8), moderate to warm temperature (15C to 24C), NO3-N presence, available organic matter (carbon), and anaerobic conditions (DO concentrations <2 mg L-1) (Messer et al., 2017a; Vymazal, 2007). Therefore, these five denitrification requirements were evaluated (table 2). The pH and temperature were within the ideal range for denitrification throughout the experiments. During the Spring experiment, all pH values were similar (7.5 to 8). During the Summer 1 experiment, all FTWs and Control Coffee mesocosms had pH values significantly lower than the Control throughout the study (a = 0.05; data not shown). Rush had significantly higher pH values in comparison to Rush Coffee after day 10 during the Summer 1 experiment, while the Control had significantly higher pH in comparison with Control Coffee after day 3 (a = 0.05; data not shown). However, Diverse and Diverse Coffee did not display this trend. During the Summer 2 experiment, pH readings could not be taken due to equipment malfunction.

In our study, one of the most notable differences between treatments was DO, which was inversely related to the presence of the coffee amendment. All treatments had significantly reduced DO in comparison with the Control during the Summer 2 experiment (a = 0.05). Although vegetation did not have significant implications for NO3-N removal rates, rush species mesocosms had significantly higher average DO in comparison to their diverse species counterparts during the Summer 1 and Summer 2 experiments (a = 0.05; data not shown). Significant differences in DO were observed between the coffee-amended treatments (Rush Coffee, Diverse Coffee, and Control Coffee) and all other treatments (Rush, Diverse, and Control) during the Summer 1 experiment (a = 0.05). Additionally, the coffee-amended mesocosms (including Control Coffee) established ideal anaerobic conditions for denitrification within three days of experiment initiation in both the Summer 1 and Summer 2 experiments (<2 mg L-1; Messer et al., 2017a, 2017c; Vymazal, 2007). Lower DO concentrations have been found in other full-scale FTW evaluations (Lane et al., 2016; White and Cousins, 2013). Therefore, further assessment is needed to determine the aquatic health implications and stratifications of lowered DO concentrations surrounding these treatments systems.

Figure 6. Chlorophyll a concentration during coffee-amended experiments: (a) Summer 1 and (b) Summer 2.

DO availability and plant uptake of NO3-N have been reported to be inversely correlated in FTWs (Garcia Chance and White, 2018). Because there were significantly lower DO concentrations in the coffee treatments during the Summer 1 experiment, and because the rush species had significantly higher DO than the diverse species in the Summer 1 and Summer 2 experiments, we expected to see impacts of increased DO on the TN content in rush species. However, the biomass analysis showed no significant differences (a = 0.05; data not shown).

Finally, the available organic carbon in each of the FTWs was evaluated. The amount of available carbon added to the system was estimated by the amount of DOC in the background water samples and the amount of available carbon in the coffee amendment (table 2). Available carbon was significantly higher in the coffee-amended treatments compared to the un-amended treatments (a = 0.05).

Algae Contribution

To assess potential N loss due to algal blooms, chlorophyll a concentrations were measured throughout the Summer 1 and Summer 2 experiments (fig. 6). Statistically significant differences in chlorophyll a concentrations were not observed by the end of the Summer 1 and Summer 2 experiments (a = 0.05). Four values were removed due to suspected probe malfunction. Regardless, all observed values were comparable to natural concentrations in Nebraska lakes, which can vary from 23 to 113 g L-1 during non-drought years and likely provided limited N loss throughout the study (Olds et al., 2011).

Benefit of Harvest and Plant Density

A common concern with plant uptake in FTWs is the potential for macrophytes to die back at the end of the growing season and thereby release N back into the water body. However, a key component of plant die-back is the increase in available organic carbon required for denitrification (Lane et al., 2016; Messer et al., 2017a). Complete harvests of FTWs are often impractical without replacing the mat and increasing overall management expenses. Recent studies have found that approximately one-third of the TN in FTW macrophytes is incorporated into the aboveground biomass, which means that harvesting only the aboveground biomass has potential to improve permanent N removal from the water during the growing season and limit the re-release of N into the water column (Lane et al., 2016; Vymazal, 2007). However, the impact of harvesting on carbon availability for denitrification must be taken into account (Van de Moortel et al., 2012).

In this study, as the plants matured, the NO3-N removal rates increased from 0.00 or 0.02 d-1 to ~0.20 d-1 for both species designs (no coffee amendment); however, the mechanism for NO3-N removal remains unknown. Therefore, realistic goals should be developed during the establishment year of FTWs, as immediate NO3-N removal benefits are unlikely until the plants are established. Further, significant differences were not observed between the FTW macrophyte designs in this study. Therefore, aboveground harvesting should be completed based on the overall nutrient management goals, and the addition of a carbon amendment to FTWs should be considered if harvests are incorporated into the FTW management to provide the required carbon source for denitrification.

Caffeine Effects

Coffee has deoxygenation effects that could be harmful for aquatic environments (Fernandes et al., 2017). The total amount of caffeine added to the mesocosms was quantified in this study by the caffeine concentration in the spent coffee grounds prior to adding the coffee to the mesocosms. The coffee-amended mesocosms had an availability of 1.97 0.096 mg L-1 of caffeine and 0.93 0.08 g L-1 of 1,7-dimethylxanthine, a byproduct of caffeine degradation. The caffeine concentrations measured in this study were one to two orders of magnitude less than the LD50 for freshwater organisms, such as C. dubia, P. promelas, and C. dilutes (Moore et al., 2008). Therefore, the caffeine concentrations applied in this study were not expected to have negative effects on aquatic organisms due to the caffeine content of the spent coffee ground amendment.


Based on the results of this study, FTWs may be a promising method of NO3-N removal for Midwestern surface waterbodies. This study provided a novel assessment throughout the establishment year of four FTW designs. NO3-N re-moval increased throughout the growing season in the FTW systems. NO3-N treatment performance was greatest once the macrophytes were established. Compared to the Control, significant NO3-N removal was found in both the rush species and diverse species FTWs; however, significant differences in NO3-N removal based on vegetation design were not observed. Spent coffee ground amendments to FTW mats resulted in greater NO3-N removal rates in comparison to the unamended controls and FTWs, presumably due to enhanced conditions suitable for denitrification. Future work should evaluate the impact of carbon additions to nutrient-rich lakes. Specifically, evaluation of the ecotoxicological factors for carbon applications to lacustrine environments is needed prior to encouraging carbon amendments as a management practice for NO3-N removal. Further assessment of established FTWs is also needed to improve our overall understanding of how plant species and carbon amendments impact FTW effectiveness following the establishment year.


This project was supported with funding from the Robert B. Daugherty Water for Food Global Institute at the University of Nebraska-Lincoln. This project is based on research that was partially supported by the Nebraska Agricultural Experiment Station with funding from the Hatch capacity funding program (Accession No. 1014685) of the USDA National Institute of Food and Agriculture. Collaborators from the Nebraska Department of Environmental Quality made this project possible. Special thanks to Alexa Davis, Autumn Dunn, Rob Schroeder, and Sam Hansen for sampling and laboratory assistance.


Adi, A. J., & Noor, Z. M. (2009). Waste recycling: Utilization of coffee grounds and kitchen waste in vermicomposting. Bioresour.Tech., 100(2), 1027-1030.

Ballesteros, L. F., Teixeira, J. A., & Mussatto, S. I. (2014). Chemical, functional, and structural properties of spent coffee grounds and coffee silverskin. Food Bioproc. Tech., 7(12), 3493-3503.

Benjamin, M. (2010). Water chemistry (1st ed.). Long Grove, IL: Waveland Press.

Bingham, M., Sinha, S. K., & Lupi, F. (2015). Economic benefits of reducing harmful algal blooms in Lake Erie. Report 66. Gainesville, FL: Environmental Consulting & Technology, Inc.

Bittencourt-Oliveira, M. C., Cordeiro-Araujo, M. K., Chia, M. A., Arruda-Neto, J. D. T., de Oliveira, E. T., & dos Santos, F. (2016). Lettuce irrigated with contaminated water: Photosynthetic effects, antioxidative response, and bioaccumulation of microcystin congeners. Ecotoxicol. Environ. Safety, 128, 83-90.

Borne, K. E., Tanner, C. C., & Fassman-Beck, E. A. (2013). Stormwater nitrogen removal performance of a floating treatment wetland. Water Sci. Tech., 68(7), 1657-1664.

Brezonik, P. L., & Arnold, W. A. (2011). Water chemistry: An introduction to the chemistry of natural and engineered aquatic systems (1st ed.). Oxford, UK: Oxford University Press.

Burchell, M. R., Skaggs, R. W., Lee, C. R., Broome, S., Chescheir, G. M., & Osborne, J. (2007). Substrate organic matter to improve nitrate removal in surface-flow constructed wetlands. J. Environ. Qual., 36(1), 194-207.

Carpenter, S. R., Caraco, N. F., Correll, D. L., Howarth, R. W., Sharpley, A. N., & Smith, V. H. (1998). Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl., 8(3), 559-568.[0559:nposww];2

Chauzat, M.-P., & Faucon, J.-P. (2007). Pesticide residues in beeswax samples collected from honey bee colonies (Apis mellifera L.) in France. Pest Mgmt. Sci., 63(11), 1100-1106.

Davies, H., & Neal, C. (2007). Estimating nutrient concentrations from catchment characteristics across the U.K. Hydrol. Earth System Sci. Disc., 11(1), 550-558.

Dewey, K. (2017). Lincoln weather and climate: Temperature data. Lincoln, NE: University of Nebraska. Retrieved from

Dodds, W. K., Bouska, W. W., Eitzmann, J. L., Pilger, T. J., Pitts, K. L., Riley, A. J., ... Thornbrugh, D. J. (2009). Policy analysis eutrophication of U.S. freshwaters: Damages. Environ. Sci. Tech., 43(1), 12-19.

Efting, A. A., Snow, D. D., & Fritz, S. C. (2011). Cyanobacteria and microcystin in the Nebraska (USA) Sand Hills Lakes before and after modern agriculture. J. Paleolimnol., 46(1), 17-27.

Fernandes, A. S., Mello, F. V. C., Thode Filho, S., Carpes, R. M., Honorio, J. G., Marques, M. R., ... Ferraz, E. R. A. (2017). Impacts of discarded coffee waste on human and environmental health. Ecotoxicol. Environ. Saf., 141, 30-36.

Franca, A. S., Oliveira, L. S., & Ferreira, M. E. (2009). Kinetics and equilibrium studies of methylene blue adsorption by spent coffee grounds. Desalination, 249(1), 267-272.

Frankowski, K. (2002). Treatment of diffuse-source pollution with small-scale constructed wetlands. Proc. 2nd Intl. Conf. on Wetlands and Remediation (pp. 169-174). Columbus, OH: Battelle Press.

Garcia Chance, L. M., & White, S. A. (2018). Aeration and plant coverage influence floating treatment wetland remediation efficacy. Ecol. Eng., 122, 62-68.

Heisler, J., Glibert, P. M., Burkholder, J. M., Anderson, D. M., Cochlan, W., Dennison, W. C., ... Suddleson, M. (2008). Eutrophication and harmful algal blooms: A scientific consensus. Harmful Algae, 8(1), 3-13.

Hilborn, E. D., & Beasley, V. R. (2015). One health and cyanobacteria in freshwater systems: Animal illnesses and deaths are sentinel events for human health risks. Toxins, 7(4), 1374-1395.

Howarth, R., Anderson, D., Cloern, J., Elfring, C., Hopkinson, C., Lapointe, B., ... Walker, D. (2000). Nutrient pollution of coastal rivers, bays, and seas. Issues in ecology. Ecology Society of America.

Iowa. (2017). Board of Water Works Trustees of the city of Des Moines Iowa v. Sac County Board of Supervisors as Trustee of drainage districts 32 42 65 79 81 83 86 Calhoun County Board of Supervisors Sac County Board of Supervisors as Joint Trustees of drainage district. Des Moines, IA: Supreme Court of Iowa.

Jones, T. G., Willis, N., Gough, R., & Freeman, C. (2017). An experimental use of floating treatment wetlands (FTWs) to reduce phytoplankton growth in freshwaters. Ecol. Eng., 99, 316-323.

Kasprzyk-Hordern, B., Dinsdale, R. M., & Guwy, A. J. (2007). Multi-residue method for the determination of basic/neutral pharmaceuticals and illicit drugs in surface water by solid-phase extraction and ultra performance liquid chromatography-positive electrospray ionisation tandem mass spectrometry. J. Chromatog. A., 1161(1-2), 132-145.

KCI. (2015). Floating treatment wetlands: Open water installations call for alternate approach (16 March 2015). Sparks, MD: KCI Technologies. Retrieved from

Keizer-Vlek, H. E., Verdonschot, P. F., Verdonschot, R. C., & Dekkers, D. (2014). The contribution of plant uptake to nutrient removal by floating treatment wetlands. Ecol. Eng., 73, 684-690.

Koustas, R. N., & Selvakumar, A. (2003). BMP cost analysis for source water protection. Proc. AWRA Intl. Congress. Middleburg, VA: American Water Resources Association.

Lane, S., Sample, D., Lazur, A., Winston, R., Streb, C., Ferrier, D., ... Wood, D. (2016). Recommendations of the expert panel to define removal rates for floating treatment wetlands in existing wet ponds: Final report. Edgewater, MD: Chesapeake Bay Consortium. Retrieved from

Ma, X., & Ouyang, F. (2013). Adsorption properties of biomass-based activated carbon prepared with spent coffee grounds and pomelo skin by phosphoric acid activation. Appl. Surface Sci., 268, 566-570.

Macch, G., Marani, D., & Tiravanti, G. (1986). Uptake of mercury by exhausted coffee grounds. Environ. Tech. Lett., 7(1-12), 431-444. 10.1080/09593338609384430

McAndrew, B., Ahn, C., & Spooner, J. (2016). Nitrogen and sediment capture of a floating treatment wetland on an urban stormwater retention pond: The case of the rain project. Sustainability, 8(10), 972.

Mercado-Borrayo, B. M., Cram Heydrich, S., Rosas Perez, I., Hernandez Quiroz, M., & Ponce de Leon Hill, C. (2015). Organophosphorus and organochlorine pesticides bioaccumulation by Eichhornia crassipes in irrigation canals in an urban agricultural system. Intl. J. Phytoremed., 17(7), 701-708.

Messer, T. L., Birgand, F., & Burchell, M. R. (2019). Diel fluctuations of high-level nitrate and dissolved organic carbon concentrations in constructed wetland mesocosms. Ecol. Eng., 133, 76-87.

Messer, T. L., Burchell, M. R., Birgand, F., Broome, S. W., & Chescheir, G. (2017a). Nitrate removal potential of restored wetlands loaded with agricultural drainage water: A mesocosm scale experimental approach. Ecol. Eng., 106, 541-554.

Messer, T. L., Burchell, M. R., Bohlke, J. K., & Tobias, C. R. (2017b). Tracking the fate of nitrate through pulse-flow wetlands: A mesocosm scale 15N enrichment tracer study. Ecol. Eng., 106, 597-608.

Messer, T., Burchell, M., & Birgand, F. (2017c). Comparison of four nitrate removal kinetic models in two distinct wetland restoration mesocosm systems. Water, 9(7), 517.

Meynendonckx, J., Heuvelmans, G., Muys, B., & Feyen, J. (2006). Effects of watershed and riparian zone characteristics on nutrient concentrations in the River Scheldt basin. Hydrol. Earth System Sci. , 10(6), 913-922.

Mitsch, W. J., & Gosselink, J. G. (2015). Wetlands (5th ed.). Hoboken, NJ: John Wiley & Sons.

Mittelstet, A. R., Gilmore, T. E., Messer, T., Rudnick, D. R., & Heatherly, T. (2019). Evaluation of selected watershed characteristics to identify best management practices to reduce Nebraskan nitrate loads from Nebraska to the Mississippi/Atchafalaya River basin. Agric. Ecosyst. Environ., 277, 1-10.

Moore, M. T., Greenway, S. L., Farris, J. L., & Guerra, B. (2008). Assessing caffeine as an emerging environmental concern using conventional approaches. Arch. Environ. Contam. Toxicol., 54(1), 31-35.

Mussatto, S. I., Machado, E. M. S., Martins, S., & Teixeira, J. A. (2011). Production, composition, and application of coffee and its industrial residues. Food Bioproc. Tech., 4(5), 661.

NDEQ. (2018). 2018 Harmful algal blooms and bacteria sampling results. Lincoln, NE: Nebraska Department of Environmental Quality. Retrieved from

Nichols, P., Lucke, T., Drapper, D., & Walker, C. (2016). Performance evaluation of a floating treatment wetland in an urban catchment. Water, 8(6), 244.

NRCS. (2011). Geospatial Data Gateway. Washington, DC: USDA National Resources Conservation Service. Retrieved from

Olds, B. P., Peterson, B. C., Koupal, K. D., Farnsworth-Hoback, K. M., Schoenebeck, C. W., & Hoback, W. W. (2011). Water quality parameters of a Nebraska reservoir differ between drought and normal conditions. Lake Reservoir Mgmt., 27(3), 229-234.

Pavlineri, N., Skoulikidis, N. T., & Tsihrintzis, V. A. (2017). Constructed floating wetlands: A review of research, design, operation and management aspects, and data meta-analysis. Chem. Eng. J., 308, 1120-1132.

Plank, C. O. (1991). Plant analysis reference procedures for the southern region of the United States. Southern Cooperative Series Bulletin 368. Athens, GA: University of Georgia Agricultural Experiment Station.

Pulou, J., Tournebize, J., Chaumont, C., Haury, J., & Laverman, A. M. (2012). Carbon availability limits potential denitrification in watercress farm sediment. Ecol. Eng., 49, 212-220.

Rabalais, N. N., Turner, R. E., & Wiseman Jr., W. J. (2002). Gulf of Mexico hypoxia, a.k.a. The Dead Zone. Ann. Rev. Ecol. System., 33(1), 235-263.

Rice, E., Baird, R., Eaton, A., & Clesceri, L. (2005). Standard method 5310: Total organic carbon (TOC). In Standard methods for the examination of water and wastewater. Washington, DC: American Public Health Association.

Saeed, T., Al-Muyeed, A., Afrin, R., Rahman, H., & Sun, G. (2014). Pollutant removal from municipal wastewater employing baffled subsurface flow and integrated surface flow-floating treatment wetlands. J. Environ. Sci. (China), 26(4), 726-736.

Saeed, T., Paul, B., Afrin, R., Al-Muyeed, A., & Sun, G. (2016). Floating constructed wetland for the treatment of polluted river water: A pilot-scale study on seasonal variation and shock load. Chem. Eng. J., 287, 62-73.

Stottmeister, U., Wiebner, A., Kuschk, P., Kappelmeyer, U., Kastner, M., Bederski, O., ... Moormann, H. (2003). Effects of plants and microorganisms in constructed wetlands for wastewater treatment. Biotech. Adv., 22(1), 93-117.

Sulyman, M., Namiesnik, J., & Gierak, A. (2017). Low-cost adsorbents derived from agricultural by-products/wastes for enhancing contaminant uptakes from wastewater: A review. Polish J. Environ. Studies, 26(2), 479-510.

Sweeney, R. A. (1989). Generic combustion method for determination of crude protein in feeds: Collaborative study. J. Assoc. Off. Anal. Chem., 72(5), 770-774.

Tanner, C. C., & Headley, T. R. (2011). Components of floating emergent macrophyte treatment wetlands influencing removal of stormwater pollutants. Ecol. Eng., 37(3), 474-486.

Temkin, A., Evans, S., Manidis, T., Campbell, C., & Naidenko, O. V. (2019). Exposure-based assessment and economic valuation of adverse birth outcomes and cancer risk due to nitrate in United States drinking water. Environ. Res., 176, article 108442.

Tokimoto, T., Kawasaki, N., Nakamura, T., Akutagawa, J., & Tanada, S. (2005). Removal of lead ions in drinking water by coffee grounds as vegetable biomass. J. Colloid Interf. Sci., 281(1), 56-61.

Tyndall, J., & Bowman, T. (2016). Iowa nutrient reduction strategy best management practice cost overview series: Constructed wetlands. Ames, IA: University of Iowa, Department of Ecology and Natural Resource Management. Retrieve from

USEPA. (2016). Water quality assessment and TMDL infromation: National summary of state information. Washington, DC: U.S. Environmental Protection Agency.

Van de Moortel, A. M. K., Du Laing, G., De Pauw, N., & Tack, F. M. G. (2012). The role of the litter compartment in a constructed floating wetland. Ecol. Eng., 39, 71-80.

Vymazal, J. (2007). Removal of nutrients in various types of constructed wetlands. Sci. Total Environ., 380(1-3), 48-65.

White, S. A., & Cousins, M. M. (2013). Floating treatment wetland aided remediation of nitrogen and phosphorus from simulated stormwater runoff. Ecol. Eng., 61, 207-215.

Winston, R. J., Hunt, W. F., Kennedy, S. G., Merriman, L. S., Chandler, J., & Brown, D. (2013). Evaluation of floating treatment wetlands as retrofits to existing stormwater retention ponds. Ecol. Eng., 54, 254-265.

Xian, Q., Hu, L., Chen, H., Chang, Z., & Zou, H. (2010). Removal of nutrients and veterinary antibiotics from swine wastewater by a constructed macrophyte floating bed system. J. Environ. Mgmt., 91(12), 2657-2661.

Xu, J., & Shen, G. (2011). Growing duckweed in swine wastewater for nutrient recovery and biomass production. Bioresour. Tech., 102(2), 848-853.

Zhang, L., Sun, Z., Xie, J., Wu, J., & Cheng, S. (2018). Nutrient removal, biomass accumulation, and nitrogen-transformation functional gene response to different nitrogen forms in enhanced floating treatment wetlands. Ecol. Eng., 112, 21-25.