1 s2.0 S0304389422001340 Main

Journal of Hazardous Materials 429 (2022) 128346
Contents lists available at ScienceDirect
Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat
Improved drinking water quality after adding advanced oxidation for

organic micropollutant removal to pretreatment of river water undergoing
dune infiltration near The Hague, Netherlands
Peer H.A. Timmers a, *, T. Slootweg b, A. Knezev b, M. van der Schans a, L. Zandvliet b, A. Reus a,
D. Vughs a, L. Heijnen a, T. Knol c, J. El Majjaoui c, P. van der Wielen a, d, P.J. Stuyfzand a, e,
K. Lekkerkerker-Teunissen c
a
KWR Water Research Institute, Groningenhaven 7, 3433 PE Nieuwegein, the Netherlands
b
The Water Laboratory N.V., J.W. Lucasweg 2, 2031 BE Haarlem, the Netherlands
c
Dunea, Utility for drinking water and nature conservancy, Plein van de Verenigde Naties 11-15, 2719 EG Zoetermeer, the Netherlands
d
Laboratory of Microbiology, Wageningen University, the Netherlands
e
Stuyfzand Hydroconsult+, Brederodestraat 138, 2042BL Zandvoort, the Netherlands
H I G H L I G H T S G R A P H I C A L A B S T R A C T
• AOP decreases the number and concen

tration of OMP and chances of negative
impact on ecology and groundwater
quality.
• AOP decreased OMP concentrations in
drinking water with no measurable
negative effect on water quality
parameters.
• MARR produces water with a highly
stable chemical and microbiological
composition and levels out seasonal
peak fluctuations.
• There is redundancy in OMP removal by
MARR and AOP, but AOP is able to
remove certain OMP that MARR is not,
and vice versa.
A R T I C L E I N F O
Keywords:
Organic micropollutants (OMP)
Advanced oxidation process (AOP)
Managed aquifer recharge and recovery
(MARR)
Dune infiltration
Drinking water production
* Corresponding author.
E-mail address: [email protected] (P.H.A. Timmers).
https://doi.org/10.1016/j.jhazmat.2022.128346
Received 14 December 2021; Received in revised form 20 January 2022; Accepted 21 January 2022
Available online 26 January 2022
0304-3894/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
P.H.A. Timmers et al. Journal of Hazardous Materials 429 (2022) 128346
1. Introduction combination of AOP followed by MARR could have several advantages

for OMP removal, such as 1) lowering the concentration and number of
The presence and concentration of organic micropollutants (OMP) in OMP before MARR (resulting in low concentrations in drinking water);
surface water sources for drinking water production form a potential 2) biological removal of transformation products and by-products of
threat for water quality today and in future. OMP, such as pesticides, AOP during MARR; 3) stimulation of the biological activity during
pharmaceutically active compounds (PhACs), endocrine disrupting MARR by the higher concentrations of AOC as a consequence of AOP;
compounds, X-ray contrast media and personal care products, have been and thereby 4) lowering the AOC concentrations in treated water. Pre
found at ng/L to low µg/L concentrations in surface waters throughout vious studies have indeed indicated increased levels of AOC due to
the world (Kolpin et al., 2002; Jürgens et al., 2002; Stolker et al., 2004; ozonation that stimulated bromate degradation in nitrate- and iron
Kasprzyk-Hordern et al., 2008; Gros et al., 2009; Houtman, 2010). reducing zones during MARR (Wang et al., 2018b; 2018c). Another
PhACs and pesticides are among OMP of concern for drinking water reason to apply AOP before MARR is that AOP is a physical-chemical
utilities (Ray et al., 2002) because of their biological activity with process with a short residence time, whereas MARR is a natural pro
possible long term effects, and sensitivity in the public media. The effect cess with a long residence time that also levels off peak concentrations
on human health is considered negligible for some OMP at low con that occur when surface water is used for drinking water production.
centrations (Schriks et al., 2010), but OMP should be avoided in In this study, a full scale combination of partial AOP pretreatment
drinking water since the toxicity of other OMP and a mixture of them is (30–40%) followed by MARR was tested at one of the drinking water
unknown. In the Netherlands, legal standards for OMP in pre-treated production locations of Dunea. The experimental serial AOP was added
river water for managed aquifer recharge and recovery (MARR) for in 2018 to treat a partial flow of water that was pretreated with rapid
drinking water production are set by the twelve provincial governments sand filtration. The serial AOP consisted of H2O2/O3 and H2O2/UV. This
(IB 1993). These standards are, for most compounds, similar to national partial AOP-treated water (30–40%) was mixed with non-AOP treated
limits set for drinking water, with a maximum allowed concentration for water (60–70%) before recharging the dune catchment area (see Fig. 1).
individual pesticides of 0.1 µg/L, and a total concentration of 0.5 µg/L. Water quality parameters were monitored during different stages of the
This could mean that when OMP concentrations increase in surface treatment process, especially aiming at a comparison between a period
waters in the future, extensive pre-treatment of surface water prior to without AOP (2016–2017) and with AOP (2019–2020). Water quality
MARR needs to be implemented. differences between both periods were evaluated for geochemistry and
Advanced oxidation processes (AOP) as pre-treatment have proven (in)organic chemistry parameters, OMP presence and concentration,
to be robust and efficient for conversion of a wide range of OMP. Pilot transformation products, biological stability and activity, toxicological
studies with a serial AOP of hydrogen peroxide, ozone and UV on water effects and effects on the microbial community composition.
pretreated with rapid sand filtration (RSF) showed a high conversion of
a set of OMP (Lekkerkerker-Teunissen et al., 2012). A drawback of AOP, 2. Materials and methods
however, may be the formation of by-products (i.e. nitrite and bromate)
and transformation products, when complete mineralization is not 2.1. Site description
achieved. These compounds could compromise subsequent treatment
processes or the drinking water quality. Furthermore, AOP has shown to The sampled locations of the treatment processes used for drinking
convert organic matter to smaller, oxidized and more biodegradable, water production are shown in Fig. 1. The RSF effluent was transported
assimilable organic carbon (AOC) (Huang et al., 2020; van der Kooij as influent for dune infiltration areas INF-M and INF-B via transport
et al., 1989, 2015). AOC can be formed as a result of direct photolysis mains BAL1 and BAL2, respectively. AOP treatment was only performed
(UV dose ≥100 mJ/cm2) of dissolved organic carbon (DOC) or through a on a fraction of the RSF effluent (30–40%) and was mixed with non-AOP
reaction of DOC with hydroxyl radicals (Ijpelaar et al., 2010; Richardson treated RSF effluent before infiltration in dune catchment area
et al., 1999; Huang et al., 2005). The organic by-products comprising Berkheide.
AOC have been linked to increased bacterial regrowth in drinking water The pre-treated dune influent (INF) was transported to dune infil
distribution systems (Kooij, 1992; LeChevallier et al., 1992). Biological tration area Berkheide north of The Hague. About 25 Mm3/a of Meuse
filtration is used to remove these by-products, creating biologically River water was infiltrated there via 30 infiltration ponds. In this
stable water prior to distribution (van der Kooij et al., 1989, 2015; Huck research, we have focused on one site of Berkheide within the dune
et al., 1991; Krasner et al., 1993). Drinking water company Dunea (The infiltration of Katwijk (Fig. 2), where the water entered infiltration
Hague, The Netherlands) uses managed aquifer recharge and recovery ponds IP-37 and IP-38 via transport mains BAL2. From here, this water
(MARR) in the coastal dune area as one of the biological filtration steps was flowing northward, while infiltrating, up to IP- 25 and IP-40, which
for surface water treatment. The distribution of drinking water with a were shut off from northern supplies. Infiltrated water was abstracted
low AOC concentration in drinking water is important in the via drains and abstraction wells (Fig. 2). All abstracted, raw water
Netherlands and several other countries (e.g. Denmark, parts of Ger (RAW) was collectively mixed near the post-treatment (Fig. 1). In the
many, Switzerland and Belgium), because water is distributed without study area, 4 shallow monitor wells were placed (MW’s, Fig. 2, red dots),
post-disinfection in these countries. Problematic bacterial regrowth in with screen depth at 1–2 m below the pond bottom. These monitor wells
drinking water distribution systems in these areas are prevented by the were equipped with loggers that measured pressure (water level), tem
limiting amount of nutrients in the drinking water and pipe materials. perature and electrical conductivity.
The current AOC levels in drinking water produced by Dunea (3.5 – 7.5
µg/L) are below the Dutch guideline value of 10 µg C/L, which should be 2.2. AOP pilot installation
maintained in the future.
MARR processes are robust and cost-effective for obtaining a safe and The applied advanced oxidation process (AOP) was a pilot scale
stable water supply, and they include a wide variety of systems for installation developed by Dunea. It consisted of subsequently H2O2
different applications (Dillon et al., 2018). MARR is an engineered addition (6 mg/l), ozonation (1.5 mg/l), and UV (0.13 kWh/m3) of
process in which surface or storm water is infiltrated into the aquifer pretreated water with a flow of 2.2 m3/h. Directly after AOP treatment,
through dug wells, ponds (basins), injection wells, etc., to augment the water was treated by activated carbon filtration (Fig. 1). It was
groundwater and is subsequently abstracted by recovery wells (Ray shown that residual concentrations > 0.25 mg/L H2O2 already affected
et al., 2002; Bouwer, 2002). Due to the physical, chemical and biological the microbial population and especially inhibited growth of anaerobic
processes involved, MARR acts as a purification step in water treatment bacteria (Wang et al., 2017). Therefore, active carbon filters were
processes (Massmann et al., 2006) Considering all the above, a operated to remove H2O2 to a maximum concentration of 0.25 mg/L in
2
Fig. 1. Schematic of the water purification process by

Meuse drinking water company Dunea with subsequent sampling
points: Intake water from Enclosed Meuse (INT), Rapid
AOP sand filtration effluent (RSF), AOP effluent (AOP), influent
Enclosed Meuse for dune infiltration area Meijendel (sampled at end of
(FeSO4 dosing) transport pipeline); period without AOP (INF-M), Influent
H2O2 O3 UV1 UV2 AC
for dune infiltration area Berkheide (sampled at end of
transport pipeline); period with AOP (INF-B), Dune infil
tration with Infiltration ponds (IP) and monitoring wells
Microsieves
(MW), collected abstracted water (RAW), and drinking
water before distribution (DW). The AOP installation
consisted of peroxide (6 mg/l H2O2), ozone (1.5 mg/l O3),
RSF two subsequent UV reactors (0.13 kWh/m3) and activated
carbon (AC).
AOP BAL2 BAL1

60-70% INF-M
30-40%
INF-B
Dune inﬁltra�on
MW-37 IP-37 IP-38 MW-38
MW-25 IP-25 IP-40 MW-40
RAW
Post-
treament
DW
the effluent. The effluent after activated carbon filtration was mixed Supplementary Table S1. During the reference period, it was chosen to
with non-AOP treated RSF effluent, which lowered the H2O2 concen sample the end of the transport pipeline to dune infiltration site Mei
tration even more. The amount of AOP treated water that was mixed jendel as it was not known yet where the AOP pilot installation would be
during transport (BAL2) varied during this research within 30–40% of placed. It was not expected that there would be a difference in the
the total infiltrated water (see Supplementary Table S1 for specific transportation to Meijendel (INF-M) and Berkheide (INF-B), since both
mixing at sampling points). sampling locations contained the same pre-treated water without AOP in
the reference period. Sampling of the monitor wells was performed ac
2.3. Resistance calculations of transport pipeline cording to the standardized norm NEN 5477/A1 (NEN, 2013). Samples
were transported and stored at 5 ± 3 ◦ C until processing. Some analyt
Resistance in the transport mains was monitored indirectly using ical procedures required sample preservation or pre-treatment (for more
relative resistance calculations based on the pressure (P), flow (Q) and details see Supplemental Table S2).
dynamic viscosity (ŋ in Pa/s). The relative resistance (¥) was calculated
using: 2.5. Travel times
¥ = P/Q*ŋ0.237
The travel time of Meuse River water was determined for 9 moni
toring stations in the dune infiltration system: the 4 infiltration ponds
(IP25, IP37, IP38 and IP40) at some distance from the dune inlet (INF),
2.4. Sampling the 4 shallow monitoring wells (MW25, MW37, MW38 and MW40) near
each pond monitoring station, and the raw, mixed water (RAW) as
Sampling was performed during a reference period without AOP recovered from the entire dune infiltration system at Katwijk. Electrical
(2016–2017) and during a period with AOP (2019–2020). Sampling conductivity (EC) fluctuations in the infiltration water could be used as
locations with corresponding abbreviations are given in Fig. 1. For tracer, because they overshadowed the minor changes due to e.g. pre
detailed information on sampling dates and frequency, see cipitation reactions in the ponds and dissolution reactions in the dune
3
Fig. 2. The southern part of dune infiltration area Berkheide with the transport mains (BAL1 and BAL2) and entry points of river water, the investigated infiltration
ponds (IP), monitor wells (MW) and local raw water (RW) recovery system.(For interpretation of the references to colour in this figure, the reader is referred to the
web version of this article.)
sand aquifer. The time shift needed to create the best overlap for the EC 2.6. Chemical analysis
signal of the inlet and 4 pond stations, and for each pond station and its
nearby monitoring well was accurately measured using data from a 2.6.1. Analytical procedures inorganics
combined EC - temperature data logger (LTC Levelogger Edge, M10/ The main constituents of water were analyzed with conventional
C80) with a time interval of 1 h. methods that need no further specification. Trace elements, among
The modal travel time in the upper aquifer system of the entire dune which arsenic and chromium, were analyzed with inductively coupled
infiltration system was estimated by measuring the time shift needed for plasma mass spectrometry (ICP-MS; Thermo-Fischer) with a method
overlap of chloride and temperature fluctuations in input and output. equivalent to NEN-EN-ISO 17294–2. Bromide was analyzed conform
For temperature, a retardation factor of 1.39 was calculated assuming a NEN-EN-ISO 10304–1 using ion exchange chromatrography with an
dune sand porosity of 41%. ICS-1100. Bromate was analyzed using ion exchange chromatography
The Centralized Moving Average (CMA) of the input signal was used followed by conductivity detection using a Dionex ICS-3000.
to account for the attenuation (dampening) of EC fluctuations in the
input as observed in the output. The required number of hours in the 2.6.2. Analytical procedures organic micropollutants
CMA is a crude dispersion indicator. A total number of 160 organic micropollutants (OMP) were analyzed
with eight quantitative analytical methods. In the Supplemental Infor
mation, the compounds are grouped per analytical method and the limit
4
of quantification (LOQ) is indicated (Supplemental Table S2). Briefly, X- (MM) were used, as described previously (Been et al., 2021a). Samples
ray contrast media were analyzed without sample pre-treatment with a were collected in 250 ml HDPE bottles (in triplicate). NTS data was
method based on a separation of the analytes by ion chromatrography processed with Compound Discoverer 3.1 (Thermo Fisher Scientific) for
(IC) and a subsequent detection by ICP-MS, as described (Sacher et al., peak picking, componentization and differential analysis. The output of
2005). The analysis of pharmaceuticals was performed in solid phase Compound Discoverer was a feature list, consisting of a table with ac
extracts of the water samples prepared with Oasis HLB columns eluted curate mass / retention time pairs (features) and their intensity reported
with MeOH that were injected into an ultra-high performance liquid as peak area. Features that were less than ten times higher than the
chromatography (UHPLC)-triple quad MS, as described previously maximum of the instrumental and bottle blanks, were removed. The
(Houtman et al., 2013). The analysis of sweeteners was performed in detection of transformation products was performed by differential
solid phase extracts of the water samples prepared with Bakerbond SDB1 analysis, i.e. comparing samples after and before AOP.
cartridges at pH 3 and analyzed by LC electrospray ionization tandem
MS operating in the negative ionization mode, as described in (Scheurer 2.7. Toxicity assessment using bioassays
et al., 2009). The analysis of perfluorinated compounds was performed
in solid phase extracts of the water samples prepared with C18 SPE 2.7.1. CALUX bioassays
Sep-Pak Vac 3 cm3 cartridges eluted with MeOH that were injected into Reporter gene assays based on selective human U2-OS cell based
a HPLC connected to a tandem MS operating in the negative ionization lines were used to measure anti-androgenic (anti-AR), estrogenic (ER)
mode, as described in (Eschauzier et al., 2010). Samples for the analysis and glucocorticoid (GR) activities with the anti-AR (antagonistic mode),
of benzotriazoles and pesticides were collected in sample vials of 40 ml ERα and GR CALUX (BioDetection Systems B.V., Amsterdam, the
containing 32 µl formic acid. Samples were analyzed without sample Netherlands) as described by (Houtman et al., 2018; van der Linden
pre-treatment and were injected directly into the system. Chromato et al., 2008). The reference compounds were flutamide (Flt),
graphic separation was achieved on a HSS T3, 2,1 × 100 mm 1,8 µm 17β-estradiol (E2) and dexamethasone (Dex) respectively.
column and samples were analyzed by LC electrospray ionization on an Anti-androgenic activity was tested in the presence of dihy
UHPLC-triple quad MS operating in the positive ionization mode. drotestosterone (DHT) at EC50 level. Furthermore, the p53 and Nrf2
The analysis of utropine was performed using a UHPLC-triple quad CALUX were used to assess the transcriptional activation, the p53
MS, operated in the positive electrospray ionization mode. The water pathway related to (in)directly actin genotoxins without metabolic
samples were filtered using a 0.20 µm regenerated cellulose filter before activation of S9 liver enzymes, and the Nrf2 pathway for oxidative stress
injecting directly into the system. The chromatographic separation was as described by (van der Linden et al., 2014). The reference compounds
performed on a Phenomenex Kinetex HILIC column (3.0 × 150 mm, were actonymcin D (Act) and curcumine (Cur), respectively. Cell
2.6 µm). viability (absence of cytotoxicity) was measured with the cytotox
Samples for the analysis of complexing agents were collected in CALUX (van der Oost et al., 2017) to check if inhibition of responses in
250 ml borosilicate glass bottles and acidified by the addition of 0.25 ml the other CALUX assays was caused by specific receptor mediated re
nitric acid 22%. Samples were filtered over a 0.45 µm spartan filter. sponses or due to cytotoxicity. Tributyltin acetate (TBT) was used as a
50 µl iron nitrate solution was added to 1 ml sample, mixed and heated reference compound. The CALUX assays were only performed in 2019
at 65 ± 2 ◦ C. Samples were injected into a LC system, whereby chro after AOP implementation. The results of the CALUX assays were
matographic separation was achieved on a ChromSep Inertsil ODS-2 C- interpreted by comparing the measured responses to effect-based
18, 250 × 4,6 mm column. Subsequently, analysis was performed by UV ‘trigger values’ (EBT) as proposed for the environment (Been et al.,
detection by a diode array detector (DAD). Volatile organic compounds 2021b) and for drinking water (Heringa et al., 2011). For more details,
were analyzed using a purge and trap injector (PTI) couples to a gas see the ‘Supplementary information on the toxicity assays’.
chromatograph mass spectrometer (GC-MS). Samples were collected in
43 ml vials containing 0.5 ml sulfuric acid. 20 µl of internal standard 2.7.2. Ames fluctuation test
was added to these vials which were subsequently placed into the The Ames fluctuation test was performed as previously (Heringa
autosampler of the GC-MS-PTI. After injection the volatile compounds et al., 2011), with minor modifications. Salmonella typhymurium strains
were trapped in the purge and trap unit and after flash heating of the TA98 and TA100 (Xenometrix, Switzerland) were exposed to a 100-fold
trap the volatile compounds were transferred into the analytic column concentrated water extract (n = 2) in culture medium. Water extracts
(Restek 60 m, 0.32 mm ID df 1.8 µm (Rtx-VMS)) and subsequently were tested with and without an exogenous metabolic activation system
detected with an MS. (rat liver S9 mix), resulting in four different test conditions (TA98-S9,
Besides the target analysis of organic micropollutants, the samples TA98 +S9, TA100-S9 and TA100 +S9). Procedure controls (extracts of
were also analyzed with a targeted screening analysis. The data obtained Evian mineral water), negative controls (dimethyl sulphoxide, DMSO)
with this method were semi-quantitative and expressed in peak areas. and positive controls were run in parallel. Two independent experiments
Peak areas of a single compound were compared between samples, were performed and all experiments were carried out in triplicate cul
giving an indication of the removal of the compound during the water tures. Cytotoxicity was measured in parallel by measuring optical den
treatment process. The database used for targeted screening contained sity of the bacterial culture (Heringa et al., 2011). For more details, see
around 2000 compounds. The target analysis was performed on an the Supplementary information on the toxicity assays.
Impact II Quadrupole Time Of Flight (QToF) MS. Samples for the
screening were collected in a 40 ml vial. Samples were filtered over a GE 2.8. Biomass Production Potential (BPP) and Assimilable Organic Carbon
Healthcare 10 syringe filter lc, 0.2 µm. After filtration 1.0 ml sample and (AOC)
50 µl internal standard were added. The samples were injected into the
UHPLC-QTOF equipped with ESI source operating in positive and Biological stability was evaluated using a Biomass Production Po
negative ionization mode. tential (BPP) test and assimilable organic carbon (AOC) analyses. The
In addition, a non-target screening (NTS) was performed for the BPP was determined according to (Hijnen et al., 2018). The BPP test
detection of highly polar transformation products. Non-target screening included water sampling in AOC free, glass-stoppered Erlenmeyer flasks
was carried out using liquid chromatography – high resolution mass and incubation of samples at 25 ◦ C for 14 days. During incubation,
spectrometry (LC-HRMS) operated in the positive and negative elec growth of the autochthonous microbial community in the water samples
trospray ionization mode. For the detection of highly polar trans was monitored by periodic ATP analysis (van der Wielen and van der
formation products, two different chromatographic conditions, i.e. Kooij, 2010). The BPP was expressed as the maximum ATP concentra
hydrophilic interaction liquid chromatography (HILIC) and mixed mode tion during the first seven days of incubation (BP7, ng ATP/L) and the
5
cumulative ATP production from day 0–14 (BPC14, d ng ATP/L). 3. Results

The concentration of easily assimilable organic carbon (AOC) was
carried out as described previously (van der Kooij et al., 1982). 3.1. Travel times
2.9. Microbiological analysis In Supplementary Fig. S1, two examples are presented of the ‘signal
overlap method’ to determine the travel time from pond to its moni
2.9.1. ATP measurements and total cell counts toring well and from dune inlet to raw dune output, respectively. They
General microbial parameters were determined by measuring the showed a short average travel time (t50) of 22 h and a centralized
total ATP concentration and the total and membrane-intact cell counts. moving area of 5 h in the dune sand from pond station IP-37 to its very
Determination of the ATP concentration was carried out as previously nearby shallow monitoring well (MW-37), and a relatively long t50 of 52
described previously (van der Wielen and van der Kooij, 2010). days and centralized moving area of 112 days in the whole dune infil
Enumeration of direct total cell counts and intact cell counts was tration system. The mixing of flow lines with diverging travel times in
performed using flow cytometry (FCM) following the protocol described the system explained the high centralized moving area.
previously (Prest et al., 2013). The results of all travel time (t50) measurements in the dune infil
tration system, for the period without and with AOP, are summarized in
2.9.2. Microbial community analysis Table 1. They showed a much longer travel time for the remote ponds IP-
From the collected water samples, a maximum of 1000 ml was 25 and IP-40 (9–13 days) than for the nearby ponds IP-37 and IP-38
filtered through a polycarbonate membrane filter with a 0.2 µm pore (1–5 h), as expected for water flowing from inlet via the nearby to the
size (Track-etch membrane, Sartorius, The Netherlands). Filters were remote ponds. Dune sand passage in the pond bank took 0.3–4.6 days for
stored at − 20 ◦ C until further DNA extraction and 16S rRNA amplicon the 4 monitoring wells, of which MW-40 showed the longest t50. In the
sequencing, which was done as described previously (Vavourakis et al., entire dune infiltration system (from inlet to output), the t50 of ~52
2020). Negative controls (water samples of DNase/RNase free water) days consisted of on average 10 days in the ponds and 42 days in the
and positive controls (Zymo DNA mock communities) were also added dune sand aquifer.
during filtration and subsequent DNA extraction. The average travel time in the entire dune infiltration system was
Processing of 16S rRNA amplicon sequence data was performed nearly equal for both periods (with and without AOP). For most infil
using Qiime2 (Bolyen et al., 2019). Firstly, paired-end sequences were tration ponds and shallow monitoring wells the average travel time
joined using vsearch. Then, sequences were quality filtered and chi decreased during the period with AOP, but not substantially.
meras were removed. Afterwards, denoising and amplicon sequence
variants (ASVs) were determined using Deblur. Taxonomy was assigned 3.2. Inorganic water quality changes
using the q2-feature-classifier plugin and the readytowear
SILVA_132_515F_806R_water-non-saline classifier (Kaehler et al., 2019; 3.2.1. General patterns during both periods (+/- AOP)
Bokulich et al., 2018) based on the SILVA reference database (Quast Dune infiltration leads to important water quality changes by various
et al., 2012). Afterwards, alpha and beta diversity analysis was per processes in the infiltration ponds, during passage of the dune sand
formed for first evaluation of results. Bray Curtis dissimilarity results aquifer, and by mixing in the recovery system (Stuyfzand, 1989). This
were log transformed and plotted using PCoA and statistical testing was can also be deduced from the monitoring data collected in this study. A
done using PERMANOVA pairwise comparison with 999 permutatons summary of the most important data is given in Supplementary table S3.
with a post-hoc Adonis test. Differential abundance analysis was done The parameters that showed significant changes in the ponds and during
using ANCOM, after filtering features (minimum frequency 50 reads in dune infiltration are depicted in Fig. 3.
minimum 4 samples) and adding pseudocounts (Mandal et al., 2015). In the infiltration ponds close to the inlet (IP-37 and IP-38) with short
Qiime2 output was imported into R studio using Qiime2R (Bisanz, retention time (‘IP-short’), minor changes occurred, but further away
2018). Further analysis and visualization was done with the R software from it (IP-25 and IP-40), with longer retention times (‘IP-long’), the
packages Ampvis2 (Andersen et al., 2018), and phyloseq (McMurdie and changes increased (Fig. 3). The main significant changes between the
Holmes, 2013). short and long retention time ponds consisted of: (i) a significant
decrease of NO3- and SiO2, possibly by uptake for biomass production by
2.10. Statistical analysis algae, diatoms and reeds, (ii) a significant increase of NH4+ and PO43-
possibly due to their remobilization from bottom muds which evidently
Statistical analysis of organic micropollutant concentrations over exceeded their uptake; (iii), a significant increase of DOC (which was
time (before and after AOP) were performed with Trendanalist (Bag strongly correlated to TOC, UV254-extinction and COD) probably due to
gelaar and Eit, 2019), software capable of performing a fully automatic
trend analysis of a data series in a database. For each data series, the
Table 1
software used the trend test that best fitted its relevant statistical char Travel time to various monitoring stations during the period without AOP (2
acteristics, being the kind of probability distribution (normal or Jan. – 26 Feb. 2018) and with AOP (3 July 2019 – 4 Dec. 2019). IP = Infiltration
non-normal) and the occurrence or absence of autocorrelation and/or pond; MW = shallow monitoring well in pond bank, RAW = abstracted water.
seasonality. The parametric linear regression test was used for normal
From to Travel time
distributed data and the distribution free Mann–Kendall test for
2 Jan. 2018–26 Feb. 2018 3 July 2019–4 Dec. 2019
non-normal distributed data.
Statistical analysis of correlations between different parameters were hour day hour day
determined using spearmann correlation tests and were performed on all INF-B IP-37 3 0.13 1 0.04
collected data using the R software where only correlations > 0.4 with INF-B IP-38 5 0.21 4 0.17
p < 0.05 were outputted. Statistical analysis of differences in parame INF-B IP-25 228 9.5 # #
INF-B IP-40 315 ## 13 ## # #
ters between periods with and without AOP or between different loca
iP.37 MW-37 22 0.9 15 0.6
tions were performed using the Wilcoxon rank test (p < 0.05). iP.38 MW-38 16 0.7 8 0.3
iP.25 MW-25 7 0.3 40 1.7
iP.40 MW-40 110 4.6 100 4.2
INF-B RAW 1200 50.0 1296 54
#: data errors ##: estimate based on various data
6
Fig. 3. Box plots showing changes in selected main constituents during transport through the MARR system. From left to right: infiltration water (INF), infiltration
ponds with short (IP-short: IP-37 and IP-38 together) and long (IP-long: IP-25 and IP-40 together) detention and their monitoring wells (MW-short, MW-long) in the
period with (red) and without AOP (green). Asterisks (*) show significant differences between the ponds with short and long detention time for both periods (+ and -
AOP), Carets (^) show significant differences between the infiltration water and monitoring wells for both periods (+ and - AOP) together, and circles (O) show
significant differences between the period with or without AOP within the different type locations. Significance tested by the Wilcoxon rank test, with p < 0.05.(For
interpretation of the references to colour in this figure, the reader is referred to the web version of this article.)
7
intensifying primary production; and (iv) a significant decrease of Ca2+ 3.2.2. Effect of AOP
and HCO3- due to possible biogenic hardness reduction (CaCO3 precip To determine the effect of AOP on these inorganic parameters, both
itation by algal CO2 uptake) (Wilcoxon rank test, p < 0.05). periods (+/-AOP) were compared on the measured inorganic parame
The data also showed that 1–1.6 m of dune sand passage in 7–110 h ters and TOC/DOC (Fig. 3 and Supplementary table S3). It was observed
resulted in larger quality changes than during pond detention (Fig. 3). on practically all observation points, that DOC and SiO2 were signifi
The changes increased with larger distance (MW-25 and MW-40) to the cantly lower and O2 and NO3 were higher in the period with AOP. The
dune inlet (INF) and with longer travel time in the dune sand (MW-40). lower PO43- and NH4+ concentrations in the infiltration ponds and
This especially was the case for the significant decrease of O2, NO3-, and monitor wells also suggested that during the period with AOP, the
DOC (and therefore TOC and COD) between infiltration water and oxidation reduction potential (deduced from the higher O2 and NO32-
monitoring wells of long retention ponds (MW-long) (Wilcoxon rank concentrations) became slightly higher. Whether this small change is
test, p < 0.05). Cu, Pb, Se and Zn also showed a decease, but these were due to AOP is hard to prove, because environmental conditions in the
not significant. In MW-40, the water became anoxic, probably allowing dune infiltration area varied over the years and seasons. The datasets for
NH4+ and PO43- to be mobilized from organic material in bottom muds. each period were too short to take these variations into account.
Anoxic water also possibly allowed Fe2+ and Mn2+ to reductively
dissolve (Supplementary table S3). 3.3. Organic micropollutants (OMP)
The overall changes in the entire dune infiltration system (RAW)
were larger than those in MW-40. This was due to a longer travel time 3.3.1. Effect of AOP
over a longer distance (on average 40 m). Also, fluctuations in Meuse From the 164 monitored OMP, 70 compounds were detected at least
River quality were far more attenuated in RAW by mixing of flowlines once in the intake water (INT) in 2019 in concentrations above the LOQ.
with different length and flow velocity, and by the admixing of ambient For 48 compounds the removal rate in AOP was calculated. These con
dune groundwater (~10%). A substantial concentration rise of SiO2, cerned industrial compounds, pharmaceuticals, X-ray compounds,
PO43-, Fe, NH4+, As, Ti and V in the mixed raw water (RAW) as sweeteners and per- and polyfluoroalkyl substances (PFAS). These
compared to the MW’s occurred (Supplementary table S3). This was compounds are listed in Supplementary table S4, where the maximum
probably due to the dissolution of diatoms, oxidation of soil organic concentration in INT is shown, the removal rate by AOP, and the
material, reductive dissolution of iron (hydr)oxides, and desorption. On removal rate between INT and INF before and after implementation of
the other hand, a substantial concentration decrease for TOC, DOC, UV- AOP. Most OMP (31 compounds) were removed by AOP for more than
extinction, K, SO42-, Al, Ba, Cd, Cs, Mo, Ni, Rb, Sb, Se, Sn, Ta and W was 80% and six OMP were removed partially (50 − 80%). Some compounds
observed in RAW (Supplementary table S3), which can probably be such as MTBE, metformin, sucralose and eight PFAS compounds were
explained by biodegradation of dissolved organic compounds, sorption not removed by AOP.
and some dilution with ambient dune groundwater. After AOP treatment, the water was infiltrated in the dunes. At the
9
Sweeteners (sum)
Urotropine
8
PFAS (sum) (conc*10)
EDTA (conc/10)
7 Benzotriazoles (sum)
X-ray contrast agents (sum)
Pharmaceu�cals (sum)
6
Concentra�on (µg/L)
0
before a�er a�er a�er a�er before a�er before a�er
INT RSF AOP INF-B RAW DW
Fig. 4. Comparison of the average sum concentrations of compounds, grouped by their application, before and after the addition of AOP.
8
infiltration point (INF-B), the AOP effluent was diluted with untreated 3.5. Toxicity assessment using bioassays
rapid sand filtrate in a ratio of approximately 40–60% (Fig. 1). The
removal rate between INT and INF increased statistically (p < 0.05) for Detailed results of the Ames fluctuation test are shown in the Sup
most compounds after implementation of AOP (Supplementary table plementary Information on toxicity assays. The samples INT, RSF, AOP,
S4). In Fig. 4 the sum of the average concentrations on the different INF-B, IP-38, MW-38 and RW showed a clear positive response in at least
sampling locations in the drinking water treatment process are clustered one sample campaign, whereas the samples IP-40, MW-40 and DW only
by the type of application. Pharmaceuticals, X-ray agents, and benzo showed negative responses.
triazoles were generally well removed. From the measured compounds, Based on the results obtained in the current study, estrogenic activ
the sweeteners were the largest contributors to the sum concentration ity, anti-androgenic activity and mutagenic activity were observed in
after AOP. This is due to sucralose which was detected in relative high intake water, but were clearly reduced, absent or below the limit of
concentrations of 0.7–2.8 µg/L after AOP. The sum concentrations in quantification (LOQ) in the drinking water. Cytotoxicity, glucocorticoid
drinking water did not decrease after the implementation of AOP. In activity, direct genotoxocity (measured as an induction in the p53-
drinking water, EDTA, urotropine and PFAS compounds were detected enzyme) and oxidative stress were absent or below the detection limit
in similar concentrations as before the implementation of AOP. The in all samples in the current study. Anti-androgenic substances were
main reason for DW showing higher sum concentrations with AOP, was efficiently removed by AOP (Supplementary Fig. S4). Estrogenic sub
the increase of sweeteners in the input, which survived MARR. In stances seemed to be not completely removed by AOP, whereas subse
addition, the raw water and drinking water also consisted of infiltrated quent MARR appeared to be successful in this (Supplementary Fig. S4).
river water with longer travel times in the MARR system and, in various The presence of mutagenic activity was variable after rapid sand
cases, with higher input concentrations in the past. filtration and AOP and plausible in one of the infiltration ponds, but in
From the compounds that were oxidized for more than 50%, five drinking water no activity of mutagenic substances was measured in the
compounds were detected in drinking water above the LOQ. Concen bioassays.
trations of these compounds are given for INT, INF and DW in a five-year
period between 2016 and 2020 (Supplementary Fig. S2). The concen 3.6. ATP, cell numbers, growth potential and natural organic matter
trations of acesulfame, amidotrizoic acid, iopamidol, EDTA and uro
tropin fluctuated strongly in INT. After MARR, concentrations were 3.6.1. Through treatment
more constant in DW. In the period of 2016–2020, the concentrations of Results showed that there were fluctuations in ATP, TCC, BP7 and
amidotrizoic acid, iopamidol and urotropin decreased significantly in BCP14 during water treatment. Most parameters (ATP, TCC, BP7 and
DW. In INT the concentrations did not decrease significantly. Concen BPC14) had highest values in INT and IP-25 and IP-40 as compared to all
trations of acesulfame and EDTA were not decreasing significantly yet, other purification steps in both periods without and with AOP taken
but for both compounds the concentration in DW were showing a together (Supplementary Fig. S5). These parameters strongly fluctuated
decreasing trend. between sampling dates, especially in the ponds with long retention (IP-
In addition to the target analysis of selected compounds, a broader 25 and IP-40). The parameters were significantly correlated to each
targeted screening was performed on the LC-QTOF. The database of the other, but also to pH, nutrients and organic fractions, such as nitrite,
targeted screening contains almost 2000 compounds. The targeted ammonium, phosphate, biopolymers, neutrals, building blocks, humic
screening is not a quantitative method, but the results expressed in peak substances, and DOC (Spearmann correlations rho >0.5 and <− 0.5,
area can be compared between sampling locations for a single com p < 0.05, Fig. 5). We also observed a strong correlation between BP7 and
pound. A comparison of the peak areas before and after AOP was BPC14, which indicated that most biological active compounds were
therefore possible. In Supplementary table S5 the removal rates are easily degradable (within 7 days) (Spearmann correlation rho 0.95,
shown for the 45 compounds that were detected in > 25% of INT sam p < 0.05). After MARR, BP7, BPC14, ATP and cell counts drastically
ples, and that were not measured with a quantitative analytical method. decreased in MW’s, RAW and DW (Supplementary Fig. S5), which was
Most compounds were removed for > 80% during AOP. Only one correlated to a decrease in nutrients, DOC and most organic fractions.
compound, the DNA component adenosine, was removed < 30%. Dune infiltration therefore seemed to form a natural barrier for organic
Adenosine was however removed during subsequent MARR. The seven material, nutrients and microorganisms.
compounds that were detected at least once in drinking water were
either well removed (gabapentin, N-formyl-4-aminoantipyrine, pyr 3.6.2. Effect of AOP
imethanil, dimethomorph) or partially removed (ritalinic acid, terbu Before dune infiltration, the BPP parameters, ATP and cell counts
tryn, melamine). were not significantly different after rapid sand filtration (RSF) and in
the dune influent (INF-B and INF-M) between the period before and after
3.3.2. AOP byproducts AOP (p > 0.05) (Fig. 6). DOC concentrations were however significantly
Bromate and bromoform are compounds that were not detected in lower during the period with AOP in both ponds with short (IP-37 en IP-
INT. During AOP, both compounds were formed and detected in AOP 38) and long (IP-25 and IP-40) retention time together and in the
effluent with concentrations up to 0.8 µg/L bromate and 0.3 µg/L bro resulting DW compared to the period without AOP (Wilcoxon rank test,
moform (Supplementary Fig. S3). Bromoform was still detected in INF-B p < 0.05) (Fig. 3, Supplementary Fig. S5). In IP-37 and IP-38, the BPC14
above the LOQ, bromate was detected below the LOQ. Both compounds was also significantly lower during the period with AOP (Wilcoxon rank
were not detected anymore in RAW after MARR. test, p < 0.05), which could have been caused by the significantly lower
DOC concentrations. Since there was a strong correlation between DOC,
3.4. Transformation products BPP and other nutrients and organic fractions, most of these parameters
were also lower (but not always statistically significant) in the period
In order to determine if highly polar transformation products were with AOP in the IP’s (Supplementary Fig. S5).
formed with AOP treatment, LC-HRMS NTS was performed on samples In the period with AOP, the BP7 values were significantly higher after
without and with AOP. All features detected after AOP installation were AOP as compared to the rapid sand filtrate effluent, but decreased again
compared (differential analysis) with features detected before AOP. No when the water reached the dunes (INF-B) (Fig. 6). This decrease in BP7
features were observed that were statistically significantly different was most probably the effect of dilution by mixing with non-AOP treated
(p < 0.05) in the samples after AOP (data not shown). Based on these RSF effluent (Fig. 1). On average 42% (v/v) AOP treated and 58% (v/v)
results, no highly polar transformation products were detected after non-AOP treated RSF effluent were mixed when these BPP measure
AOP. ments were performed (feb-dec 2019). The theoretical BP7 and BPC14
9
DOC (mg/l)
Humics (ug/l)
Building blocks (ug/l)
Neutrals (ug/l)
Phosphate (mg/l)
Biopolymers (ug/l)
pH
Oxygen (mg/l)
BPC14 (ng ATP/l)
BP7 (ng ATP/l)
TCC (cells/ml)
Live intact CC (cells/ml)
ATP (ng/l)
Nitrite (mg/l)
Ammonium (mg/l)
Faith's PD index
Carbonate (mg/l)
l) l)
g/
g/ )
i u )
g tra (m l)
xy ng T )
Hu ck (u l)
O (u l)
l)
C g )
(m /l)
x
l)
ild N at (u pH
n TP )
l)
(n l) /m
(m /l
m s ( g/l
O 4 ( A ml
TO (mg/l
ge A P/l
de
in u e g/
D cs g/
/
g/
g/
P g/
bl ls g
(m lls
g /
in
AT (m
C1 (n lls
e
(c
PD
m BP 7 ce
ph rs
iu
C
ite
os me
BP (
on CC
's
o
itr
C
th
ct
TC
Ph oly
m
N
e
i
Fa
Am ta
op
Bi in
ve
Bu
Li
Fig. 5. Significant Spearman correlations (rho >0.5 and <− 0.5) between measured parameters for all samples taken in both periods with and without introduction of
AOP (p < 0.05). Colors represent different Spearman correlation classes (see legend).(For interpretation of the references to colour in this figure, the reader is
referred to the web version of this article.)
values that would be achieved after mixing were calculated according to 40) was highly variable and was significantly different from cluster A
these mixing proportions (Fig. 6, T42 values). These theoretical BP7 and with shorter detention time ponds (INT, RSF, INF, IP-37, IP-38) (PER
BPC14 values (T42) were significantly higher than the values measured MANOVA pairwise difference, p < 0.05; Fig. 7). This high variation in
at INF (Wilcoxon rank test, p < 0.05). The measured values were IP-25 and IP-40 was reflected in the microbial community composition
therefore probably lower due to biological activity during transportation of MW-25, MW-40, which made that these were not significantly
to the dunes. The flow resistance of the transport mains should increase different from the communities in the ponds. For MW-40, this was also
with increasing biofilm growth, but it did not show significant changes reflected in high fluctuations in the microbial diversity (Fig. 7 and
between the two periods with and without AOP (Supplementary Fig. Supplementary Fig. S7). This temporal variation was also observed for
S6). Furthermore, intact cell counts were significantly lower in INF-B in the biological parameters and (in)organic parameters (Supplementary
the period with AOP than without AOP. This was probably caused by Fig. S5). For IP-37 and IP-38, temporal variations were much smaller
AOP, as the cell counts after AOP were dramatically lower compared to and communities of MW-37 and MW-38 were significantly different
RSF (Fig. 6). The subsequent increase in intact live cell counts was also from the communities in the ponds itself (PERMANOVA pairwise dif
probably an effect of mixing with non-AOP treated RSF effluent. ference, p < 0.05). The microbial community composition was also less
variable over time in RAW and DW as compared to the infiltration
ponds. This showed that dune infiltration produced oligotrophic drink
3.7. Microbial diversity and community composition ing water with a high microbial diversity and increased microbiological
stability.
3.7.1. Through treatment
The Faith’s PD, as indicator for microbial diversity, was lowest in INT 3.7.2. Effect of AOP
and IP’s and increased significantly during and after dune infiltration (in The differences in Faith’s PD between each of the IP’s or the MW’s
MW’s, RAW and DW) (Fig. 7 and Supplementary Fig. S7, Kruskall Wallis, were larger during the period with AOP than without AOP, and were
p < 0.05). The microbial diversity was significantly and negatively therefore all significantly different from each other (Kruskall Wallis,
correlated to biological activity parameters, DOC/TOC, organic frac p < 0.05, Supplementary Fig. S7). From all sample locations, only DW
tions, nutrients and pH (Fig. 5) (Spearman correlation, p < 0.05). showed a significantly higher Faith’s PD (factor 4) during the period
Indeed, these parameters were much higher before entering the dune with AOP as compared to the period without AOP (Kruskall Wallis,
sand, such as in the IP’s, where microbial diversity was much lower. This p < 0.05, Supplementary Fig. S7), which correlated to a lower amount
relationship between the microbial diversity and other parameters was of DOC and TOC (Fig. 5 and Supplementary Fig. S5). This was probably
strongest in the ponds with long retention time (IP-25 and IP-40) as not an effect of AOP, as the Faith’s PD was not significantly different
compared to IP-37 and IP-38 and INT. Ponds IP-25 and IP-40 showed a with or without AOP treatment at INF (INF-M and INF-B). As stated
significantly lower diversity than IP-37 and IP-38 in both periods with previously, DOC concentrations were significantly higher in the IP’s
and without AOP, with a higher concentration of organics and nutrients during the period without AOP. The organic material in the IP’s there
and higher biological production potential. fore probably affected the organic material in the MW’s and the
The microbial communities were also significantly different in Bray- resulting DW, and could explain the difference in Faith’s PD as these
Curtis dissimilarity before and after MARR (Fig. 7, PERMANOVA pair parameters showed a strong and significant negative correlation (Fig. 5).
wise difference, p < 0.05). The microbial community composition of Microbial community compositions were not significantly different
cluster B with the long detention time ponds (IP-25, IP-40, MW-25, MW-
10
ATP (ng/ml)
BPC14 (ng ATP/l)

BPC7 (ng ATP/l)
* ^ ^
^ ^
*
*
1.5x106
6
2x10
Live intact CC (cells/ml)
TCC (cells/ml)
1x106
1.5x106
5x105
6
1x10
Fig. 6. Values for biological production potential (BP7, BPC14), ATP and total cell counts (TCC), live intact cell counts (CC) in the period with AOP (green) and
without AOP (red). The theoretical BP7 and BPC14 values that would be achieved after mixing AOP-treated with non-AOP treated RSF effluent (T42) are also shown
(BP7 = 59 (SD 4.7), BPC14 = 556 (SD 38.9). Standard deviations of these theoretical values were based on the relative standard deviations of the BPP analysis. These
relative SDs of BP7 (8%) and BPC14 (7%) were calculated as the median relative SD of all BP analysis performed. Values that were significantly different from each
other are indicated with the same symbol (Wilxocon rank test, p < 0.05).(For interpretation of the references to colour in this figure, the reader is referred to the web
version of this article.)
between sampling locations in the period with or without AOP (Bray producers in the IP’s at certain periods. However, higher sampling fre
Curtis distance, PERMANOVA, pairwise, p < 0.05). The variation in the quency and sampling at the same dates in the periods with and without
microbial community composition was indeed only for a very small part AOP are needed to ascertain the significance of differences between
explained by the effect of AOP (3.3%, Adonis test, p < 0.05) and mostly these communities, especially with respect to the primary producers.
explained by temporal differences (18.6%, Adonis test, p < 0.05). It was The microbial diversity showed significant differences between both
expected that the higher concentrations of organic material during the periods because Faith’s PD index is very sensitive for loss of species,
period without AOP was reflected in the abundance of primary whereas Bray Curtis distances are less sensitive to changes in the less
11
Fig. 7. Principal Coordinates Analysis (PCoA) plot based

on Bray Curtis distance for all samples that were analyzed
using 16S rRNA gene sequencing. Blanc samples and mock
communities were excluded for this analysis. Both axes
show 33% of the variation in microbial composition that
could be explained by the period before (triangles) or after
(circles) introduction of AOP. Sample colors represent the
microbial diversity (Faith PD index). Bray Curtis distances
that were significantly different are shown with clusters of
different colors (PERMANOVA (p < 0.05)). Cluster A (INT,
RSF, INF-B, INF-M, IP-37, IP-38) and cluster B (IP-25, IP-
40, MW-25, MW-40) represent treatment steps before and
during MARR.(For interpretation of the references to
colour in this figure, the reader is referred to the web
version of this article.)
abundant species of a community. polyfluoroalkyl substances (PFAS) and sucralose. These compounds are
The microbial community of the AOP effluent was significantly known to be very mobile and persistent during water treatment
different in Faith’s PD (Kruskall Wallis, p < 0.05) and composition (Bray (Scheurer et al., 2009; Rahman et al., 2014). Oxidation steps with
Curtis, PERMANOVA, p < 0.05) compared to all other treatment steps, UV/H2O2 and ozonation can partially oxidize sucralose, probably at
but after mixing with non-AOP treated RSF effluent, INF was not higher doses, but they are less efficient in oxidizing sucralose compared
significantly different anymore in microbial diversity and composition to the sweetener acesulfame-K (Scheurer et al., 2010; Wang et al.,
(p < 0.05). However, the microbial communities of IP-37 and IP-38 in 2018a). PFAS are resistant to oxidation because of the strong C-F bond
the period with AOP were more similar to the microbial communities of together with the electron withdrawing functional groups –COOH and
the AOP effluent than in the period before AOP (Supplementary Fig. S8). –SO3H in the structures of perfluoroalkyl carboxylic acids and per
Since these observations were not significant, this could mean that a fluoralkyl sulfonic acids, respectively (Rahman et al., 2014).
minor part of the total microbial community was affected by AOP. It was For the OMP that were partially or completely oxidized, the con
therefore examined which amplicon sequence variants (ASVs) showed a centration in INF-B was reduced accordingly. Before MARR, the AOP
difference in relative abundance before and after AOP, and if these ASVs effluent was diluted with approximately 40–60% untreated RSF effluent
proliferated after AOP, and eventually end up in DW. ASVs belonging to water. Due to the OMP concentration fluctuations in the intake water
the genera Gallionella, Bosea and Porphyrobacter were indeed signifi over time and the fluctuation in the exact ratio between AOP and un
cantly more present in IP-37 and IP-38 during the period with AOP than treated water at INF, it was not possible to determine exactly if the
without AOP, but not in IP-25 and IP-40 (ANCOM differential abun reduction in concentration between INT and INF was completely due to
dance testing, W-value >1919, Supplementary Fig. S9). These genera AOP. However, the OMP concentration reduction in INF in the period
belonged to the relatively most abundant genera of the AOP effluent with AOP compared to the period without AOP was on average nearly
(Fig. 8). Some of these taxa were already present in the RSF effluent, but 40% for OMP compounds that were oxidized > 80% by AOP. This means
most seemed to be specifically enriched after AOP treatment, such as that the concentrations of those OMP in INF would become equal to the
Hyphomicrobium, Gallionella, Bosea, Porphyrobacter, Hydrogenophaga, concentration in the AOP effluent if the oxidation step would treat 100%
Legionella, Sphingomonas and Afipia. During transport and mixing with of the dune influent. MARR itself also reduced the concentration of
non-AOP treated water, these taxa decrease in relative abundance but many OMP. Some of the OMP that were poorly oxidized by AOP, such as
they did arrive in the first IP’s. As soon as the water was infiltrated, or metformin, MTBE and adenosine, were completely removed by MARR.
guided to the infiltration ponds with longer detention, the effect was not Previous research showed poor removal of MTBE during MARR because
visible anymore (Fig. 8). The relative abundance of these taxa fluctuated of poor sorption characteristics (Segers and Stuyfzand, 2007), but
during the next dune infiltration and treatment steps, and some were removal highly depends on influent concentrations, residence time
also present in the drinking water, such as ASVs belonging to the (especially in ponds), media sorption characteristics, water temperature
Hyphomicrobium and Legionella genus. This was, however, also the case and redox conditions. It was hypothesized that AOP could have a posi
during the period without AOP, where they showed similar relative tive effect on the degradation rate of OMP during MARR due to AOC
abundances and was therefore not an effect of AOP treatment. formation during AOP that results in increased biological activity after
AOP. MTBE was only detected twice above the LOQ in INF-B, IP-38 and
4. Discussion MW-38. Metformin was only detected once above the LOQ in INF-B. It
was therefore not possible to investigate the hypothesis of the positive
4.1. Effect of AOP on organic micropollutant oxidation and removal effect of AOP on OMP oxidation by MARR. However, biological activity
analysis before and after introduction of AOP did give more insights into
Implementation of AOP during drinking water treatment clearly this, which is discussed in Section 4.3).
resulted in a high oxidation of most OMP. From the detected compounds Although most OMPs were oxidized and removed during AOP and
that were either measured quantitatively or qualitatively via targeted subsequent MARR, some compounds were regularly detected in drink
screening, the majority was oxidized for more than 80%. OMP that were ing water. This concerned the sweetener acesulfame-K, the X-ray
neither oxidized during AOP, nor removed during MARR were per- and contrast agents amidotrizoic acid and iopamidol, and the industrial
12
Actinobacteria - HGCL clade

Bacteroidia - Flavobacterium
Gammaproteobacteria - Hydrogenophaga
Alphaproteobacteria - Sphingorhabdus
Alphaproteobacteria - Brevundimonas
Gammaproteobacteria - Gallionella
Subgroup 6 - Bacterium gp6
Gammaproteobacteria - Undibacterium
Gammaproteobacteria - Legionella
Alphaproteobacteria - Hyphomicrobium
Alphaproteobacteria - Sphingomonas
Alphaproteobacteria - Sphingomonadaceae uncultured
Gammaproteobacteria - Leptothrix
Alphaproteobacteria - Novosphingobium
Alphaproteobacteria - Sphingopyxis
Alphaproteobacteria - Rhirzorhapis
Alphaproteobacteria - Bosea
Alphaproteobacteria - Afipia
Alphaproteobacteria - Polymorphobacter
T F -M 7 37 5 25 38 8 4 0 40 W W
IN RS NF P-3 W- P-2 W- IP- W-3 IP- W- RA D
I I M I M M
M
Actinobacteria - HGCL clade

Bacteroidia - Flavobacterium
Alphaproteobacteria - Hyphomicrobium
Gammaproteobacteria - Gallionella
Alphaproteobacteria - Bosea
Alphaproteobacteria - Porphyrobacter
Gammaproteobacteria - Hydrogenophaga
Gammaproteobacteria - Legionella
Alphaproteobacteria - Sphingorhabdus
Subgroup 6 - Bacterium gp6
Alphaproteobacteria - Sphingomonadaceae uncultured
Alphaproteobacteria - Brevundimonas
Blastocatellia - Blastocatella
Gammaproteobacteria - Pseudoduganella
Gammaproteobacteria - Leptothrix
Gammaproteobacteria - Rhizobacter
Alphaproteobacteria - Sphingomonas
Alphaproteobacteria - Afipia
Gammaproteobacteria - Undibacterium
F OP -B 37 37 25 25 38 38 40 40 W W
RS A INF IP- W- IP- W- IP- W- IP- W- RA D
M M M M
Fig. 8. Heatmap of the average relative read abundance (%) of the most abundant genera occurring in the AOP effluent and their relative abundance in other
purification steps. The average relative abundance was calculated for all samples taken throughout the sampling period before (top) and after (bottom) introduction
of AOP.
compounds EDTA and urotropin. The maximum concentrations in DW (https://rvszoeksysteem.rivm.nl/). Although the intake of these com
of 0.3 µg/L acesulfame-K, 0.07 µg/L amidotrizoic acid, 0.06 µg/L iopa pounds via drinking water does not pose a risk for human health, it is
midol, 13 µg/L EDTA and 0.5 µg/L urotropin in the period 2016–2020 still desirable that concentrations are lowered because drinking water
were well below the provisional drinking water guideline values that are utilities strive to achieve drinking water concentrations that are as low
established by the Dutch National Institute for Public Health and the as reasonably achievable (ALARA) (van den Berg et al., 2017).
Environment, i.e. 3200 µg/L acesulfame-K, 250,000 µg/L amidotrizoic
acid, 415 µg/L iopamidol, 600 µg/L EDTA, and 500 µg/L urotropin
13
4.2. Effects of AOP on bromate and bromoform formation 4.4. Effects of natural variations on (micro)biological growth potential
and communities
Bromate and bromoform are known to be formed from bromide
during AOP (Haag and Hoigne, 1983). Bromide was present in INT in Results showed that the surface waters (the intake water and infil
concentrations between 90 and 150 µg/L. Bromide concentrations in tration ponds) showed highest values and strongest fluctuations for ATP,
INT were negatively correlated with the discharge of the Meuse River TCC, BP7 and BCP14 in both periods without and with AOP taken
(Supplementary Fig. S10). On the sampling dates with higher bromide together, especially the ponds with long retention times. These param
concentrations in INT, higher concentrations of bromate and bromoform eters were significantly and positively correlated to each other and to
were detected after AOP treatment. According to the Dutch govern pH, nutrients and organic fractions, such as nitrite, ammonium, phos
mental decision on infiltration based on the Soil Protection Act, water phate, biopolymers, neutrals, building blocks, humic substances, and
that is infiltrated should not contain more than 2 µg/L of tri DOC. The microbial diversity (Faith’s PD index) was also significantly,
halomethanes determined as the sum of bromoform, chloroform, bro but negatively correlated to the above mentioned parameters. The mi
modichloromethane and dibromochloromethane. The last three crobial diversity was much lower in the infiltration ponds before MARR,
compounds were not detected above their LOQ of 0.03 µg/L. The con especially in the ponds with long retention times. The variation in the
centration of bromoform in INF was well below this concentration. For microbial community composition throughout the year was also espe
bromate a ‘predicted no effect concentration’ (PNEC) of 11 µg/L was cially observed in the infiltration ponds with longer detention times.
derived by Environment and Health Canada from the lowest acceptable Since these effects were mostly observed in the ponds with long
toxicity value identified for a freshwater organism, i.e. 1.1 mg/L for retention times, it seems to be caused by external seasonal effects (i.e.
Hylalella azteca (H.C.E. Environment Canada, 2010). For bromoform, a precipitation, sun hours, temperature, presence of birds or other fauna
PNEC value of 13 µg/L is derived for freshwater organisms and a PNEC in and around the ponds). These have an effect on the availability of
of 2.26 µg/kg soil dw is derived for soil organisms (E.C.A. ECHA, 2021). nutrients (nitrate, ammonium, phosphate), which determine the pri
Based on these toxicity values, bromate and bromoform in the detected mary production, the pH and the concentration of organic compounds
concentrations were not likely to cause a concern for the ecology in the and the biological production potential (BPP) and microbial activity
dune filtration areas. (ATP, cell counts). Growth of phototrophic primary producers caused a
Health based drinking water guideline values were established for lower microbial diversity in stratified light-driven ecosystems (Bernstein
bromate and bromoform by the World Health Organization. For bro et al., 2017). Longer detention times therefore could facilitate more
moform a drinking water guideline of 100 µg/L was derived (W.H.O. , phototrophic primary production and competitive growth due to the
2005a). For bromate a concentration of 2 µg/L in drinking water was higher concentrations of nutrients and easily degradable compounds,
associated with an upper-bound excess lifetime cancer risk of 10− 5. resulting in lower microbial diversity.
However, based on analytical and technological feasibility, the sug Most of these above mentioned parameters were also lower (but not
gested guideline value by WHO is 10 µg/L (W.H.O. , 2005b). Since always statistically significant) in the period with AOP in the infiltration
bromate and bromoform were removed during MARR, and not detected ponds, indicating that the period with AOP was a less productive period
in drinking water, an increased health risk via intake of drinking water for the infiltration ponds as compared to the period without AOP. This
was not likely. also explained the observed, significantly lower silicate concentrations
in the ponds in the period with AOP, which was probably due to a lower
diatom production. A higher sampling frequency is however necessary
4.3. Effects of AOP on (micro)biological growth potential and to statistically confirm the causes for these seasonal fluctuations.
communities
4.5. Effects of AOP on toxicity bioassays
AOP increased the biological production potential by introducing
easily degradable compounds in the AOP effluent. These compounds Effects of AOP on the presence of OMP was assessed biologically
decreased to similar concentrations as in the period without AOP due to using the Ames fluctuation test for mutagenicity (Heringa et al., 2011)
mixing with non-AOP treated RSF effluent, and partly by biological and six CALUX reporter gene assays to cover additional toxicological
degradation during transport to the MARR system. The biological mechanisms such as endocrine disruption, cytotoxicity, direct geno
degradation in the transport mains did not (yet) result in substantial toxicity and oxidative stress (van der Linden et al., 2008, 2014). The
biofilm formation according to flow resistance measurements. It is results obtained in the current study provided insights in the effectivity
therefore probable that AOP did not stimulate biological degradation of of the water treatment process and demonstrated that AOP in combi
OMP during MARR. This could be different if in the future when a higher nation with MARR was able to completely remove substances with
proportion of AOP-treated water will be infiltrated. biological toxicity based on the toxicological endpoints measured.
Introduction of AOP did however result in lower cell numbers of the In the performed CALUX assays, indications were found for presence
infiltrated water. This effect was cancelled out in the first infiltration of estrogenic and anti-androgenic substances at the intake water as
ponds by primary production and other biological activity. Lastly, mi detected with the ERα and anti-AR CALUX respectively. ER and anti-AR
crobial taxa that were increased in relative abundance after AOP treat activity in the intake water were detected at levels below the environ
ment did have a significantly higher relative abundance in the first mental EBT (van der Oost et al., 2017), with the exception of one outlier
infiltration ponds as compared to the period without AOP. These taxa of 35 µg/L Flt-eq. In drinking water, ER and anti-AR activity levels were
decreased in relative abundance during MARR and were of similar below the EBT derived based on human health (Been et al., 2021b).
relative abundance in the resulting drinking water between both pe ER activity decreased during the pretreatment with rapid sand
riods, showing that AOP did not affect the microbial communities of the filtration and was only detected in two out of seven samples after this
resulting drinking water. Whether these taxa concerned DNA of dead or step. Estrogenic activity in AOP effluent was also found in only two out
live cells is not clear. However, AOP did increase AOC concentrations in of seven samples in levels around 0.1 ng/L E2-eq. The low detection
the AOP effluent, which mainly includes carboxylic acids and aldehydes frequency of ER activity before and after AOP made it difficult to esti
(Magic-Knezev et al., 2009). Bacterial isolates belonging to the same mate the impact of this treatment on the removal of estrogenic com
genera as found in this study (i.e. Hydrogenophaga, Sphingomonas and pounds. In Dutch surface waters, natural hormones estrone (E1), E2,
Afipia) were previously collected from activated carbon filters for water estriol (E3), and the synthetic hormone ethinylestradiol (EE) are found
treatment and at least Hydrogenophaga and Afipia prefer carboxylic acids to contribute to the biological response found in the ERα CALUX (Zwart
and/or amino acids as growth substrates (Magic-Knezev et al., 2009). et al., 2020). These compounds have a hydrophobic nature and a
14
relatively high sorption potential and partition to sediment (Lai et al., during treatment. This however needs to be reassessed when AOP
2000), which explained why these were partly removed by rapid sand treatment would be performed on 100% of the water to be infiltrated in
filtration. E1, E2, E3 and EE can be removed by AOP (Silva et al., 2012), MARR systems.
but in this study the compounds causing the ER activity were not always
removed completely by AOP. CRediT authorship contribution statement
Anti-AR responses are known to be caused by a large and diverse
group of compounds. Houtman et al. performed effect directed analysis Peer H.A. Timmers, Tineke Slootweg, Aleksandra Knezev, Astrid
to identify compounds that are responsible for the anti-AR activity Reus and Pieter Stuyfzand analyzed data and wrote the manuscript,
detected in INT (Houtman et al., 2021). Different pesticides were found Dennis Vughs and Leo Heijnen analyzed data, Martin van de Schans,
to contribute to the activity, indicating that mixture toxicity was Peer H.A. Timmers and Luc Zandvliet coordinated sampling, analysis
important for anti-AR activity at this sampling point. The removal of and project deliverables. Ton Knol and Jamal El Majjaoui designed the
anti-AR activity during AOP was consistent with the efficient removal experiments and gave feedback on the manuscript, Paul van de Wielen
that was found for most OMPs during this treatment step. gave feedback on the manuscript and Karin Lekkerkerker-Teunissen
The results of the Ames fluctuation test did not permit conclusions on designed the experiments, gave feedback on the manuscript and coor
the removal or neoformation of potentially mutagenic substances in dinated project deliverables.
some samples after rapid sand filtration and subsequent AOP. It is
known that substances with mutagenic properties can be present in Declaration of Competing Interest
surface waters (Ohe et al., 2004) and the formation of potentially toxic
transformation products during drinking water treatment has also been The authors declare that they have no known competing financial
reported (Parker et al., 2017; Zoeteman et al., 1982). Others observed an interests or personal relationships that could have appeared to influence
increase in mutagenic activity of TA98 after UV/H2O2 oxidation, which the work reported in this paper.
was absent after granular active carbon (GAC) filtration (Heringa et al.,
2011). In the present study, activated carbon filtration after AOP and Acknowledgements
subsequent MARR could have exerted a similar effect. Because the
identity, potency and concentration of the potentially mutagenic sub This work was supported by Dunea, utility for drinking water and
stances in some of the samples were largely unknown, the risk for human nature conservancy. We would like to thank Goffe Elsinga, René van
health and the environment cannot be assessed. Doorn, Kay Hup, Margo Kooi, Claudia Kooijman and Leonie Pap from
Although the Ames fluctuation test showed mutagenic activity in KWR Water Research Institute for sample pre-treatment and perfor
some samples, none of the samples showed direct genotoxicity in the mance of the Ames fluctuation test, DNA extraction and 16S rRNA
p53 CALUX assay. This can be explained by differences in test system sequencing.
(bacteria for the Ames fluctuation test vs. human cell line for the p53
CALUX), mechanism (heritable changes in the DNA vs. induction of a Novelty Statement
signaling cascade to prevent DNA damage) and assay-specific differ
ences in sensitivity and specificity (Pinter et al., 2020). This work describes for the first time the implementation of
Since water is a complex mixture of natural and anthropogenic advanced oxidation processes (AOP) combined with natural dune infil
substances, it is plausible that multiple substances were accountable for tration for organic micropollutant removal during drinking water pro
the observed estrogenic, anti-androgenic and mutagenic activity. duction. The combination of (in)organic compounds, transformation
Chemical analysis demonstrated that AOP and MARR were generally products, biological activity, microbial communities, and toxicity effects
able to remove substances, either completely or partially, which was in that were measured to assess the effects of AOP on water quality is also a
line with the observed decrease in biological effects. With more infor novelty.
mation on chemical identity, it would be possible to identify (groups of) Organic micropollutants can be toxic for ecological systems and
substances that are responsible for specific biological effects. humans. It was therefore decided to investigate pretreatment of source
water with AOP and subsequent dune infiltration, so that most micro
5. Conclusions pollutants and AOP byproducts do not end up in the environment or
drinking water.
Introduction of AOP decreased the number and concentration of
OMP that were introduced during MARR, which decreased the risk of Appendix A. Supporting information
negative impacts on ecology and groundwater. The many uncertainties,
such as seasonal and annual fluctuations of the intake water and in the Supplementary data associated with this article can be found in the
infiltration ponds, made it difficult to determine if factors were directly online version at doi:10.1016/j.jhazmat.2022.128346.
affected by the AOP introduction. However, AOP did increase the
amount of easily biological degradable compounds, but these were References
diluted with non-AOP treated infiltration water and degraded during
transport to the dunes, making the effects on MARR negligible. Several Andersen, K.R.H., K.S, Karst, S.M., Albertsen, M., 2018. Ampvis2: an R package to
analyse and visualise 16S rRNA amplicon data. bioRxiv. https://doi.org/10.1101/
OMP that were not removed during AOP, were removed during MARR. 299537.
The overall OMP concentrations in resulting drinking water therefore Baggelaar, P.K.v.d.M. , Eit, C.J. , 2019. Trendanalist. Gebruikershandleiding, in, AMO -
decreased after introduction of AOP, without showing a measurable Icastat, 2019.
Been, F., Kruve, A., Vughs, D., Meekel, N., Reus, A., Zwartsen, A., Wessel, A., Fischer, A.,
negative effect on other water quality parameters. Another study in ter Laak, T., Brunner, A.M., 2021. Risk-based prioritization of suspects detected in
parallel on the same dune infiltration area revealed that implementation riverine water using complementary chromatographic techniques. Water Res. 204,
of partial AOP had no effects on the Water Framework Directive (Pen 117612.
Been, F., Pronk, T., Louisse, J., Houtman, C., van der Velden-Slootweg, T., van der
ders, 2014). These results showed that partial AOP treatment and MARR
Oost, R., Dingemans, M.M.L., 2021. Development of a framework to derive effect-
were complementary in OMP removal. The combination of partial AOP based trigger values to interpret CALUX data for drinking water quality. Water Res.
treatment and MARR is therefore a promising technology for lowering 193, 116859.
the amount and concentration of OMP, and for producing a stable Bernstein, H.C., Brislawn, C., Renslow, R.S., Dana, K., Morton, B., Lindemann, S.R.,
Song, H.-S., Atci, E., Beyenal, H., Fredrickson, J.K., Jansson, J.K., Moran, J.J., 2017.
drinking water quality without negative effects on inorganic chemistry, Trade-offs between microbiome diversity and productivity in a stratified microbial
toxicology, biological stability, activity and microbial populations mat. ISME J. 11, 405–414.
15
J.E. Bisanz, qiime2R: Importing QIIME2 artifacts and associated data into R sessions, Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B.,
unpublished, (2018). Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater
Bokulich, N.A., Kaehler, B.D., Rideout, J.R., Dillon, M., Bolyen, E., Knight, R., Huttley, G. contaminants in U.S. Streams, 1999− 2000: a national reconnaissance. Environ. Sci.
A., Gregory Caporaso, J., 2018. Optimizing taxonomic classification of marker-gene Technol. 36, 1202–1211.
amplicon sequences with QIIME 2′ s q2-feature-classifier plugin. Microbiome 6, 90. Kooij, Dvd, 1992. Assimilable organic carbon as an indicator of bacterial regrowth.
Bolyen, E., Rideout, J.R., Dillon, M.R., Bokulich, N.A., Abnet, C.C., Al-Ghalith, G.A., J. AWWA 84, 57–65.
Alexander, H., Alm, E.J., Arumugam, M., Asnicar, F., Bai, Y., Bisanz, J.E., Krasner, S.W., Sclimenti, M.J., Coffey, B.M., 1993. Testing biologically active filters for
Bittinger, K., Brejnrod, A., Brislawn, C.J., Brown, C.T., Callahan, B.J., Caraballo- removing aldehydes formed during ozonation. J. Am. Water Works Assoc. 85, 62–71.
Rodríguez, A.M., Chase, J., Cope, E.K., Da Silva, R., Diener, C., Dorrestein, P.C., Lai, K.M., Johnson, K.L., Scrimshaw, M.D., Lester, J.N., 2000. Binding of waterborne
Douglas, G.M., Durall, D.M., Duvallet, C., Edwardson, C.F., Ernst, M., Estaki, M., steroid estrogens to solid phases in river and estuarine systems. Environ. Sci.
Fouquier, J., Gauglitz, J.M., Gibbons, S.M., Gibson, D.L., Gonzalez, A., Gorlick, K., Technol. 34, 3890–3894.
Guo, J., Hillmann, B., Holmes, S., Holste, H., Huttenhower, C., Huttley, G.A., LeChevallier, M.W., Becker, W.C., Schorr, P., Lee, R.G., 1992. Evaluating the
Janssen, S., Jarmusch, A.K., Jiang, L., Kaehler, B.D., Kang, K.B., Keefe, C.R., Keim, P., performance of biologically active rapid filters. J. AWWA 84, 136–146.
Kelley, S.T., Knights, D., Koester, I., Kosciolek, T., Kreps, J., Langille, M.G.I., Lee, J., Lekkerkerker-Teunissen, K., Knol, A.H., van Altena, L.P., Houtman, C.J., Verberk, J.Q.J.
Ley, R., Liu, Y.X., Loftfield, E., Lozupone, C., Maher, M., Marotz, C., Martin, B.D., C., van Dijk, J.C., 2012. Serial ozone/peroxide/low pressure UV treatment for
McDonald, D., McIver, L.J., Melnik, A.V., Metcalf, J.L., Morgan, S.C., Morton, J.T., synergistic and effective organic micropollutant conversion. Sep. Purif. Technol.
Naimey, A.T., Navas-Molina, J.A., Nothias, L.F., Orchanian, S.B., Pearson, T., 100, 22–29.
Peoples, S.L., Petras, D., Preuss, M.L., Pruesse, E., Rasmussen, L.B., Rivers, A., Magic-Knezev, A., Wullings, B., Van der Kooij, D., 2009. Polaromonas and
Robeson 2nd, M.S., Rosenthal, P., Segata, N., Shaffer, M., Shiffer, A., Sinha, R., Hydrogenophaga species are the predominant bacteria cultured from granular
Song, S.J., Spear, J.R., Swafford, A.D., Thompson, L.R., Torres, P.J., Trinh, P., activated carbon filters in water treatment. J. Appl. Microbiol 107, 1457–1467.
Tripathi, A., Turnbaugh, P.J., Ul-Hasan, S., van der Hooft, J.J.J., Vargas, F., Vázquez- Mandal, S., Van Treuren, W., White, R.A., Eggesbø, M., Knight, R., Peddada, S.D., 2015.
Baeza, Y., Vogtmann, E., von Hippel, M., Walters, W., Wan, Y., Wang, M., Warren, J., Analysis of composition of microbiomes: a novel method for studying microbial
Weber, K.C., Williamson, C.H.D., Willis, A.D., Xu, Z.Z., Zaneveld, J.R., Zhang, Y., composition. Microb. Ecol. Health Dis. 26, 27663.
Zhu, Q., Knight, R., Caporaso, J.G., 2019. Reproducible, interactive, scalable and Massmann, G., Greskowiak, J., Dünnbier, U., Zuehlke, S., Knappe, A., Pekdeger, A., 2006.
extensible microbiome data science using QIIME 2. Nat. Biotechnol. 37, 852–857. The impact of variable temperatures on the redox conditions and the behaviour of
Bouwer, H., 2002. Artificial recharge of groundwater: hydrogeology and engineering. pharmaceutical residues during artificial recharge. J. Hydrol. 328, 141–156.
Hydrogeol. J. 10, 121–142. McMurdie, P.J., Holmes, S., 2013. phyloseq: an R package for reproducible interactive
Dillon, P., Stuyfzand, P., Grischek, T., Lluria, M., Pyne, R.D.G., Jain, R.C., Bear, J., analysis and graphics of microbiome census data. PLoS One 8, e61217.
Schwarz, J., Wang, W., Fernandez, E., Stefan, C., Pettenati, M., van der Gun, J., NEN, Soil quality - Sampling of groundwater, NEN, NEN 5744:2011/A1:2013 nl (2013)
Sprenger, C., Massmann, G., Scanlon, B.R., Xanke, J., Jokela, P., Zheng, Y., 6.
Rossetto, R., Shamrukh, M., Pavelic, P., Murray, E., Ross, A., Bonilla Valverde, J.P., Ohe, T., Watanabe, T., Wakabayashi, K., 2004. Mutagens in surface waters: a review.
Palma Nava, A., Ansems, N., Posavec, K., Ha, K., Martin, R., Sapiano, M., 2018. Sixty Mutat. Res. 567, 109–149.
years of global progress in managed aquifer recharge. Hydrogeol. J. 27, 1–30. Parker, A.M., Lester, Y., Spangler, E.K., von Gunten, U., Linden, K.G., 2017. UV/H(2)O
E.C.A. (ECHA), 2021. Registration dossier Bromoform, in, 2021. (2) advanced oxidation for abatement of organophosphorous pesticides and the
Eschauzier, C., Haftka, J., Stuyfzand, P.J., de Voogt, P., 2010. Perfluorinated compounds effects on various toxicity screening assays. Chemosphere 182, 477–482.
in infiltrated river rhine water and infiltrated rainwater in coastal dunes. Environ. Penders, E.J.M. , 2014. Water Framework Directive evaluation of AOP (In Dutch:
Sci. Technol. 44, 7450–7455. KaderRichtlijnWater beoordeling GOBAM studie), projectreport 202104, Het
Gros, M., Petrović, M., Barceló, D., 2009. Tracing pharmaceutical residues of different Waterlaboratorium.
therapeutic classes in environmental waters by using liquid chromatography/ Pinter, E., Rainer, B., Czerny, T., Riegel, E., Schilter, B., Marin-Kuan, M., Tacker, M.,
quadrupole-linear ion trap mass spectrometry and automated library searching. 2020. Evaluation of the suitability of mammalian in vitro assays to assess the
Anal. Chem. 81, 898–912. genotoxic potential of food contact materials. Foods 9, 237.
H.C.E. Environment Canada, 2010. Screening Assessment for the Challenge Bromic acid, Prest, E.I., Hammes, F., Kötzsch, S., van Loosdrecht, M.C., Vrouwenvelder, J.S., 2013.
potassium salt (Potassium bromate) Chemical Abstracts Service Registry Number Monitoring microbiological changes in drinking water systems using a fast and
7758–01-2, in, 2010. reproducible flow cytometric method. Water Res. 47, 7131–7142.
Haag, W.R., Hoigne, J., 1983. Ozonation of bromide-containing waters: kinetics of Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J.,
formation of hypobromous acid and bromate. Environ. Sci. Technol. 17, 261–267. Glöckner, F.O., 2012. The SILVA ribosomal RNA gene database project: improved
Heringa, M.B., Harmsen, D.J., Beerendonk, E.F., Reus, A.A., Krul, C.A., Metz, D.H., data processing and web-based tools. Nucleic Acids Res. 41, D590–D596.
Ijpelaar, G.F., 2011. Formation and removal of genotoxic activity during UV/H(2)O Rahman, M.F., Peldszus, S., Anderson, W.B., 2014. Behaviour and fate of perfluoroalkyl
(2)-GAC treatment of drinking water. Water Res. 45, 366–374. and polyfluoroalkyl substances (PFASs) in drinking water treatment: a review. Water
Hijnen, W.A.M., Schurer, R., Bahlman, J.A., Ketelaars, H.A.M., Italiaander, R., van der Res. 50, 318–340.
Wal, A., van der Wielen, P., 2018. Slowly biodegradable organic compounds impact Ray, C., Soong, T.W., Lian, Y.Q., Roadcap, G.S., 2002. Effect of flood-induced chemical
the biostability of non-chlorinated drinking water produced from surface water. load on filtrate quality at bank filtration sites. J. Hydrol. 266, 235–258.
Water Res. 129, 240–251. Richardson, S.D., Thruston, A.D., Caughran, T.V., Chen, P.H., Collette, T.W., Floyd, T.L.,
Houtman, C.J., 2010. Emerging contaminants in surface waters and their relevance for Schenck, K.M., Lykins, B.W., Sun, G.-r, Majetich, G., 1999. Identification of new
the production of drinking water in Europe. J. Integr. Environ. Sci. 7, 271–295. ozone disinfection byproducts in drinking water. Environ. Sci. Technol. 33,
Houtman, C.J., ten Broek, R., de Jong, K., Pieterse, B., Kroesbergen, J., 2013. 3368–3377.
A multicomponent snapshot of pharmaceuticals and pesticides in the river Meuse Sacher, F., Raue, B., Brauch, H.J., 2005. Analysis of iodinated X-ray contrast agents in
basin. Environ. Toxicol. Chem. 32, 2449–2459. water samples by ion chromatography and inductively-coupled plasma mass
Houtman, C.J., ten Broek, R., Brouwer, A., 2018. Steroid hormonal bioactivities, culprit spectrometry. J. Chromatogr. A 1085, 117–123.
natural and synthetic hormones and other emerging contaminants in waste water Scheurer, M., Brauch, H.-J., Lange, F.T., 2009. Analysis and occurrence of seven artificial
measured using bioassays and UPLC-tQ-MS. Sci. Total Environ. 630, 1492–1501. sweeteners in German waste water and surface water and in soil aquifer treatment
Houtman, C.J., Brewster, K., ten Broek, R., Duijve, B., van Oorschot, Y., Rosielle, M., (SAT). Anal. Bioanal. Chem. 394, 1585–1594.
Lamoree, M.H., Steen, R.J.C.A., 2021. Characterisation of (anti-)progestogenic and Scheurer, M., Storck, F.R., Brauch, H.-J., Lange, F.T., 2010. Performance of conventional
(anti-)androgenic activities in surface and wastewater using high resolution multi-barrier drinking water treatment plants for the removal of four artificial
effectdirected analysis. Environ. Int. 153, 106536. sweeteners. Water Res. 44, 3573–3584.
Huang, G., Ng, T.W., Chen, H., Chow, A.T., Liu, S., Wong, P.K., 2020. Formation of Schriks, M., Heringa, M.B., van der Kooi, M.M., de Voogt, P., van Wezel, A.P., 2010.
assimilable organic carbon (AOC) during drinking water disinfection: a Toxicological relevance of emerging contaminants for drinking water quality. Water
microbiological prospect of disinfection byproducts. Environ. Int. 135, 105389. Res. 44, 461–476.
Huang, W.J., Fang, G.C., Wang, C.C., 2005. The determination and fate of disinfection Segers, W.C.J. , Stuyfzand,P.J., 2007. Appereance and behaviour of emerging substances
by-products from ozonation of polluted raw water. Sci. Total Environ. 345, 261–272. during dune filtration (in Dutch). M.Sc. thesis, VU University Amsterdam., The
Huck, P.M., Fedorak, P.M., Anderson, W.B., 1991. Formation and removal of assimilable Netherlands, (2007).
organic carbon during biological treatment. J. AWWA 83, 69–80. Silva, C.P., Otero, M., Esteves, V., 2012. Processes for the elimination of estrogenic
Ijpelaar, G.F., Harmsen, D.J.H., Beerendonk, E.F., van Leerdam, R.C., Metz, D.H., steroid hormones from water: a review. Environ. Pollut. 165, 38–58.
Knol, A.H., Fulmer, A., Krijnen, S., 2010. Comparison of low pressure and medium Stolker, A.L., Niesing, W., Fuchs, R., Vreeken, R.J., Niessen, W.M., Brinkman, U.A., 2004.
pressure UV Lamps for UV/H2O2 treatment of natural waters containing micro Liquid chromatography with triple-quadrupole and quadrupole-time-of-flight mass
pollutants. Ozone Sci. Eng. 32, 329–337. spectrometry for the determination of micro-constituents - a comparison. Anal.
Jürgens, M.D., Holthaus, K.I., Johnson, A.C., Smith, J.L., Hetheridge, M., Williams, R.J., Bioanal. Chem. 378, 1754–1761.
2002. The potential for estradiol and ethinylestradiol degradation in English rivers. Stuyfzand, P.J. , 1989. Quality changes of river Rhine and Meuse water upon basin
Environ. Toxicol. Chem. 21, 480–488. recharge in The Netherlands’ coastal dunes: 30 years of experience., In: Artificial
Kaehler, B.D., Bokulich, N.A., McDonald, D., Knight, R., Caporaso, J.G., Huttley, G.A., recharge of Groundwater, Proc. Int. Symp. Anaheim USA, 21- 28 august 1988,
2019. Species abundance information improves sequence taxonomy classification Johnson, A.I. & D.J. Finlayson (eds), Am. Soc. Civil Eng., New York, (1989)
accuracy, Nature. Nat. Commun. 10, 4643. 233–245.
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2008. The occurrence of van den Berg, H.H.J.L. , Friederichs, L. , Versteegh, J.F.M. , Smeets, P. , de Roda Husman,
pharmaceuticals, personal care products, endocrine disruptors and illicit drugs in A.M. , 2017. Risicoanalyse en risicomanagement van drinkwaterproductie in
surface water in South Wales, UK. Water Res. 42, 3498–3518. Nederland, in, Rijksinstituut voor Volksgezondheid en Milieu, Bilthoven, 2017, pp.
58.
16
van der Kooij, D., Visser, A., Hijnen, W.A.M., 1982. Determining the concentration of communities in three drinking water distribution systems with variable biological
easily assimilable organic carbon in drinking water. J. AWWA 74, 540–545. stability. Environ. Sci. Technol. 54, 14535–14546.
van der Kooij, D., Hijnen, W.A.M., Kruithof, J.C., 1989. The effects of ozonation, W.H.O. (WHO), 2005a. Bromate in drinking-water. Background document for the
biological filtration and distribution on the concentration of easily Assimilable development of WHO Guidelines for Drinking-water Quality, in, World Health
Organic Carbon (AOC) in drinking water. Ozone Sci. Eng. 11, 297–311. Organization, 2005.
van der Kooij, D., Martijn, B., Schaap, P.G., Hoogenboezem, W., Veenendaal, H.R., van W.H.O. (WHO), 2005b. Trihalomethanes in Drinking-water Background document for
der Wielen, P.W., 2015. Improved biostability assessment of drinking water with a development of WHO Guidelines for Drinking-water Quality, in, World Health
suite of test methods at a water supply treating eutrophic lake water. Water Res 87, Organization, 2005.
347–355. Wang, C., Zheng, Q., Yang, X., 2018. Occurrence and removal of acesulfame and
van der Linden, S.C., Heringa, M.B., Man, H.-Y., Sonneveld, E., Puijker, L.M., sucralose in the drinking water treatment plants along the Yangtze River. Water
Brouwer, A., van der Burg, B., 2008. Detection of multiple hormonal activities in Supply 19, 1305–1312.
wastewater effluents and surface water, using a panel of steroid receptor CALUX Wang, F., van Halem, D., Gang, L., Lekkerkerker-Teunissen, K., van der Hoek, J.P., 2017.
bioassays. Environ. Sci. Technol. 42, 5814–5820. Effect of residual H2O2 from advanced oxidation processes on subsequent biological
van der Linden, S.C., von Bergh, A.R.M., van Vught-Lussenburg, B.M.A., Jonker, L.R.A., water treatment: a laboratory batch study. Chemosphere 185, 637–646.
Teunis, M., Krul, C.A.M., van der Burg, B., 2014. Development of a panel of high- Wang, F., van Halem, D., Ding, L., Bai, Y., Lekkerkerker-Teunissen, K., van der Hoek, J.
throughput reporter-gene assays to detect genotoxicity and oxidative stress. Mutat. P., 2018. Effective removal of bromate in nitrate-reducing anoxic zones during
Res. Genet. Toxicol. Environ. Mutagen. 760, 23–32. managed aquifer recharge for drinking water treatment: laboratory-scale
van der Oost, R., Sileno, G., Suárez-Muñoz, M., Nguyen, M.T., Besselink, H., Brouwer, A., simulations. Water Res. 130, 88–97.
2017. SIMONI (smart integrated monitoring) as a novel bioanalytical strategy for Wang, F., Salgado, V., van der Hoek, J., van Halem, D., 2018. Bromate reduction by Iron
water quality assessment: Part I–model design and effect-based trigger values. (II) during managed aquifer recharge: a laboratory-scale study. Water 10, 370.
Environ. Toxicol. Chem. 36, 2385–2399. Zoeteman, B.C., Hrubec, J., de Greef, E., Kool, H.J., 1982. Mutagenic activity associated
van der Wielen, P.W.J.J., van der Kooij, D., 2010. Effect of water composition, distance with by-products of drinking water disinfection by chlorine, chlorine dioxide, ozone
and season on the adenosine triphosphate concentration in unchlorinated drinking and UV-irradiation. Environ. Health Perspect. 46, 197–205.
water in the Netherlands. Water Res. 44, 4860–4867. Zwart, N., Jonker, W., ten Broek, R., de Boer, J., Somsen, G., Kool, J., Hamers, T.,
Vavourakis, C.D., Heijnen, L., Peters, M.C.F.M., Marang, L., Ketelaars, H.A.M., Hijnen, W. Houtman, C.J., Lamoree, M.H., 2020. Identification of mutagenic and endocrine
A.M., 2020. Spatial and temporal dynamics in attached and suspended bacterial disrupting compounds in surface water and wastewater treatment plant effluents
using high-resolution effect-directed analysis. Water Res. 168, 115204.
17

1 s2.0 S0304389422001340 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0304389422001340 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0304389422001340 Main

Uploaded by

Copyright:

Available Formats

Journal of Hazardous Materials 429 (2022) 128346

Contents lists available at ScienceDirect

Journal of Hazardous Materials

Improved drinking water quality after adding advanced oxidation for

• AOP decreases the number and concen

1. Introduction combination of AOP followed by MARR could have several advantages

Fig. 1. Schematic of the water purification process by

AOP BAL2 BAL1

MW-37 IP-37 IP-38 MW-38

MW-25 IP-25 IP-40 MW-40

cumulative ATP production from day 0–14 (BPC14, d ng ATP/L). 3. Results

#: data errors ##: estimate based on various data

BPC14 (ng ATP/l)

Fig. 7. Principal Coordinates Analysis (PCoA) plot based

Actinobacteria - HGCL clade

Actinobacteria - HGCL clade

You might also like

1 s2.0 S0304389422001340 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0304389422001340 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0304389422001340 Main

Uploaded by

Copyright:

Available Formats

Journal of Hazardous Materials 429 (2022) 128346

Contents lists available at ScienceDirect

Journal of Hazardous Materials

Improved drinking water quality after adding advanced oxidation for

• AOP decreases the number and concen­

1. Introduction combination of AOP followed by MARR could have several advantages

Fig. 1. Schematic of the water purification process by

AOP BAL2 BAL1

MW-37 IP-37 IP-38 MW-38

MW-25 IP-25 IP-40 MW-40

cumulative ATP production from day 0–14 (BPC14, d ng ATP/L). 3. Results

#: data errors ##: estimate based on various data

BPC14 (ng ATP/l)

Fig. 7. Principal Coordinates Analysis (PCoA) plot based

Actinobacteria - HGCL clade

Actinobacteria - HGCL clade

You might also like

• AOP decreases the number and concen