Removal of 17α-ethynylestradiol and β-estradiol using bench-scale constructed wetlands Open Access

The estrogens, β-estradiol and 17α-ethynylestradiol are estrogenic endocrine-disrupting compounds that have raised concern in the wastewater industry due to the inability of traditional methods to efficiently remove them to below concentrations of concern (Song et al. 2009). β-Estradiol is a naturally produced hormone that is often used in hormone replacement therapy (HRT), and 17α-ethynylestradiol is a synthetic hormone that can be found in human contraceptives (Aris et al. 2014; Roby 2019).

Chronic exposure to β-estradiol or 17α-ethynylestradiol has been linked to suppressed immune responses in humans, as well as a notable decrease in both male and female fertility (McAvoy 2008). Increased rates of breast cancer, endometriosis, and birth defects have been reported in populations with high exposure to these compounds (McAvoy 2008). The impacts are also prevalent for aquatic organisms, decreased fertility and sex changes favoring female organisms are most notable (Jiang et al. 2020). The compounds enter the environment through contaminated wastewater and then proceed to find their way into different environmental media (surface water, groundwater, marine waters, soils, sediment, and drinking water) (Lubliner et al. 2010). Microbial biodegradation of synthetic estrogen can occur in activated sludge, soil, and sediments (Song et al. 2009).

β-Estradiol and 17α-ethynylestradiol are typically found in aquatic environments where wastewater effluent has been discharged. Wetlands where these compounds could be discharged have optimal temperatures between 25 and 30 °C (Kadlec & Reddy 2001) and pH between 6 and 8 at which those plants are able to grow (Seybold et al. 2002). After entering the water, estrogens may sorb to soil, and once these compounds are in the soil, they may slowly biodegrade (de Mes et al. 2005). Synthetic estrogens are more prevalent in water than in soil and biodegrade more slowly (de Mes et al. 2005). Therefore, 17α-ethynylestradiol is found in much higher concentration than β-estradiol in water, and the opposite is true for soil. If any wildlife is present in the environment it is highly likely that these compounds will bioaccumulate into fatty animal tissues.

It is estimated that the maximum concentration in wastewater of natural estrogens (including β-estradiol) is 1 μg/L and of 17α-ethynylestradiol is 13.4 ng/L, with common concentrations being lower than those (de Mes et al. 2005). β-Estradiol and 17α-ethynylestradiol are not able to be removed below toxic levels by traditional sewage treatment works (STWs) despite this removal being as high as 90% by some plants (Song et al. 2009). According to Koh et al. (2008), secondary wastewater treatment is where these compounds are removed, primarily via biodegradation. Commonly this is done through a conventional activated sludge process, with the key factor in effectiveness being solids retention time. The study also analyzed trickling filters, which were shown to be relatively ineffective, removing less than 33% of β-estradiol and 17α-ethynylestradiol. Additionally, membrane bioreactors and biological nutrient removal plants are less common options for removing estrogens (Zhu et al. 2017). Membrane bioreactors are useful alternatives to activated sludge especially when solids retention time cannot be of the length that is needed for effective removal of estrogens (Zhu et al. 2017). Biological nutrient removal plants exist primarily for the removal of nitrogen and phosphorus, but they have been also proven to be effective at reducing estrogen levels as well (Koh et al. 2008). Constructed wetlands are a relatively new wastewater treatment technology and is considered an effective ecotechnology replacement for small-scale municipal facilities (Chen et al. 2006). They are particularly useful in correcting pollutants sourced from wastewater, stormwater, agricultural runoff, coal mine drainage, and more (Davis 1995). There are three main components of constructed wetlands; water, substrate, and vascular plants. In addition to these major components, there are some minor components including aquatic invertebrates and microorganisms that develop naturally as the ecosystem ages. Processes such as photodegradation, sorption, and biodegradation all aid in the removal of estrogenic compounds in constructed wetland treatment operations. This study primarily focuses on estrogenic removal by selected isolated plant species: Eichhornia crassipes (water hyacinth), Pistia stratiotes (water cabbage), and Lemna minor (duckweed), along with a composite sediment media (gravel, sand, and silty-clay layers). Estrogen removal by photodegradation, phytodegradation, sorption, and biodegradation will be accounted for by the use of a control, lacking in substrate and vegetation. A constructed wetland was analyzed on the bench scale for the effectiveness of β-estradiol and 17α-ethynylestradiol removal while being controlled for sediment and plant types. The plants assessed were grown in separate 30 L tanks that had either a single estrogen compound or a combination of the two compounds. For the purpose of accurately monitoring the disappearance of both estrogen compounds investigated, higher concentrations of both compounds were used compared with that reported in wastewater (Jiang et al. 2020). The limits of detection (LOD) for both compounds are very low and, hence, spiking at higher concentrations enable more accurate kinetic values to be observed. The observed LOD for combined estrogens during high pressure liquid chromatography (HPLC) analysis ranged from 0.2 to 0.5 mg/L in water (Bliss 2017).

Evaporation effects, standard chemical removal, and disappearance due to sediment were controlled for with the use of control tanks that were both plant matter and sediment free along with routine freshwater top offs in all tanks to offset evaporation. After several weeks, sediment and plant matter extractions were performed to determine removal efficiency.

In soil 17α-ethynylestradiol will be readily absorbed (Tang et al. 2021), whereas in water, 17α-ethynylestradiol (K_oc = 510 (SRC)) will absorb onto suspended solids, but β-estradiol (K_oc = 30,000 (SRC)) will not due to their correspondent K_oc values (Pubchem 2021a, 2021b). These compounds will not volatilize as seen in Table 1. These compounds bioconcentrate in aquatic organisms and they may be susceptible to photolysis via direct sunlight. β-Estradiol and 17α-ethynylestradiol are hydrophobic as seen in Table 1. Their log(K_ow) is greater than 1, indicating an affinity towards n-octanol rather than water. Also, log(K_ow) approaching 4.5 indicates a higher risk for bioaccumulation. Therefore, β-estradiol and 17α-ethynylestradiol have an affinity for bioaccumulation in aquatic ecosystems.

β-Estradiol and 17α-ethynylestradiol were purchased from Sigma Aldrich with greater than 98% purity (St. Louis, MO, USA). Table 1 illustrates the chemical and physical properties of these selected estradiol compounds. Pure acetone (certified HPLC grade) was purchased from Fisher Scientific (Fair Lawn, NJ, USA) alongside certified HPLC grade methanol. Deionized (DI) water and city distribution tap water (Troy, NY, USA) were used in the experiments, as well as standard preparation, and calibration standards. Seachem Flourish Phosphorus product is a solution of potassium phosphate having available phosphate (⁠⁠) of 0.3% and soluble potassium (⁠⁠) at 0.2% with 4,500 mg/L phosphate concentration (obtained from Eddie's Aquarium Centre, Cohoes, NY). Seachem Flourish Nitrogen product is a mix of nitrogen sources concentrated at 15,000 mg/L. This product has soluble potash (⁠⁠) of 2% and analyzed available total nitrogen (N) of 1% (obtained from Eddie's Aquarium Centre, Cohoes, NY). Liquid nitrogen was obtained from Air Gas (Radnor, PA, USA). The pH standardizing buffers of 4, 7, and 10 were obtained from Fisher Scientific (Fair Lawn, NJ, USA). All chemicals were used as received.

Quikrete All-Purpose Washed Gravel (Manufactured in Atlanta, GA, USA) was used as a bottom layer for the media trial aquarium. It has a uniform particles size of 10 mm and is made of crystalline silicate and limestone with a density of 2.5 g/cm³. Quikrete Play Sand (Manufactured in Atlanta, GA, USA) was used as the middle or top layer of the media trial aquarium. It has uniform particle size and a homogenous makeup of crystalline silica (quartz) sand with a white-tan colour and density of 2.5–2.7 g/cm³. The top sediment layer was lake sediment with a main component of fines (clay and silt) with a density of 2.65 g/cm³.

Eichhornia crassipes (water hyacinth) quantity of six, Pistia stratiotes (water cabbage) quantity of eight, and Lemna minor (common duckweed) quantity of approximately 1,000 were purchased from Eddie's Aquarium Centre (Cohoes, NY, USA). Plants were grown at room temperature (23 ± 2 °C) under grow lights that mimic full sunlight conditions. Each species was grown in its own tank containing, sediment, water, aquarium air stones, and thermometers.

For pH analysis, about 5.0 g (mass known) of each washed sediment were added to three separate vials. Next, 5 mL of DI water were added to each vial. The samples were then mixed with an end-over-end rotator for several days. The pH was measured with a calibrated Accumeter AB15 pH meter, Fisher Scientific (Fair Lawn, NJ, USA). To measure organic carbon (OC) content gravimetrically, two ceramic crucibles were heated at 103 °C and 550 °C in an oven and furnace respectively (Schumacher 2002). The OC content was achieved gravimetrically by dividing the difference of the mass of the oven dried (103 °C) and combusted (550 °C) sediment by the mass of the oven dried sediment. Particle size distribution (PSD) was determined gravimetrically using a stack of sieves consisting of 850 μm, 425 μm, 250 μm, 150 μm, and 75 μm sieve sizes and a pan (ordered largest on top to smallest on bottom) (Mc Lean 1982).

Five glass aquariums (22″ L × 12″ W × 13.5″ H) were setup, each containing an isolated component of a constructed wetland along with estradiol compounds (β-estradiol and/or 17α-ethynylestradiol). The first three aquariums housed a single plant species (six water hyacinths, eight water cabbages, or 1,000 duckweed), sediment media, and the estradiol compounds. The fourth aquarium was filled with sediment media (2 kg gravel, a middle layer of 2 kg sand, and a top layer of 2 kg lake sediment) and the estradiol compounds. The fifth aquarium, the control, contained only tap water and the estradiol compounds. All aquariums were filled with 28 L of dechlorinated water, and underwent a seven-day equilibration period. At this point a sample was taken from each aquarium, such would be later used as the standard during the trial. Ten solutions were prepared and stored in 1 L glass bottles. A combined theoretical concentration of 10.0 mg/L of 17α-ethynylestradiol and 3.3 mg/L β-estradiol. The final solution volume was 30 L (after addition of the estradiol and additional tap water) in each of the five tanks.

The systems were open; therefore, to offset evaporation losses the aquariums were periodically filled to the 30 L line. Each aquarium was under continuous mixing, this was achieved through air bubbling using the bubble stones in the aquariums. For the aquariums that contained plants two nutrient solutions were made: (1) Seachem Flourish Phosphorus product at 4,500 mg/L phosphate concentration and (2) Seachem Flourish Nitrogen product at 15,000 mg/L concentration. These solutions were diluted to 108 and 360 mg/L for phosphorous and nitrogen, respectively. A 10 mL aliquot of each nutrient solution was added every 2–3 days to maintain plant health.

Over 15 weeks, samples were taken from each aquarium every 48–72 hours using a Pasteur pipette, with a new pipette used for each aquarium and sampling time. Before samples were taken, the tanks were returned to 30 L by adding dechlorinated tap water. Samples were placed in 9 mL glass vials and mixed well. Approximately 1.5 mL of each sample was added to an HPLC amber glass vial. All samples were stored in a laboratory refrigerator at 5.0 ± 1.0 °C immediately after being taken and the samples remained there until analysis.

Two weeks after the 15-week trial was completed plant matter extractions were performed on the remaining plant species (water hyacinth, healthy duckweed, and visibly stressed duckweed). In total, 1 g of each species was harvested from the aquariums and freeze dried in liquid nitrogen then pulverized with a mortar and pestle into a fine powder. Resulting sample masses ranging from 50.0 mg to 76.0 mg (exact mass known) were transferred to 1.5 mL clear Eppendorf tubes and were continuously stored in liquid nitrogen to prevent plant cell degradation. This process was repeated until there were 10 samples for each of the plant specimens (water hyacinth, healthy duckweed, and visibly stressed duckweed). Exact masses for these samples can be found in Supplementary Materials. Final concentrations of chlorophylls a and b, total carotenoids, and residual contaminants (e.g., β-estradiol and 17α-ethynylestradiol) were measured by the addition of 1 mL methanol to eight of the ten vials for contaminant extraction and 1 mL pure acetone to the remaining two vials for chlorophyll and carotenoid extraction. All vials were vortexed and centrifuged so plant matter could have ample contact time with the solvent and then the solution could be separated into liquid and solid phases. The liquid phase was extracted and used for analysis.

Wet composite medium removed from the top 1.5 inches of layered media was added to six 100 mL amber glass vials to about two-thirds volume. Samples were taken from various locations on the surface to account for variability across the surface. Samples were centrifuged at 1,395 rpm for 30 minutes, separating residual water from the sediment. The separated water was removed with Pasteur pipettes. Methanol (5 mL) was added to each vial. The vials were then mixed for five days on an end-over-end rotator. The vials were centrifuged again, and the liquid was added to HPLC vials and analyzed using the method outlined below.

 Standards were made to relate concentration to absorbance. Ten samples between 0.1 and 10 mg/L of β-estradiol or 17α-ethynylestradiol in HPLC grade methanol were made and absorbance was measured via UV-VIS spectroscopy. Wavelengths of 290 μm and 291 μm were used for β-estradiol and 17α-ethynylestradiol, respectively, based on a previous UV-VIS scan for maximum absorbance for each compound investigated. Thirty-nine samples were taken from the aquariums and analyzed using UV-VIS spectroscopy at a wavelength of 290 nm. A cuvette with water from the ‘equilibration’ phase (without the compounds) of each aquarium was used to zero the spectrophotometer. Calibration curves were made for each estrogen compound and they were measured every time analysis occurred. An example of the curves created can be seen in the supplemental attachments.

Ten standards were made for the purpose of calibration curves for each chemical with concentrations ranging from 0.1 to 10 mg/L. A Shimadzu Prominence-I LC-2030C 3D Liquid Chromatograph was used for analysis (Shimadzu Scientific Instruments, Columbia, MD). The detection wavelength was 291 nm for both compounds. A methanol:water 70:30 v/v mobile phase, operating at a flow rate of 1.0 mL/min from two pumps was passed through the stationary phase of Phenomenex C18 column with 5 cm with 5 μm particle size (Torrance, CA, USA) at ambient conditions. Injection volume was 5 μL. The unknown concentrations were obtained using an external calibration curve. Standard known concentrations of each compound and of mixed compounds ranging from 0.1 mg to 11.0 mg were used to produce three separate calibration curves (e.g., β-estradiol, 17α-ethynylestradiol, and both combined). All samples were injected twice and an average area was used to convert to concentrations.

Important sediment characteristics included sediment OC content and pH. Particle size distribution was also found to be important for the top layer of silty lake sediment. The OC content is useful in determining sediment adsorption capacities of the estrogen compounds investigated this study. The unwashed gravel sediment layer proved to have the highest OC content, averaging at about 1.2% OC by weight. The sand layer had the lowest OC content at approximately 0.03% OC by weight, and the silty lake sediment layer had about 0.3% OC by weight. The ideal pH would be between 4.1 and 8.8 for the plant species included in this study to survive and be healthy (Seybold et al. 2002). Sand had the highest average pH value of 8.0, while silt had the lowest average pH value of 7.2. Gravel pH was fairly close to the sand with an average pH value of 8.0. This proved that these sediment types would not put the selected plant species at risk in the case of a complete constructed wetland setup.

The top layer of silty lake sediment needed to be tested for a PSD because, unlike the sand and gravel, there was not a uniform manufacturer specification for lake sediments. All of the sample mass of lake sediment proved to have a grain size less than 250 μm.

The surface area of each sediment medium (mass of 2 kg) was calculated assuming spherical particle shape. The sand and gravel accounted for 5 × 10⁻³% and 1 × 10⁻³% of the total surface area respectively. The remainder of the surface area was made of the silty lake media. The total surface area data obtained were useful for reaching conclusions about the adsorption mechanisms of the selected estrogen compounds.

Essential parameters for plant survival were monitored and reported for each plant species, both at the beginning and at the end of the study. Such parameters included: water pH, electrical conductivity (μSi/cm), light intensity (μmol photons/m² s), and temperature (°C). The temperature was recorded at 20 °C, and it remained constant throughout the course of the study for all plant species investigated. Likewise, the light intensity remained at 5 μmol photons/m² s for all plants both at the beginning and at the end of the study. The electrical conductivity at the beginning of the study was below the target range of 0.4–0.6 μSi/cm, thus prompting the addition of a potassium and nitrogen supplement plan, which was used during the course of the experiment. The last parameter, water pH, was deemed as suitable for plant growth both at the beginning and at the end of the study. The water pH for water hyacinth and for water cabbage did not fluctuate throughout the study, and was measured at 6.8 and at 6.4, respectively. Lastly, the water pH for duckweed decreased from 6.8 to 6.7. The 15-week study had limited seasonal variations and thus both lighting and temperature data remained constant within their appropriate ranges throughout the study.

Samples were taken from each trial aquarium over 15 weeks. UV-VIS analysis was unable to separate concentrations of β-estradiol from that of 17α-ethynylestradiol as the maximum absorbance wavelengths only had a difference of 1 nm. A combination of UV-VIS and HPLC analysis was used to find combined estrogen removal efficiencies and combined estrogen concentration over time. HPLC was used to determine removal efficiency and concentration over time for each compound separately.

A comparison of removal rates was implemented, rather than a comparison of end result concentrations, because solid β-estradiol and 17α-ethynylestradiol are somewhat difficult to dissolve into water. This created slight variations in the initial concentrations of the estrogen compounds from the intended 3.3 or 5 mg/L at the start of the experiment. Removal efficiencies (%) were calculated by subtracting the concentration at a given time t (C_t) (mg/L) from the initial concentration (C_i) (mg/L) and then dividing by the initial concentration (C_i) and multiplied by 100%. This value is then reported as a percent removal and is reported in Table 2.

Combined estrogen concentration over time was analytically measured using a UV-VIS spectrophotometer with peak absorbance wavelengths of 290 μm for β-estradiol and 291 μm for 17α-ethynylestradiol. Varying dilutions of combined chemicals in methanol were used to generate calibration curves, which were used to calculate effluent sample concentrations. An initial combined estrogen concentration of 10 mg/L was expected for each trial aquarium. Table 2 includes a summary of the initial and final concentrations determined through the calibration curve relationship between concentration and absorbance in the case of the UV-VIS spectrophotometer or area in the case of the HPLC.

The hyacinth trial had to be ended prematurely due to a significant amount of algal bloom (sudden increase in algal biomass) due to a suspected nutrient overload in the water column. This was evidenced by a corresponding change in the water parameters. The nutrient overload was most likely caused by the slow growing nature of water hyacinth and the steady decline to 50% of the starting plant population by the end of the study. As a result of this, the water hyacinth trial was only 1,093.5 hours, rather than the complete 2,530.3 hours. Despite this, water hyacinth had the second highest overall removal efficiency at 72% at the end of the hyacinth trial. Figure 1 illustrates the removal efficiency over time for water hyacinth, along with the other two plants, and it demonstrated a dramatic early decline in the total estrogen concentration before slowly levelling off towards the end of the trial.

This trial had an observed second-order reaction rate constant of approximately 0.002 Ls/mg. The initial combined estrogen concentration of the water hyacinth tank was 8.3 mg/L. The final concentration, reported after 1,093.5 hours of study, of the combined estrogen compounds was 2.3 mg/L (see Table 2).

Water cabbage was selected for this experiment due to its widespread use in decorative constructed ponds and the resultant ease of acquisition. However, this plant species proved unable to thrive in a laboratory setting and thus the trial had to be prematurely ended after 12 days (248 hours) due to decaying plant matter skewing the UV-VIS absorbance values of water samples. Despite the reduced trial time, water cabbage was still able to achieve a removal efficiency of 35%. It is likely that, given a more conducive growing environment, the water cabbage would have been able to reach a higher removal efficiency over the complete study. Figure 1 shows yet another rapid early decline in total estrogen concentration, but not quite as dramatic as the water hyacinth. This trial followed zero-order reaction kinetics with a reaction rate of 0.0137 mg/L h. The initial concentration of combined estrogen in the water cabbage tank was 11.9 mg/L. Over the course of the 284-hour trial this value was reduced to a final concentration of 7.8 mg/L.

Duckweed was by far the hardiest and easily maintained of all of the tested plant species in this trial. The entire population remained healthy throughout the experiment and achieved an almost complete total estrogen removal efficiently of 96%. Figure 1 shows the removal efficiency of duckweed throughout the complete 2,530-hour trial. The duckweed was able to remove over 40% of total estrogens after only a few hours of experimentation, and reached over 80% by the 1,000-hour mark. However, it seems as though the duckweed's removal efficiency slowed down by the end of the trial, granted there was not much estrogen left to remove at that point of the trial. This trial had an observed first-order reaction rate of 77.8 ⁠. The combined estrogen initial concentration in the duckweed tank was 12.8 mg/L and the final concentration was 0.5 mg/L over the course of the complete 2,530.3-hour trial.

Figure 2 shows a removal efficiency greater than 90% from the sediment media aquarium within the first 409 hours of experimentation. However, the absorbance values corresponding to these data suggested an initial combined estrogen concentration greatly exceeding the actual initial concentration recorded at the start of the trial. Effluent samples in the first 409 hours produced absorbance values between 0.100 and 0.220, up to ten times greater than the expected 0.050 absorbance. This implied that there was a significant amount of turbidity within the sediment media tank interfering with the collection of absorbance data. This same high turbidity was found in this particular aquarium during the wetland construction phase, and as previously stated, a full 14 days were required to allow sediment particles to settle out of the water column. It is believed that this same phenomenon occurred when the estrogen was added into the aquarium, thus interfering with the first 14 days (409 hours) of effluent data collection. Using data beyond hour 409 produced a more realistic removal efficiency value of 9%. This trial had an observed first-order reaction rate of 0.001 ⁠. The initial concentration was 10.6 mg/L for the sediment media aquarium recorded at hour 485.5 due the turbidity problem explained above (see Table 2). A final combined estrogen concentration was found to be 9.6 mg/L at the end of the complete 2,530.3-hour study. The sediment medium was only effective at removing less than 1 mg/L of combined estrogen during this trial.

The control aquarium, containing only tap water and estrogen compounds, had the greatest variability over the trial period. Removal rates increased and decreased seemingly randomly over time as illustrated in Figure 2. This control aquarium had a large surface area open to the atmosphere, especially when compared with the planted aquariums. After the complete 2,530.3-hour trial the effluent absorbance value was a 56% removal rate. Laboratory temperature fluctuations, relatively low solubility of estrogen compounds and large surface areas open to the atmosphere all provide possible reasons for the wide removal efficiency fluctuations observed in this trial. This trial had an observed first-order reaction rate of 0.0002 ⁠. A reasonable initial concentration of combined estrogen was found to be 8.26 mg/L. The final concentration was 3.67 mg/L by the end of the 2,530.3-hour study.

The mass of each extracted chemical was determined using Lichtenthaler & Buschmann (2001) linear equations between absorbance and concentration. Negative values indicated zero concentration. The stressed (brown) duckweed had values of 0.02 μg/mg of fresh weight (FW) of chlorophyll a, and no chlorophyll b. The stressed duckweed also had the highest concentration of total carotenoids with a value of 0.08 μg/mg FW. These values are indicative of high stress. Water hyacinth had lower values of chlorophylls a and b and total carotenoids in comparison to the stressed and non-stressed duckweed as illustrated in Figure 3.

Stressed duckweed (brown leaves), healthy duckweed (bright green leaves), and water hyacinth (green leaves) were subjected to methanol extraction to determine if a residual concentration of estrogen compounds was present. HPLC analysis was used and applied to the three HPLC calibration curves as described previously. Healthy and stressed duckweed exhibited similar concentrations of selected estrogen compounds, whereas water hyacinth had significantly lower residual concentrations. These are illustrated in Figure 4.

Sediment samples were taken from the sediment media aquarium ranging from 50 to 65 g (known mass) and analyzed for estrogen compound concentrations. The analysis found an average concentration of 6.9 × 10⁻⁵ mg/g of sediment across the samples. Considering the approximately 6,000 g of sediment, 0.4 mg of estrogen compound removal could be attributed to sorption to the sediment. This mass was less than 2% of the removed compounds according to UV-VIS analysis of the liquid phase. This indicated relatively low efficiency of the methanol extraction method described previously.

This 15-week study aimed to study isolated components of constructed wetlands to better understand their estrogen compound treatment capabilities and to bring to light the potential complexities of sustaining the required aquatic plant species in a laboratory setting. UV-VIS analysis was used to quantify the removal efficiencies of the following wetland components: (i) duckweed, (ii) water hyacinth, (iii) water cabbage, and (iv) a sediment composite with the respective resulting removal efficiencies of 96, 72, 35, and 9%. Final combined estrogen concentrations were found to be 0.5 mg/L, 2.3 mg/L, 7.8 mg/L, 9.6 mg/L for duckweed, water hyacinth, water cabbage, and the sediment composite, respectively, down from the initial 8.6 mg/L concentration. Whereas the water hyacinth and water cabbage had abbreviated study durations due to a biomass influence, there is hope that higher removal efficiencies would be observed if given longer contact time. HPLC analysis was used to reduce researcher error, to allow for a second method of data capture in the first 409 hours of study, and to determine the detection limit of 0.2–0.5 mg/L combined estrogen concentration in water. The study of the effectiveness of isolated wetland component's estrogen removal proved important in seeking a solution for reducing the concentrations of persistent synthetic estrogen compounds in our aquatic environments.

The authors would like to thank the Department of Civil and Environmental Engineering at Rensselaer Polytechnic Institute (RPI) for providing a nurturing learning environment for SNB. In addition, Dr Eyosias Ashenafi is acknowledged for his thorough review of the manuscript and helpful suggestions. Lastly, the authors would like to acknowledge the Editor and anonymous reviewer for improving this manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.