Environmental Science and Pollution Research (2021) 28:52306–52318

https://doi.org/10.1007/s11356-021-14483-4

RESEARCH ARTICLE

Crushed recycled glass as a substrate for constructed wetland

wastewater treatment: a case study of its potential to facilitate
pharmaceutical removal
Luis G. Chaves-Barquero 1,2 & Braedon W. Humeniuk 1 & Kim H. Luong 3 & Nazim Cicek 4 & Charles S. Wong 1,2,3,5 &
Mark L. Hanson 1

Received: 8 December 2020 / Accepted: 14 May 2021 / Published online: 18 May 2021
# The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract
The use of recycled glass as a substrate for constructed wetlands was assessed through two studies. The first study examined the
dissipation of atenolol, carbamazepine, and sulfamethoxazole in mesocosm-modeled wetlands using glass or limestone gravel as
substrates, with or without cattails (Typha spp.). Following pseudo-first-order kinetics, atenolol dissipated the fastest from the
water surface of the mesocosms (t1/2~1 day), followed by sulfamethoxazole (t1/2~14 days), and carbamazepine (t1/2~48 days),
with no significant differences across treatments. Increased half-lives were observed at greater depth, likely due to light screening.
A Monte Carlo sensitivity analysis diagnosed sunlight absorption rates and second-order hydroxyl-mediated indirect photolysis
rates to be the main sources of uncertainty in our dissipation rate estimates, compared to our observed rates. The second study
examined in situ pharmaceutical removal in tertiary pilot-scale subsurface filters made of crushed recycled glass or sand in a
wastewater treatment facility in Manitoba, Canada. Glass and sand showed no significant differences for pharmaceutical re-
movals; atenolol and metoprolol were removed below limits of detection, while carbamazepine and sulfamethoxazole persisted
over a retention time of 24 h. Overall, recycled glass performed similarly to traditional substrates for wetland-based wastewater
treatment.

Keywords Pharmaceuticals . Constructed wetlands . Recycled glass . Mesocosms . Wastewater . Subsurface filtration

Introduction into surface waters. Pharmaceuticals and personal care prod-

ucts have been detected ubiquitously in surface and wastewa-
The entry of pharmaceuticals into the aquatic environment ters worldwide as a consequence of their incomplete removal
occurs mostly through the discharge of wastewater effluents in wastewater treatment systems (Oulton et al. 2010; Uslu
et al. 2013; Vidal-Dorsch et al. 2012) typically in the ng/L to
Responsible Editor: Alexandros Stefanakis μg/L range (Cizmas et al. 2015). Potential impacts of these
contaminants, especially their chronic effects on the environ-
* Luis G. Chaves-Barquero ment and human health are not fully understood (Li et al.
lchaves@itcr.ac.cr 2015; Zhu et al. 2013). Currently, there are international ef-
forts to develop wastewater guidelines for these ubiquitous
1
Department of Environment and Geography, University of emerging contaminants (Kolpin et al. 2002) since these are
Manitoba, Winnipeg, MB R3T 2N2, Canada typically unregulated (Ávila and García 2015).
2
Escuela de Química, Instituto Tecnológico de Costa Rica, Constructed wetlands (CWs) have provided a low-cost al-
Cartago 30101, Costa Rica ternative to wastewater treatment in rural and remote regions,
3
Richardson College for the Environment, The University of achieving adequate levels of treatment and reducing nutrients
Winnipeg, Winnipeg, MB R3B 2E9, Canada and other contaminant releases (Conkle et al. 2008; Li et al.
4
Department of Biosystems Engineering, University of Manitoba, 2014; Matamoros et al. 2007). CWs have been reported to
Winnipeg, MB R3T 5V6, Canada remove pharmaceuticals on four categories of efficiency:
5
Southern California Coastal Water Research Project Authority, Costa readily removed (over 70% mean removal, e.g., sulfamethox-
Mesa, CA 92626, USA azole, atenolol, metoprolol); moderately removed (50–70%
Environ Sci Pollut Res (2021) 28:52306–52318 52307

mean removal, e.g., ibuprofen, naproxen, gemfibrozil); limit- of the filter bed and potentially support biofilm growth while
edly removed (20–50% mean removal, e.g., diclofenac, increasing backwash ability through increased percolation
clofibric acid, carbamazepine) and hardly removed (less than when compared to sand (Elliott 2001; Gill et al. 2009; Gill
20% mean removal, e.g., ampicillin, erythromycin) (Li et al. et al. 2011). Increased aeration can enhance contaminant re-
2014). Current knowledge gaps exist around understanding moval in CW-based application through promotion of biodeg-
the influence of design parameters (e.g., aeration conditions, radation (Camacho-Munoz et al. 2012).
hydraulic retention time, type of substrate used) on the perfor- CRG has been tested as a wastewater filter medium within
mance of pilot- and full-scale treatment facilities for pharma- intermittent recirculating biofilters treating septic tank effluent
ceutical removal (Li et al. 2014; Verlicchi and Zambello (Elliott 2001; Healy et al. 2010; Hu and Gagnon 2006) and in
2014). polishing filters for domestic wastewater (Gill et al. 2011;
Gravel and sand are the most common CW substrates re- Healy et al. 2010; Horan and Lowe 2007). In recirculating
ported in the literature for the removal of nutrients and other biofilters, CRG has been reported to retain 79% to 98% of
contaminants in horizontal, vertical, and hybrid subsurface total suspended solids and remove over 94% and 96% of
filter beds (Li et al. 2014). Crushed rock, gravel and sand BOD and ammonium nitrogen, respectively (Elliott 2001;
mixtures, glass, anthracite, dolomite, and soil have been tested Hu and Gagnon 2006). As a tertiary filter media, Gill et al.
as well (Wu et al. 2015). Retention behavior is influenced by (2009) reported removal of 73% of chemical oxygen demand
hydraulic properties, mechanical and chemical resistance, ad- and 28% of total nitrogen by CRG. Salzmann et al. (2020)
equacy for supporting the development of plants and micro- studied CRG as a tertiary filter media for municipal lagoon
organisms, inertness, potential for release/leaching of addi- water treatment, obtaining equal capacity for removal of
tional contaminants, and ability for desorption/regeneration suspended solids as achieved with sand, and similar perfor-
(Dordio and Carvalho 2013). Important factors to determine mance for ammonia and chemical oxygen demand
the applicability of candidate substrates in CWs are cost, local parameters.
availability, saturation time (i.e., the time needed to occupy all To our knowledge, no previous studies have explored CRG
available sorption sites), and recyclability of saturated filter as a substrate for the removal of pharmaceuticals from waste-
media (Dordio and Carvalho 2013). The combination of hy- water. This study sought to evaluate CRG as a substrate in
draulic conductivity and adsorption capacity of the substrate, CW-based applications and compare its performance to tradi-
as determined by its surface area, are two factors that can drive tional materials such as gravel or sand. The studies were con-
the usefulness of a given substrate for CW applications. ducted at two scales of work (i.e., mesocosm and pilot scale)
Moreover, particle size distributions, porosity, and effective to allow investigation of CRG removal of pharmaceutical
particle size are all important factors that affect system perfor- analytes from both real and synthetic wastewaters.
mance (Kadlec et al. 2009; Vymazal 2011). Mesocosm systems offer a controllable experimental setup
Crushed recycled glass (CRG) is an abundant material in to study and quantify processes and mechanisms (e.g., phys-
many jurisdictions. In Canada, it represents approximately 3% ical, chemical, biological, biogeochemical) for assessing re-
of national domestic solid waste (Statistics Canada 2015). In moval and fate of wastewater contaminants. On the other
Manitoba, Canada, residential and non-residential sources hand, pilot-scale studies allow for observation of these pro-
produced more than 9000 tonnes of waste glass in 2016, and cesses using more realistic matrices, flow conditions, and hy-
over 8000 tonnes of glass were diverted from disposal sites for draulic retention times. We hypothesized that removal of phar-
recycling purposes (Statistics Canada 2018). The volume of maceutical contaminants can be attained by CRG, and perfor-
waste transformed into new glass containers is minimal. Most mance of CRG would not be significantly different from that
collected glass are crushed and reused as sidewalk and road of traditional substrates (i.e., gravel, sand), for both
base aggregate (Morawski et al. 2016; Reindl 2003), occa- mesocosm- and pilot-scale systems.
sionally serving as fill-in for water and sewer trench systems.
The limited use for CRG is not only an issue in Canada, but
globally. In the USA, three million tonnes of container glass Material and methods
were recycled in 2017, which represents 27% of the total of
glass waste generated for that year, while 60% was landfilled Study designs and treatment
(U.S. EPA 2018). In Europe, 30 billion glass containers were
recycled in 2017, for an average rate of 76% across 160 plants, Mesocosm-scale study
with approximately a quarter of all glass being wasted during
that year (European Container Glass Federation 2018). This study took place at the Prairie Wetland Research Facility
Considering the physical structure and properties of CRG, (PWRF) at the University of Manitoba in Winnipeg, Canada
it has been proposed as a potentially suitable substrate for (49° 48′ 35.9′′ N, 97° 07′ 33.0′′ W). Circular, flat-bottomed,
CWs. CRG is durable and inert, which can promote aeration low-density polyethylene tanks (2.7 m diameter × 0.72 m
52308 Environ Sci Pollut Res (2021) 28:52306–52318

height; 3.49 m 3 total volume, Figure S1) from ACE comparison with previous field studies (Anderson et al.
Rotomolds (Hospers, Iowa) were installed in an array of 2015, Chaves-Barquero et al. 2018). All mesocosms also re-
twelve (Figure S2). The setup allowed for individual, isolated ceived a one-time 1-L amendment of secondary wastewater
wetland mesocosms with quiescent waters having no inflows from a local wastewater treatment facility (Village of
or outflows. Mesocosms were randomly assigned to one of Dunnottar, Manitoba) to provide microorganism amend-
three treatments in triplicate (Figure S2): limestone gravel ments. Based on mesocosm volumes, this represents approx-
(LG), glass (PCRG), or unplanted glass (CRG). Limestone imately a 2000-fold dilution by volume.
gravel (3/4 inch) was sourced from a local materials vendor
(Reimer Soils, Winnipeg), while CRG (3/4 inch) was sourced Pilot-scale study
from a local recycling facility (Cascades Recovery+,
Winnipeg). Glass was sieved through a tumbling screen to This study was conducted at the Village of Dunnottar, located
remove large waste impurities (e.g., plastic fragments, gravel on the southwestern shore of Lake Winnipeg, Manitoba. The
particles, finer organic materials) originated from the village has approximately 763 permanent residents, with a
recycling process. Some of these materials passed through several-fold increase during the summer due to the presence
the screen and were consequently combined with the crushed of vacationers (Anderson et al. 2015). Holding tanks have
glass media. These materials accounted for approximately been installed in the community households, and sewage is
12% of the glass media by weight (Salzmann et al. 2020). hauled and discharged into the primary lagoon using septic
Physical characterization of the glass is reported in Table S1. trucks. The lagoon treatment system is comprised of a primary
The glass was given minimal treatment prior to use as filter lagoon, a secondary lagoon, and a secondary overflow lagoon
media, in an attempt to employ post-consumer waste glass that (Figure S3). After secondary treatment, water is polished by
could be converted into wastewater filter media at low cost to sending it through a pilot-scale tertiary subsurface filter (Fig.
facilitate system implementation. A layer of substrate (~27 1) comprised of four cells (10-m long × 1.8-m wide × 1-m
cm) was placed at the bottom of the tanks. Winnipeg tap water deep, bed volume of 18 m3). In 2016, two of the cells were
was added on June 25th, 2018, to fill the tanks (average depth packed with three-quarter inch CRG (Cascades Recovery+,
~ 40 cm). Floating debris from CRG (e.g., plastic caps, labels, Winnipeg) and two with local river bottom sand as substrates.
corks) were removed from the tanks upon filling and then Cells were separated with plywood and individually lined
periodically for 3 weeks. Mesocosms were then left unplanted with a polyethylene liner. Limestone cobble was placed on
or planted with approximately five cattails (Typha spp.) per the bottom of the array, while a layer of wood chips was
square meter (25 plants per mesocosm). Cattails were collect- spread on top of the filter beds. Wastewater was pumped from
ed from Oak Hammock Marsh (Stonewall, MB, 50° 11′ 15″ the secondary lagoon through a transfer pipe, splitting water
N, 97° 7′ 30″ W) on July 19th, 2018. Macrophytes were accli- into the four filter beds with two parallel perforated PVC pipes
mated for 26 days prior to the start of the experiment. delivering the water to the surface of each cell. Wastewater
Mesocosms were treated with synthetic wastewater and a then percolated vertically through the bed, reaching a collector
stock solution of pharmaceuticals on August 14, 2018. The pipe at the bottom. A wastewater pump (Burcam 400500,
wastewater (2 L per mesocosm) contained (per liter): 32.0 g Burke Water Systems Manufacturing, Inc.) was placed in
peptone, 19.0 g Lab Lemco powder meat extract, 3.0 g yeast the secondary lagoon to supply water to the filter system. It
extract, 3.0 g urea, 6.7 g (NH4)2SO4, 2.9 g K2HPO4, 2.3 g worked at a head of 1.5 m and a flow rate of 17,000 L/h. A
KH 2PO4, 0.27 g CaCl2·2H2O, and 0.2 g MgSO4·2H 2O timer controlled the pump, leaving it on for 5 min and off for
(Knapp and Graham 2007). Synthetic wastewater was pre- 99 min, resulting in a filter hydraulic residence time (HRT) of
pared to control loadings of nutrient and BOD in the 24 h.
mesocosms, and to minimize human health risks associated
with handling raw wastewater. The pharmaceutical additions Monitoring, sample collection, and analysis
consisted of 1 L of a stock solution of three common pharma-
ceuticals: atenolol, carbamazepine, and sulfamethoxazole. Mesocosm-scale study
Pharmaceutical initial target concentrations were between 5
and 10 μg/L, selected both for ease of quantification and to A suite of standard water quality parameters (dissolved oxy-
reflect reported levels in wastewater influents in Manitoba gen, chlorophyll-a, pH, temperature, oxidation-reduction po-
municipal wastewater treatment facilities (Anderson et al. tential, and conductivity) were measured every weekday
2013; Anderson et al. 2015; Chaves-Barquero et al. 2018) morning, and once a week in the afternoon, using a YSI
and elsewhere (Verlicchi and Zambello 2014). These pharma- 6600 V2 Sonde (Yellow Springs, OH). Measurements began
ceuticals were selected based on their frequency of detection on July 16th and continued until October 10th. Temperature
in wetland-based treatment facilities, ease of quantification, was also monitored every 30 min using HOBO (Bourne, MA)
low likelihood of acute effects to aquatic organisms, and water temperature data loggers. Depths were measured bi-
Environ Sci Pollut Res (2021) 28:52306–52318 52309

Fig. 1 Top view of the pilot-scale subsurface filter system located in Dunnottar, Manitoba, Canada (left), and cross-sectional view of an individual filter
bed (right). Diagram based on Salzmann et al. 2020

weekly in five different locations for each tank, and then av- preserved on ice and transported to the laboratory for analysis
eraged to account for evaporation and to monitor the water within 48 h.
volume. Photosynthetically active radiation (PAR) was mea-
sured at mid-day once a week. Sample processing and analysis
Grab water samples for pharmaceutical analysis were taken
pre-treatment on day −1 and post-treatment on days 0, 2, 7, 14, Water samples were stored in the dark at 4 °C and processed
28, and 56, with additional samples collected at 8 h, 24 h, and within 48 h. Samples were solid-phase extracted using 3 cc,
30 h post-treatment. Samples were integrated from four to six 60 mg HLB cartridges from Waters Corporation (Milford,
random spots over the depth of the water column of each MA). Ultrahigh performance liquid chromatography–tandem
mesocosm tank and collected into pre-ashed amber glass bot- mass spectrometry (UHPLC-MS/MS) was performed on an
tles, using a PVC integrative water sampler (6-cm diameter × Agilent 1200 UHLPC coupled to an Agilent 6410 triple-
1.6-m length; 4.5-L total capacity) retrofitted with PVC tubing quadrupole mass spectrometer, following the methods de-
(4-cm diameter × 30-cm length), based on previous models scribed on previous works (Carlson et al. 2013). Relevant
(Solomon et al. 1982). Piezometer wells were built using ½” QA/QC parameters (e.g., limits of detection and quantifica-
PVC piping and installed on each tank to monitor the concen- tion, instrumental parameters for chemical analysis) can be
tration of pharmaceuticals at two depths: 2 cm and 22 cm found in Tables S2 and S3.
below the depth of glass or gravel. Samples were collected
from the wells on weeks 0, 1, 2, and 4 using an Masterflex E/S Calculation of rate constants for fate processes
portable sampler (Cole-Parmer, IL). Cattail growth was mon-
itored by measuring the height of the longest leaf at week 0 Estimates for sorption and direct/indirect photolysis were cal-
and week 4 after the pharmaceutical spike, and then calculat- culated by determining the compound-specific rates and half-
ing growth rate. lives (Schwarzenbach et al. 2017). The equations are de-
scribed as follows:

Pilot-scale study Sedimentation

Grab water samples for pharmaceutical analysis were taken in For sedimentation, the calculated rate constant (ks*) was as
pre-ashed amber glass during the summer of 2018 on follows:
June 26th, July 24th, July 31st, August 7th, August 16th,
August 21st, September 4th, and September 11th at the input  
f sus
oc rsw K oc
of the filters and at each of the four output wells, with corre- k *s ¼ ð1−f w Þk s ¼ ks ð1Þ
sponding field blanks for each sampling date. Samples were 1 þ f sus
oc r sw K oc
52310 Environ Sci Pollut Res (2021) 28:52306–52318

where fw is the fraction of chemical dissolved, f sus oc is the literature (Table 3). Photolysis parameters (i.e., Φ) obtained
fraction of organic carbon within the suspended particles, from experiments under natural sunlight, or sunlight simulator
and rsw is the solid-to-solution phase ratio, equivalent to the with a cutoff filter below 290 nm, were favored for the calcu-
total suspended solids in the water column (Schwarzenbach lation of our estimates. Direct photolysis is dependent on the
et al. 2017). The sedimentation rate was estimated from: incoming solar radiation and the ability of the medium to
vs absorb some of this radiation.
ks ¼ ð2Þ
Z mix
Indirect photolysis
where vs is the particle settling velocity and zmix is the average
depth of the mesocosms (Schwarzenbach et al. 2017). For our Indirect photolysis was estimated by assuming that pharma-
purposes, zmix is 0.4 m, vs was estimated at 1.5 m day−1 for ceutical degradation occurred by reacting with hydroxyl rad-
particulate organic carbon (Gustafsson et al. 2000) and focsus icals •OH (kI∙OH):
was estimated at 0.4 since most of the particulate matter was 0

of plant origin (Schwarzenbach et al. 2017). No other sorption k I;OH ¼ k p; OH ½OH SS ð8Þ
mechanisms were considered for our analysis. We estimated
where k’p•OH is the second-order hydroxyl radical rate con-
Koc using a linear free energy relationship (Schwarzenbach
stant and [•OH]ss is the steady-state concentration of •OH.
et al. 2017) for alkylated and chlorinated benzenes, used
Second-order rate constants k’p•OH were obtained from the
previously by Carballa et al. (2008) as surrogate for sorption
literature and are displayed in Table 2 for each pharmaceuti-
estimations on pharmaceuticals:
cal. Steady-state concentrations of hydroxyl radicals have
log K oc ¼ 0:74 log Dow þ 0:15 ð3Þ been reported in the literature ranging between 1 × 10−15
and 1 × 10−17 M (Schwarzenbach et al. 2017). The steady-
where log Dow is the pH-dependent octanol-water distribution state of hydroxyl radical concentration was assumed to be 1 ×
coefficient: 10−16 M in natural surface waters (Mabury and Crosby 1994).
 
log Dow ¼ log Kow–log 1 þ 10ðpH−pKaÞ ð4Þ Overall dissipation rate

An overall dissipation rate can be estimated by adding up the

Direct photolysis rates calculated above:

k est ¼ k s * þ k D þ k I;OH ð9Þ

Direct photolysis rates (kD) were estimated by first determin-
ing the compound’s wavelength-dependent specific rate of
light absorption ka(λ) (Schwarzenbach et al. 2017): Sensitivity analysis for the overall dissipation rate
predictions
W ðλÞεðλÞ
k ta ðλÞ ¼ ð5Þ Oracle Crystal Ball software version 11.1.2.4.850 (Oracle
Z mix αðλÞ
Corporation, Redwood Shores, CA), was used to conduct sen-
where W(λ) is the spectral photon fluence rate (Einstein cm−2 sitivity analyses of our fate predictions based on the estimated
day−1) for latitude 50°N during the fall (OECD, 1997), ε(λ) is rate of dissipation defined in equation 9. We defined uncer-
the molar extinction coefficient of the compound (M−1 cm−1), tainty distributions for every parameter based on our assump-
and α(λ) is the attenuation coefficient of the medium. The tions, and the available information in the literature. For ks*,
total rate of light absorption (kat) was estimated by summation we defined a conservative two order of magnitude triangular
of the rates over all wavelengths (Schwarzenbach et al. 2017): distribution. For kD values, we considered horizontal distribu-
tions for the possible range of light absorption rates (kat) con-
1 W ðλÞεðλÞ sidering from no medium absorption to full medium absorp-
k ta ¼ ∑k ta ðλÞ ¼ ∑ ð6Þ
Z mix αðλÞ tion, as well as the reported uncertainty from measured quan-
tum yields in the literature (whenever reported). For kI,·OH, a
Finally, the direct photolysis rate constant (kD) was calcu- conservative triangular distribution considering a variability
lated by the following: of two orders of magnitude was used. Then, we performed a
forecast simulation (100,000 iterations) for the estimated rate
k D ¼ k ta Φ ð7Þ value (kest) for each analyte and then obtained a sensitivity
analysis information (i.e., how much each of the variables
where Φ is the quantum yield, which was retrieved from the affects the final estimation) as described in Figure S11.
Environ Sci Pollut Res (2021) 28:52306–52318 52311

Statistical analysis of synthetic wastewater and pharmaceuticals (Figure S7). A

similar behavior was observed for the evolution of dissolved
Results for pharmaceutical concentrations in both studies were oxygen in the mesocosms, with a quick decline and rebound
assessed using GraphPad Prism 6 (Graph Pad Software, La upon spike (Figure S8). Treated mesocosms showed a brief
Jolla, CA). Concentrations, half-lives, and rate constants are increase in chlorophyll-a concentration following the addition
reported as mean value ± standard deviation (SD) unless oth- of synthetic wastewater and pharmaceuticals, likely due to the
erwise stated. Differences were deemed statistically signifi- increased availability of nutrients (Figure S9). As well, glass
cant at p < 0.05. Normality was assessed via Shapiro-Wilk treatments experienced the greatest increase in chlorophyll-a
test. following the pulse exposure of synthetic effluent.
For the mesocosm study, Student’s t-tests were applied to
evaluate plant growth data, as well as general water quality pa- Dissipation of pharmaceuticals
rameters. One-way analysis of variance (ANOVA) followed by
Tukey’s test were applied to compare pharmaceutical concentra- Atenolol dissipated rapidly from the water column in the
tions and half-lives in the water column across treatments, as well tanks, while sulfamethoxazole and carbamazepine experi-
as pharmaceutical concentrations and half-lives in the water col- enced greater half-lives compared to atenolol in the water
umn, water-substrate interface, and within the substrate. column (Fig. 2 and Table 1). The half-life of atenolol averaged
For the pilot-scale study, one-way ANOVA tests followed 0.8 days in gravel treatments, 1 day in planted glass treat-
by Tukey’s post-hoc test were used to compare the concentra- ments, and 1 day in unplanted glass treatments.
tions of each pharmaceutical at the input and at each of the Carbamazepine had average half-lives, respectively, of 46
four outputs of the system. days, 58 days, and 41 days. Sulfamethoxazole had average
half-lives of 14 days, 14 days, and 15 days. Among treat-
ments, no significant differences (p > 0.05) were found for
Results the dissipation rates of atenolol and carbamazepine, while
sulfamethoxazole had a slightly shorter half-life in unplanted
Mesocosm-scale project glass, compared to the other two treatments (p < 0.05).
Dissipation of pharmaceuticals was also examined near the
Plant growth water-substrate interface (i.e., 2-cm depth) and within the sub-
strate (i.e., 22-cm depth) in all tanks (Table 2). We observed
After 4 weeks, no statistically significant differences were significant differences (p < 0.05) between the half-lives of
found for Typha spp. growth among substrates (p > 0.05; atenolol at different depths for all treatments. Briefly, a signif-
Figure S4). Plants were able to establish in both substrates icantly lower half-life (~1 day) of atenolol was observed in the
and grow roots within the bulk of the materials with senes- water surface, compared to the other depths near or within the
cence of the macrophytes commencing during week 4 substrate (2 to 3 days) pooled across treatments. No significant
(September 14th, 2018) for both substrates. During the course differences were found between atenolol half-lives at the in-
of the study, maintenance of the macrophytes was necessary terface and at the bulk of the substrates. In the case of carba-
during the course of the study, with some replanting labor mazepine and sulfamethoxazole, no statistically significant
conducted weekly due to tilting of the plants as a consequence differences were found between the observed half-lives at dif-
of windy conditions. ferent depths (p > 0.05).

Total phosphorus and water quality Pilot-scale study

Phosphorus was added to achieve an in-tank concentration of From all pharmaceuticals of interest for this study (Table S2),
between 1.5 and 2 mg/L. After 2 weeks, it dissipated from the only four were detected in the background at least once at the
water column in all treatments (Figure S5). The greatest re- input of the filter beds or at any of the outputs of the pilot-scale
moval rates were observed for gravel treatments, followed by system: atenolol, carbamazepine, metoprolol, and sulfameth-
unplanted glass, and planted glass treatments. No significant oxazole, with maximum concentrations at the input of 13 ng/
differences (p > 0.05) in phosphorus removal rates were found L, 247 ng/L, 72 ng/L, and 27 ng/L, respectively (Table 6).
between planted glass and unplanted glass treatments. Overall, carbamazepine was not effectively removed by any
Temperatures for the mesocosms declined over time as the of the substrates during the season. Concentrations at the out-
summer progressed into autumn (Figure S6). As expected, pH puts of both types of subsurface filters were not significantly
and dissolved oxygen values remained relatively consistent different from the levels detected at the input (p > 0.05).
throughout the duration of the study, except for a quick de- Sulfamethoxazole was neither removed by either sand or glass
cline and rebound in all spiked treatments, due to the addition during this study (p > 0.05). Metoprolol was generally well
52312 Environ Sci Pollut Res (2021) 28:52306–52318

Fig. 2 Dissipation over time of a carbamazepine; b atenolol, and c sulfamethoxazole in the water column of modeled mesocosm-scale wetlands at the
Prairie Wetland Research Facility using limestone gravel and either planted or unplanted crushed recycled glass as substrates

removed, when detected, by both materials over the season. (Cardinal et al. 2014; Cardinal et al. 2016) and with other
No significant differences were observed in the performance locations in Manitoba (Anderson et al. 2013; Anderson et al.
of the studied substrates with regard to pharmaceutical 2015; Carlson et al. 2013; Chaves-Barquero et al. 2018). This
removal. demonstrates that the mesocosm system was representative of
real systems (i.e., CWs) found in the province.
Our results suggest that a greater phosphorus removal
Discussion was promoted in gravel (>90%) than glass due to co-
precipitation and adsorption of calcium phosphates onto
Mesocosm-scale project the calcite surface of the substrate, as observed elsewhere
(Aulenbach and Nie 1988; Xu et al. 2014). The crushed
Water quality and phosphorus glass showed less affinity for phosphate removal during
the course of the study (~70%), which agrees with a sim-
Temperature, pH, and conductivity values in the current study ilar study where 50% removal was achieved for total phos-
agreed with previous studies conducted at the same facility phorus (Gill et al. 2011).

Table 1 Observed mean half-lives (t1/2) and dissipation rates (kobs) of pharmaceuticals in the water column of modeled wetlands at the Prairie Wetland
Research Facility using either gravel or crushed recycled glass as substrates. Values are presented as mean ± standard deviation

Compound Gravel Planted glass Unplanted glass

Half-life (day) kobs (day−1) Half-life (day) kobs (day−1) Half-life (day) kobs (day−1)

Atenolol 0.8 ± 0.1 0.9 ± 0.1 1.0 ± 0.2 0.7 ± 0.1 0.97 ± 0.02 0.71 ± 0.01
Carbamazepine 46 ± 3 0.0150 ± 0.0006 58 ± 18 0.012 ± 0.004 41 ± 4 0.017 ± 0.002
Sulfamethoxazole 14 ± 1 0.050 ± 0.004 14 ± 1 0.050 ± 0.004 15 ± 1 0.046 ± 0.004
Environ Sci Pollut Res (2021) 28:52306–52318 52313

Table 2 Observed mean half-lives (in days) of selected pharmaceuticals Facility using either gravel or crushed recycled glass as substrates.
near the water-substrate interface (depth = 2 cm) and in the water around Values are presented as mean ± standard deviation
the bulk substrate (depth = 22 cm) at the Prairie Wetland Research

Compound Gravel interface Gravel bulk Planted glass interface Planted glass bulk Unplanted glass interface Unplanted glass bulk

Atenolol 2.8 ± 0.1 3.1 ± 0.3 3.5 ± 0.7 2.9 ± 0.2 2.7 ± 0.3 2.1 ± 0.2
Carbamazepine 65 ± 17 72 ± 24 40 ± 11 74 ± 12 47 ± 2 53 ± 27
Sulfamethoxazole 12 ± 2 14 ± 3 11 ± 1 11 ± 1 15 ± 1 17 ± 1

Dissipation of pharmaceuticals through the wavelength-dependent specific rate of light ab-

sorption for latitude 50° N, the total rate of light absorption
Atenolol, carbamazepine, and sulfamethoxazole presented was estimated by summation of the rates over all wavelengths;
pseudo-first-order dissipation kinetics over the course of the and indirect photolysis assumed pharmaceutical degradation
study, from the water column to the bottom of the system. No due to hydroxyl radicals only.
significant differences were found among treatments for ob- Overall, the calculated and observed half-lives of all
served half-lives, even in the case of planted and unplanted analytes in the water column were not in agreement, with
glass. Observed half-lives of carbamazepine at the water col- observed values for all three compounds being significantly
umn were greater by approximately 6-fold compared to the different than predicted ones (Table 4). Briefly, for atenolol,
half-lives found in previous studies in the same facility using we estimated a half-life of 10 days, versus an observed half-
topsoil as substrate during summer (Cardinal et al. 2014) and life of 0.8 days; for carbamazepine, we estimated a half-life of
fall (Cardinal et al. 2016). Lam et al. (2004) observed a mean 8.7 days, versus an observed half-life of 47.4 days; and for
half-life of 82 days for carbamazepine in 12,000 L outdoor sulfamethoxazole, we estimated a half-life of 0.8 days, versus
microcosms over the summer, which approximately doubles an observed half-life of 14 days.
the observed half-lives in this study, likely due to increased Sorption was quantified by the organic carbon normalized
light attenuation in their system compared to the mesocosm partition coefficient (log Koc), which assumes neutral species
tanks used in our study. For sulfamethoxazole, we observed binding mainly to the organic carbon content of the particles.
half-lives similar to the ones reported by Cardinal et al. (2016) We used a single parameter linear free energy relationship
during the fall season and 2-fold greater compared to the ones (Schwarzenbach et al. 2017) to relate logDow to logKoc. The
reported during the summer (Cardinal et al. 2014). In the case analytes in this study are polar, and can be charged at the pH
of atenolol, our results suggest that a subsurface concentration values in the mesocosms (e.g., atenolol can be partially posi-
gradient occurred, with greater concentrations near the water- tively charged, and sulfamethoxazole negatively charged).
interface substrate and within the substrate than in those found Corresponding log Koc values were calculated to be 0.16 for
in the water column during early stages of the study. atenolol, 1.96 for carbamazepine, and 0.004 for sulfamethox-
Incomplete mixing within the water column has been reported azole within the pH range of our study. To determine the
to affect the observed concentrations (Cardinal et al. 2014; importance of sorption for dissipation of the analytes, we cal-
Kunkel and Radke 2011), particularly with the possibility of culated compound-specific sedimentation rates (ks*). We ob-
sorption to biofilms with subsequent release; however, it is not tained a sedimentation rate of 1.3 × 10− day−1 for carbamaz-
clear if this is the case for atenolol. epine, and values lower than 1.2 × 10−5 day−1 for the other
compounds, suggesting that sorption via sedimentation of par-
Pharmaceutical fate ticles could play a role for carbamazepine, as it was previously
observed by Cardinal et al. (2014 and 2016). However, we
In this study, the importance of fate processes responsibility expected photolysis to be dominant under the conditions of
for the removal of pharmaceuticals and was estimated. We this study.
used literature data from previous laboratory-based experi- Direct photolysis of carbamazepine was estimated to con-
ments to assess the processes that may affect removal efficien- tribute in 9% of its overall removal, and approximately 95%
cy of the three studied compounds. The processes considered for sulfamethoxazole (Table 4). The extent of direct photolysis
for our predictions were sedimentation through sorption to is dictated by both the quantum yield and the rate of light
particles, and direct and indirect photolysis. Our assumptions absorption, with compounds having a greater degree of con-
to conduct these fate predictions are described in the material jugation being able to absorb more light in the UV-A range
and methods section. Briefly, sorption was estimated through (Table 3). Atenolol is a compound that does not absorb radi-
particle sedimentation rates assuming most of the particulate ation between 290 and 800 nm, so no direct photolysis in the
matter being of plant origin; direct photolysis was estimated natural environment was expected (Liu and Williams 2007).
52314 Environ Sci Pollut Res (2021) 28:52306–52318

Table 3 Estimation of first-order direct photolysis rate constant (kD) quantum yield, k’p·OH = second order hydroxyl radical rate constant.
and pseudo-first order indirect photolysis rate constant due to reaction N/A, the compound does not absorb radiation in the spectral range of
with hydroxyl radicals (kI·OH) for pharmaceuticals in mesocosm water 290 nm to 800 nm
from Eq. 8 (see SI). kat = compound-specific absorption rate, Ф =

Compound Direct photolysis Indirect photolysis

kat Ф kD k’p·OH kI·OH a

(Eins mol−1 day−1) (mol Eins−1) (day−1) (M−1 s−1) (day−1)

Atenolol N/A 3.6 × 10-2 b N/A 8.0 × 109c 6.9 × 10-2

Carbamazepine 122 4.8 × 10-5 0.006 8.8 × 109 7.6 × 10-2
Sulfamethoxazole 9 0.09d 0.81 5.5 × 109 4.8 × 10-2
a
Assuming [·OH]ss = 1 × 10−16 M (Mabury and Crosby 1994); b Salgado et al. (2013); c Dong et al. (2015); d Boreen et al. (2004).

Challis et al. (2014) highlighted the importance of identifying photolysis in the mesocosm tanks, as previously observed at
uncertainties in the estimated kinetic data, which can enhance the same facility (Lu et al. 2015), or inaccurate estimations of
research transparency in terms of precision of the experimen- the amount of light available during the course of the study.
tal techniques and data variability. For instance, the incident From the Monte Carlo analysis, uncertainties associated with
light measurement can be a major source of uncertainty in second-order rate constants for reaction with hydroxyl radicals
studies addressing the fate of pharmaceuticals than can suffer (k’p·OH) proved to be important for atenolol (100%), carba-
photodegradation (Challis et al. 2014). mazepine (100%), and sulfamethoxazole (33%).
With the aim to understand and identify the greatest uncer- Sorption mechanisms, including electrostatic interactions
tainty sources in our estimations, we conducted a sensitivity between analytes and sorbents, were probably in place to
analysis of our fate predictions through a Monte Carlo (see some extent to promote a delay in dissipation. Covalent and
Figure S4). For direct photolysis, we considered the potential irreversible binding to sediments and suspended particulate
variability in sunlight absorption rates for both carbamazepine matter has previously been observed for sulfonamide-type an-
and sulfamethoxazole, as well as the uncertainty in quantum tibiotics (Carstens et al. 2013) and carbamazepine (Zhang
yields obtained from the literature, whenever it was reported. et al. 2011), where sorption and photodegradation were im-
Sunlight absorption rates variability proved to be important portant removal processes. The mineral surfaces used as sub-
for sulfamethoxazole estimations (65.8%), and quantum yield strates tend to be coated with negatively charged natural or-
uncertainty had a minor importance (2%). For carbamazepine, ganic matter (Schwarzenbach et al. 2017) and this was prob-
direct photolysis uncertainties in our estimations did not prove ably the case for our study. That said, we were unable to
to be important for the overall estimated rates, according to the quantify electrostatic sorption of the compounds by the
sensitivity analysis. modeling approaches used.
Hydroxyl radicals were expected to dominate indirect pho-
tolysis processes in this study. Their production was likely
Pilot-scale project
mediated by dissolved organic matter (Lam et al. 2005). As
mentioned previously, carbamazepine and sulfamethoxazole
The number of consistently detected background pharmaceu-
dissipated considerably slower than predicted in this study
ticals during the course of the study was less than in previous
(Table 4). This could be due to a number of processes includ-
efforts at the same facility (Anderson et al. 2015; Chaves-
ing reactions with radical species such as singlet oxygen or
Barquero et al. 2018). Though, their detected levels (atenolol:
carbonate radicals, increased light screening for direct
1–13 ng/L; carbamazepine: 5.7–247 ng/L; metoprolol: 11–73

Table 4 Comparison of the estimated photolysis rate constants (kest) Research Facility, and their associated half-lives for studied pharmaceu-
with the observed dissipation rates (kobs, averaged from values in ticals; kest was calculated from kD + kI·OH + ks*
Table 1) in the water column of mesocosm tanks at the Prairie Wetland

Compound kD kI·OH ks* kest kobs Estimated half-life (day) Observed half-life (day)
(day−1) (day−1) (day−1) (day−1) (day−1)

Atenolol N/A 6.9 × 10−2 ≤1.2 × 10−5 0.07 0.9 10 0.8

−2
Carbamazepine 0.006 7.6 × 10 1.3 × 10−4 0.08 0.02 8.7 47.4
Sulfamethoxazole 0.81 4.8 × 10−2 ≤1.2 × 10−5 0.85 0.05 0.8 14
Environ Sci Pollut Res (2021) 28:52306–52318 52315

ng/L; sulfamethoxazole 0.8–27 ng/L) were comparable to we observed slow kinetics of removal, for typically re-
those measured in other municipal wastewater facilities in calcitrant pharmaceutical contaminants such as carba-
Manitoba and elsewhere in North America (Table 5). The mazepine and sulfamethoxazole, while atenolol was dis-
ability of studied wastewater treatment system in removing sipated efficiently. These dynamics agreed with the re-
wastewater contaminants continues to be relevant and poten- sults obtained for removal of pharmaceuticals at the
tially beneficial to the community. pilot-scale study, where efficient removals were ob-
Atenolol and metoprolol were efficiently removed (< 99%) served for metoprolol, while carbamazepine and sulfa-
from the pilot-scale subsurface beds, for both glass and sand, methoxazole were not removed by the substrates at the
likely from biodegradation and sorption inside the filter beds subsurface filter. Moreover, pharmaceutical half-lives
as main drivers. Carbamazepine and sulfamethoxazole re- obtained from the mesocosm-scale study were well
movals were non-existent across the season, with no signifi- above the hydraulic retention time that pharmaceuticals
cant differences between the performance of both studied sub- experienced in the pilot-scale system. Overall, CRG
strates (p < 0.05) (Table 6). Inefficient removal of carbamaz- showed a good potential as substrate for CW applica-
epine (i.e., very limited to no removal) by these types of sys- tions, with similar performances compared to traditional
tems has been reported in previous efforts at the same facility materials such as sand or gravel. We recommend new
in Manitoba (Anderson et al. 2015, Chaves-Barquero et al. studies consider the use of CRG either by itself or in
2018), and elsewhere (Anderson et al. 2020; Carlson et al. mixtures with traditional materials to study the removal
2013; Chaves-Barquero et al. 2016; Li et al. 2014). of nutrients and other wastewater contaminants though
Carbamazepine is known for its recalcitrant behavior and CW-based technologies in the field.
low biodegradation previously reported in municipal waste-
water treatment systems (Metcalfe et al. 2003; Ngoc Han et al.
2018; Yang et al. 2017). Despite the susceptibility of sulfa-
methoxazole to undergo photodegradation in wetland systems Conclusions
(Cardinal et al. 2014), this was not the case because of the
light-limited conditions present in the subsurface system, Our results suggest that CRG can be a suitable substrate for
which was covered by wooden chips. Variations in concen- wetland-based wastewater treatment applications. Overall,
trations could be observed for all analytes depending on the CRG showed similar behavior compared to gravel in terms
sampling date, which is to be expected considering the use of of dissipation and obtained half-lives for studied pharma-
grab-sampling on punctual dates to determine the concentra- ceuticals in the mesocosm-scale systems, with photolysis
tions, and the variability in population and in wastewater dis- being a predominant removal mechanism for sulfamethox-
charge from the community due to the of touristic activities azole, based on our estimations. A similar behavior in
during the summer as noted in previous studies (Anderson terms of interaction with pharmaceuticals was also ob-
et al. 2015, Chaves-Barquero et al. 2018). served for CRG and sand in the pilot-scale system. The
Finally, a comparison between the results obtained in observed performance of CRG suggests its potential to be
the pilot scale to the results reported in the mesocosm- used by itself or mixed with other traditional materials in
scale study is presented. These systems were different in substrates for CW application, especially in post-lagoon
design: the mesocosms were configured as quiescent treatment systems such as the pilot-scale system studied
modeled wetlands with no inflows or outflows, treating on this manuscript. Lack of removal for carbamazepine
spiked levels of contaminants, while the pilot-scale sys- and sulfamethoxazole in the pilot-scale system indicates a
tem was a vertical-flow subsurface filter treating second- need for future studies that aim to optimize the operational
ary municipal wastewater with background levels of the conditions (e.g., substrate particle size, hydraulic retention
contaminants of interest. In the mesocosm-scale project, time, flow rate) for achieving greater removal rates and

Table 5 Ranges of target concentrations (in ng L−1) detected in grab-water samples collected in the current pilot-scale study and in previous studies of
rural wastewater lagoon systems in Manitoba

Compound Dunnottara Grand Maraisb Winkler/Mordenc Norway housed Misipawistike Current study

Atenolol 4–856 < LOD < 2–26 < LOD 44–2184 1–13
Carbamazepine 105–426 59–500 4–48 247–2770 6–150 6–247
Metoprolol 10–1122 < LOD < LOD 1210–1230 30-600 11–73
Sulfamethoxazole 4–1253 10–21 6–48 186–7180 185–3701 0.8–27
a
Anderson et al. (2015); b Anderson et al. (2013); c Carlson et al. (2013); d Challis et al. (2018); e Anderson et al. (2020)
52316 Environ Sci Pollut Res (2021) 28:52306–52318

Table 6 Mean concentrations (ng L−1, ± standard deviation, n = 3) of Ackerman from the Biosystems Engineering Department of the
consistently detected target compounds before and after the tertiary pilot- University of Manitoba for their contribution on setup and maintenance
scale filter at the Dunnottar wastewater treatment facility during the sum- of the pilot-scale filter, as well as their help through the field sampling
mer of 2018. Subsurface filter cells were made of recycled glass or sand. season; Danielle Desrochers, Carlie Lau, Alistair Brown, Kevin Stroski,
LOD and LOQ (ng L−1) are indicated below each analyte’s name. ND, Jonathan Challis, and Billy Hong for their help with setting up the
non-detectable mesocosm tanks. Thank you to the anonymous reviewers for taking time
to review our manuscript and provide helpful comments that contributed
s Sampling Site to the improvement of this manuscript.

Compound Date Input Glass Sand Authors’ contributions LGCB and MLH designed the studies. LGCB
and BWH set up, maintained, and took samples on the mesocosm-scale
Atenolol Jun 26 13 ± 2 2.5 ± 0.6 1.2 ± 0.6 experiment, with support from MLH. LGCB did all the sampling at the
LOD: 0.29 Jul 24 4±2 3±1 4±2 pilot-scale system. LGCB performed all the pharmaceutical analyses with
LOQ: 0.96 support from KHL and CSW. NC supported the pilot-scale system design
Jul 31 <LOQ 2.0 ± 0.3 ND
and construction. LGCB wrote the manuscript with inputs from all
Aug 7 3±1 1.2 ± 0.2 ND
authors.
Aug 16 4±2 1.2 ± 0.2 ND
Aug 21 4±1 ND ND Funding Support for this project was provided by Manitoba Liquor &
Sep 4 4±1 1.4 ± 0.3 ND Lotteries, the Natural Sciences and Engineering Council of Canada, the
Sep 11 6±2 1.2 ± 0.2 2±1 Canada Research Chairs Program, the University of Manitoba, and
Vicerrectoría de Investigación y Extensión del Instituto Tecnológico de
Carbamazepine Jun 26 120 ± 5 120 ± 3 90 ± 10 Costa Rica.
LOD: 1.7 Jul 24 247 ± 31 52 ± 1 128 ± 6
LOQ: 5.7
Jul 31 115 ± 12 125 ± 8 99 ± 13 Data availability All data generated or analyzed during this study are
Aug 7 99 ± 4 107 ± 6 101 ± 4 included in this published article, and its supplementary information file.
Aug 16 88 ± 9 66 ± 2 78 ± 13
Aug 21 70 ± 6 59 ± 15 73 ± 18 Declarations
Sep 4 ND ND ND
Competing interests The authors declare that they have no competing
Sep 11 ND ND ND
interests.
Metoprolol Jun 26 24 ± 2 22 ± 1 26 ± 1
LOD: 3.4 Jul 24 72 ± 5 ND ND
LOQ: 11
Jul 31 19 ± 2 11 ± 2 ND
References
Aug 7 30 ± 1 <LOQ ND
Aug 16 <LOQ ND ND Anderson JC, Carlson JC, Low JE, Challis JK, Wong CS, Knapp CW,
Aug 21 <LOQ ND ND Hanson ML (2013) Performance of a constructed wetland in Grand
Sep 4 ND ND ND Marais, Manitoba, Canada: removal of nutrients, pharmaceuticals,
and antibiotic resistance genes from municipal wastewater. Chem
Sep 11 21 ± 4 <LOQ ND Central J 7:15. https://doi.org/10.1186/1752-153x-7-54
Sulfamethoxazole Jun 26 24 ± 5 22 ± 1 26 ± 1 Anderson JC, Joudan S, Shoichet E, Cuscito LD, Alipio AEC, Donaldson
LOD: 0.25 Jul 24 15 ± 2 12 ± 3 27 ± 4 CS, Khan S, Goltz DM, Rudy MD, Frank RA, Knapp CW, Hanson
LOQ: 0.83 ML, Wong CS (2015) Reducing nutrients, organic micropollutants,
Jul 31 ND ND ND
antibiotic resistance, and toxicity in rural wastewater effluent with
Aug 7 ND ND ND subsurface filtration treatment technology. Ecol Eng 84:375–385.
Aug 16 13 ± 3 9±4 14 ± 3 https://doi.org/10.1016/j.ecoleng.2015.08.005
Aug 21 23 ± 3 23 ± 3 21 ± 1 Anderson JC, Jabari P, Parajas A, Loeb E, Luong KH, Vahedi A, Wong
Sep 4 9±1 12 ± 2 9±1 CS (2020) Evaluation of cold-weather wastewater nitrification tech-
nology for removal of polar chemicals of emerging concern from
Sep 11 19 ± 4 17 ± 2 16 ± 3 rural Manitoba wastewaters. Chemosphere 253:11. https://doi.org/
10.1016/j.chemosphere.2020.126711
Aulenbach DB, Nie MS (1988) Studies on the mechanism of phosphorus
better understanding removal mechanisms in the laborato- removal from trated wastewater by sand. J Water Pollut Control Fed
ry as well as in the field. 60:2089–2094
Ávila C, García J (2015) Pharmaceuticals and Personal Care Products
(PPCPs) in the environment and their removal from wastewater
through constructed Wetlands. Compr Anal Chem 67:195–244.
https://doi.org/10.1016/B978-0-444-63299-9.00006-5
Supplementary Information The online version contains supplementary Boreen AL, Arnold WA, McNeill K (2004) Photochemical fate of sulfa
material available at https://doi.org/10.1007/s11356-021-14483-4. drugs in the aquatic environment: Sulfa drugs containing five-
membered heterocyclic groups. Environ Sci Technol 38:3933–
Acknowledgements We thank Mayor Rick Gamble and Mitch 3940. https://doi.org/10.1021/es0353053
McPherson at the Village of Dunnottar for granting access to their waste- Camacho-Munoz D, Martin J, Santos JL, Aparicio I, Alonso E (2012)
water treatment facility; Rena Salzmann, Daniel Flores, and Joe Effectiveness of conventional and low-cost wastewater treatments in
Environ Sci Pollut Res (2021) 28:52306–52318 52317

the removal of pharmaceutically active compounds. Water Air Soil European Container Glass Federation (2018): Record collection of glass
Pollut 223:2611–2621. https://doi.org/10.1007/s11270-011-1053-9 containers hits 76% in the EU. URL: https://feve.org/record-
Carballa M, Fink G, Omil F, Lema JM, Ternes T (2008) Determination of collection-of-glass-containers-for-recycling-hits-76-in-the-eu/,
the solid-water distribution coefficient (K-d) for pharmaceuticals, Accessed June 15, 2020.
estrogens and musk fragrances in digested sludge. Water Res 42: Gill L, Doran C, Misstear D, Sheahan B (2009) The use of recycled glass
287–295. https://doi.org/10.1016/j.watres.2007.07.012 as a filter media for on-site wastewater treatment. Desalin Water
Cardinal P, Anderson JC, Carlson JC, Low JE, Challis JK, Beattie SA, Treat 4:198–205. https://doi.org/10.5004/dwt.2009.376
Bartel CN, Elliott AD, Montero OF, Lokesh S, Favreau A, Kozlova Gill LW, Veale PL, Murray M (2011): Recycled glass compared to sand
TA, Knapp CW, Hanson ML, Wong CS (2014) Macrophytes may as a media in polishing filters for on-site wastewater treatment.
not contribute significantly to removal of nutrients, pharmaceuticals, Water Practice and Technology 6(3):1–2. https://doi.org/10.2166/
and antibiotic resistance in model surface constructed wetlands. Sci wpt.2011.058
Total Environ 482:294–304. https://doi.org/10.1016/j.scitotenv. Gustafsson O, Duker A, Larsson J, Ingri J (2000) Functional separation of
2014.02.095 colloids and gravitoids in surface waters based on differential set-
Cardinal P, Anderson JC, Carlson JC, Low JE, Challis JK, Wong CS, tling velocity: coupled cross-flow filtrations-split flow thin-cell frac-
Hanson ML (2016) Late season pharmaceutical fate in wetland tionation (CFF-SPLITT). Limnol Oceanogr 45:1731–1742. https://
mesocosms with and without phosphorous addition. Environ Sci doi.org/10.4319/lo.2000.45.8.1731
Pollut Res 23:22678–22690. https://doi.org/10.1007/s11356-016- Healy MG, Burke P, Rodgers M (2010) The use of laboratory sand, soil
7433-3 and crushed-glass filter columns for polishing domestic-strength
Carlson JC, Anderson JC, Low JE, Cardinal P, MacKenzie SD, Beattie synthetic wastewater that has undergone secondary treatment.
SA, Challis JK, Bennett RJ, Meronek SS, Wilks RPA, Buhay WM, Journal of Environ Sci Health Part a-Toxic/Hazard Substances
Wong CS, Hanson ML (2013) Presence and hazards of nutrients and Environmental Engineering 45:1635–1641. https://doi.org/10.
emerging organic micropollutants from sewage lagoon discharges 1080/10934529.2010.506130
into Dead Horse Creek, Manitoba, Canada. Sci Total Environ 445: Horan NJ, Lowe M (2007) Full-scale trials of recycled glass as tertiary
64–78. https://doi.org/10.1016/j.scitotenv.2012.11.100 filter medium for wastewater treatment. Water Res 41:253–259.
Carstens KL, Gross AD, Moorman TB, Coats JR (2013) Sorption and https://doi.org/10.1016/j.watres.2006.08.028
photodegradation processes govern distribution and fate of sulfa- Hu ZF, Gagnon GA (2006) Impact of filter media on the performance of
methazine in freshwater-sediment microcosms. Environ Sci full-scale recirculating biofilters for treating multi-residential waste-
Technol 47:10877–10883. https://doi.org/10.1016/10.1021/ water. Water Res 40:1474–1480. https://doi.org/10.1016/j.watres.
es402100g 2006.01.041
Challis JK, Hanson ML, Friesen KJ, Wong CS (2014) A critical assess- Kadlec RH, Wallace S, ebrary Inc (2009) Treatment wetlands. CRC
ment of the photodegradation of pharmaceuticals in aquatic environ- Press, Boca Raton
ments: defining our current understanding and identifying knowl- Knapp CW, Graham DW (2007) Nitrite-oxidizing bacteria guild ecology
edge gaps. Environ Sci-Processes Impacts 16:672–696. https://doi. associated with nitrification failure in a continuous-flow reactor.
org/10.1039/c3em00615h FEMS Microbiol Ecol 62:195–201. https://doi.org/10.1111/j.1574-
Challis JK, Stroski KM, Loung KH, Hanson ML, Wong CS (2018) Field 6941.2007.00380.x
evaluation and in situ stress testing of the organic-diffusive gradients Kolpin DW, Furlong ET, Meyer MT, Thurman EM, Zaugg SD, Barber
in thin-films passive sampler. Environ Sci Technol 52:12573– LB, Buxton HT (2002) Pharmaceuticals, hormones, and other or-
12582. https://doi.org/10.1021/acs.est.8b03622 ganic wastewater contaminants in US streams, 1999-2000: a nation-
Chaves-Barquero LG, Luong KH, Mundy CJ, Knapp CW, Hanson ML, al reconnaissance. Environ Sci Technol 36:1202–1211. https://doi.
Wong CS (2016) The release of wastewater contaminants in the org/10.1021/es011055j
Arctic: a case study from Cambridge Bay, Nunavut, Canada. Kunkel U, Radke M (2011) Reactive tracer test to evaluate the fate of
Environ Pollut 218:542–550. https://doi.org/10.1016/j.envpol. pharmaceuticals in rivers. Environ Sci Technol 45:6296–6302.
2016.07.036 https://doi.org/10.1021/es104320n
Chaves-Barquero LG, Luong KH, Rudy MD, Frank RA, Hanson ML, Lam MW, Young CJ, Brain RA, Johnson DJ, Hanson MA, Wilson CJ,
Wong CS (2018) Attenuation of pharmaceuticals, nutrients and tox- Richards SM, Solomon KR, Mabury SA (2004) Aquatic persistence
icity in a rural sewage lagoon system integrated with a subsurface of eight pharmaceuticals in a microcosm study Environ Toxicol
filtration technology. Chemosphere 209:767–775. https://doi.org/ Chem 23:1431–1440
10.1016/j.chemosphere.2018.06.119 Lam MW, Young CJ, Mabury SA (2005) Aqueous photochemical reac-
Cizmas L, Sharma VK, Gray CM, McDonald TJ (2015) Pharmaceuticals tion kinetics and transformations of fluoxetine. Environ Sci Technol
and personal care products in waters: occurrence, toxicity, and risk. 39:513–522. https://doi.org/10.1021/es0494757
Environ Chem Lett 13:381–394. https://doi.org/10.1007/s10311- Li Y, Zhu G, Ng WJ, Tan SK (2014) A review on removing pharmaceu-
015-0524-4 tical contaminants from wastewater by constructed wetlands: de-
Conkle JL, White JR, Metcalfe CD (2008) Reduction of pharmaceutically sign, performance and mechanism. Sci Total Environ 468:908–
active compounds by a lagoon wetland wastewater treatment system 932. https://doi.org/10.1016/j.scitotenv.2013.09.018
in Southeast Louisiana. Chemosphere 73:1741–1748. https://doi. Li JA, Cheng WX, Xu LK, Strong PJ, Chen H (2015) Antibiotic-resistant
org/10.1016/j.chemosphere.2008.09.020 genes and antibiotic-resistant bacteria in the effluent of urban resi-
Dong MM, Trenholm R, Rosario-Ortiz FL (2015) Photochemical degra- dential areas, hospitals, and a municipal wastewater treatment plant
dation of atenolol, carbamazepine, meprobamate, phenytoin and system. Environ Sci Pollut Res 22:4587–4596. https://doi.org/10.
primidone in wastewater effluents. J Hazard Mater 282:216–223. 1007/s11356-014-3665-2
https://doi.org/10.1016/j.jhazmat.2014.04.028 Liu QT, Williams HE (2007) Kinetics and degradation products for direct
Dordio AV, Carvalho AJP (2013) Organic xenobiotics removal in con- photolysis of beta-blockers in water. Environ Sci Technol 41:803–
structed wetlands, with emphasis on the importance of the support 810. https://doi.org/10.1021/es0616130
matrix. J Hazard Mater 252:272–292. https://doi.org/10.1016/j. Lu Z, Challis JK, Wong CS (2015) Quantum yields for direct photolysis
jhazmat.2013.03.008 of neonicotinoid insecticides in water: implications for exposure to
Elliott RW (2001) Evaluation of the use of crushed recycled glass as a nontarget aquatic organisms. Environ SciTechnol Lett 2:188–192.
filter medium: Part 2. Water-Eng Manag 148:17–20 https://doi.org/10.1021/acs.estlett.5b00136
52318 Environ Sci Pollut Res (2021) 28:52306–52318

Mabury SA, Crosby DG (1994): The relationship of hydroxyl reactivity Statistics Canada (2018) Disposal of waste, by source (Table 38-10-0032-
to pesticide persistence. Aquat Surf Photochem 149–161. 01). Environmentl Accounts and Statistics Division, System of
Matamoros V, Arias C, Brix H, Bayona JM (2007) Removal of pharma- National Accounts, Ottawa URL: https://www150.statcan.gc.ca/t1/
ceuticals and personal care products (PPCPs) from urban wastewater tbl1/en/tv.action?pid=3810003201. Accessed July 25, 2020
in a pilot vertical flow constructed wetland and a sand filter. Environ U.S. EPA (2018): Facts and figures about materials, waste and recycling.
Sci Technol 41:8171–8177. https://doi.org/10.1021/es071594+ URL: https://www.epa.gov/facts-and-figures-about-materials-
Metcalfe CD, Koenig BG, Bennie DT, Servos M, Ternes TA, Hirsch R waste-and-recycling/glass-material-specific-data, Accessed August
(2003) Occurrence of neutral and acidic drugs in the effluents of 2, 2020.
Canadian sewage treatment plants. Environ Toxicol Chem 22: Uslu MO, Jasim S, Arvai A, Bewtra J, Biswas N (2013) A survey of
2872–2880. https://doi.org/10.1897/02-469 occurrence and risk assessment of pharmaceutical substances in the
Morawski C, Wilcox J, Millette S (2016): Who pays that — an analysis of Great Lakes Basin. Ozone Sci Eng 35:249–262. https://doi.org/10.
beverage container collection and costs in Canada. CM Consulting, 1080/01919512.2013.793595
Barcelona, URL: https://www.cmconsultinginc.com/wpcontent/ Verlicchi P, Zambello E (2014) How efficient are constructed wetlands in
uploads/2016/12/WPW2016-FINAL-with-cover.pdf. Accessed removing pharmaceuticals from untreated and treated urban waste-
July 7, 2020. waters? A review. Sci Total Environ 470:1281–1306. https://doi.
Ngoc Han T, Reinhard M, Gin KY-H (2018) Occurrence and fate of org/10.1016/j.scitotenv.2013.10.085
emerging contaminants in municipal wastewater treatment plants Vidal-Dorsch DE, Bay SM, Maruya K, Snyder SA, Trenholm RA,
from different geographical regions-a review. Water Res 133:182– Vanderford BJ (2012) Contaminants of emerging concern in munic-
207. https://doi.org/10.1016/j.watres.2017.12.029 ipal wastewater effluents and marine receiving water. Environ
Organisation for Economic Co-operation And Development (OECD) Toxicol Chem 31:2674–2682. https://doi.org/10.1002/etc.2004
(1997) Guidance document on direct phototransformation of
Vymazal J (2011) Constructed wetlands for wastewater treatment: five
chemicals in water. OECD Series on Testing and Assessment.
decades of experience. Environ Sci Technol 45:61–69. https://doi.
OECD Publishing
org/10.1021/es101403q
Oulton RL, Kohn T, Cwiertny DM (2010) Pharmaceuticals and personal
care products in effluent matrices: a survey of transformation and Wu JM, Xu D, He F, He J, Wu ZB (2015) Comprehensive evaluation of
removal during wastewater treatment and implications for wastewa- substrates in vertical-flow constructed wetlands for domestic waste-
ter management. J Environ Monit 12:1956–1978. https://doi.org/10. water treatment. Water PractTechnol 10:625–632. https://doi.org/
1039/c0em00068j 10.2166/wpt.2015.077
Reindl J (2003) Reuse/recycling of glass cullet for non-container uses. Xu N, Chen M, Zhou KR, Wang YL, Yin HW, Chen ZG (2014)
Dane County Department of Public Works, Madison URL: https:// Retention of phosphorus on calcite and dolomite: speciation and
archive.epa.gov/wastes/conserve/tools/greenscapes/web/pdf/glass. modeling. RSC Adv 4:35205–35214. https://doi.org/10.1039/
pdf. Accessed July 23, 2020 c4ra05461j
Salgado R, Pereira VJ, Carvalho G, Soeiro T, Gaffney V, Almeida C, Yang Y, Ok YS, Kim K-H, Kwon EE, Tsang YF (2017) Occurrences and
Cardoso VV, Ferreira E, Benoliel MJ, Ternes TA, Oehmen A, Reis removal of pharmaceuticals and personal care products (PPCPs) in
MAM, Noronha JP (2013) Photodegradation kinetics and transfor- drinking water and water/sewage treatment plants: a review. Sci
mation products of ketoprofen, diclofenac and atenolol in pure water Total Environ 596:303–320. https://doi.org/10.1016/j.scitotenv.
and treated wastewater. J Hazard Mater 244:516–527. https://doi. 2017.04.102
org/10.1016/j.jhazmat.2012.10.039 Zhang DQ, Tan SK, Gersberg RM, Sadreddini S, Zhu JF, Nguyen AT
Salzmann RD, Ackerman JN, Cicek N (2020): Pilot-scale, on-site inves- (2011) Removal of pharmaceutical compounds in tropical construct-
tigation of crushed recycled glass as tertiary filter media for munic- ed wetlands. Ecol Eng 37:460–464. https://doi.org/10.1016/j.
ipal lagoon wastewater treatment. Environ Technol 9. doi:https:// ecoleng.2010.11.002
doi.org/10.1080/09593330.2020.1775711 Zhu SC, Chen H, Li JN (2013) Sources, distribution and potential risks of
Schwarzenbach R, Gschwend P, Imboden D (2017) Environmental or- pharmaceuticals and personal care products in Qingshan Lake basin,
ganic chemistry. John Wiley & Sons, Inc., New Jersey Eastern China. Ecotoxicol Environ Saf 96:154–159. https://doi.org/
Solomon KR, Smith K, Stephenson GL (1982) Depth integrating sam- 10.1016/j.ecoenv.2013.06.033
plers for use in limnocorrals. Hydrobiologia 94:71–75
Statistics Canada (2015): Human activity and the environment., Ottawa, Publisher’s note Springer Nature remains neutral with regard to jurisdic-
ON. URL: https://www150.statcan.gc.ca/n1/en/pub/16-201-x/16- tional claims in published maps and institutional affiliations.
201-x2005000-eng.pdf?st = LvPhnqMU. Accessed July 26, 2020.