Documente Academic
Documente Profesional
Documente Cultură
Water Research
journal homepage: www.elsevier.com/locate/watres
a r t i c l e i n f o a b s t r a c t
Article history: Glyphosate is the most widely-used pesticide for weed control in agriculture and in urban and residential
Received 25 June 2019 areas. This is the first study to quantify glyphosate and AMPA levels in surface water in Australia from
Received in revised form different land uses. Glyphosate and AMPA was measured in surface water from 10 rural streams, 30
13 September 2019
urban stormwater wetlands and 9 urban streams located in and around the city of Melbourne, Australia
Accepted 27 September 2019
on five occasions between October 2017 and February 2018. Glyphosate and AMPA were present in most
Available online 30 September 2019
of the urban surface water samples. The frequency of detection of glyphosate was 77% in wetlands and
79% in urban streams, whereas it was only detected in 4% of the rural streams. Similarly, AMPA detection
Keywords:
Glyphosate
was 91% in wetlands and 97% in urban streams, whereas it was only present in 6% of the rural stream
Wetlands samples. In both urban streams and wetlands, the highest average glyphosate concentrations occurred in
Stormwater November (1.8 ± 2.2 mg L1). Overall, wetlands and streams associated with urban land use are vulnerable
Streams to glyphosate contamination. These results highlight the importance of screening for contaminants in
Urban land use urban stormwater to identify the source of pollutants that may end up in aquatic ecosystems and the
risks therefrom.
© 2019 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.watres.2019.115139
0043-1354/© 2019 Elsevier Ltd. All rights reserved.
2 E. Okada et al. / Water Research 168 (2020) 115139
Fig. 2. A) Area of study in the State of Victoria, Australia, B) major catchments and streams, C) sampling sites and land use description.
into local streams and rivers. Consequently, surface water in this Lower Yarra, and Bunyip Catchments were sampled (Fig. 2C,
area receives large amounts of urban runeoff, which reduces the Appendix B). New residential developments incorporate storm-
water quality and changes the water flow rates. In general, the water wetland systems built to deliver effective stormwater flow
metropolitan area of the catchment has poor quality vegetation due management, and pollutant reduction, as well as a secondary
to urbanization and industrial impact. In residential areas, the community amenity and habitat outcomes (Melbourne Water,
waterway channels are significantly altered by urbanization (i.e. 2017b); some are also built to provide a source for irrigation wa-
straightening, channeling, and concrete-lining) which reduces the ter for nearby parks and sports fields. These constructed wetlands
amount of riparian vegetation. The sampling sites of the urban help retard the water flow of around 540 GL/year of stormwater
streams were located within sub-catchments that receive mostly that enters natural drainage systems (Allinson et al., 2017).
runoff water from urban areas (see Appendix B; Fig. B1), although Stormwater wetlands are diverse in size and design. They are built
influence from upstream cannot be disregarded, since the water site-specifically to capture and treat the run-off generated by
samples were taken from the stream and not exclusively from run- impervious or hard surfaces, and to reduce nutrient, sediment
off water. loading of natural waterways. In this sense, wetlands are buffer
zones that can intercept and partly retain run-off related pesticides
(Imfeld et al., 2013). The selected sampled wetlands were located
2.1.3. Stormwater wetlands within sub-catchments that receive mostly runoff water from ur-
Thirty stormwater treatment wetlands in urban and peri-urban ban areas (see Appendix B, Fig. B1).
developments located in the Moorabol, Werribee, Maribyrnong,
4 E. Okada et al. / Water Research 168 (2020) 115139
Samples were collected monthly from October 2017 to February 3.1. Glyphosate and AMPA occurrence
2018 using 500 mL polypropylene bottles, rinsed three times with
the sampled water. The temperature, pH, electrical conductivity Glyphosate and AMPA were present in the majority of the urban
(EC) and dissolved oxygen (DO) of each sample was measured in samples from wetlands and streams, but rarely in stream samples
situ with a field meter (556 MPS Multi Probe System, YSI Inc., Yel- taken from rural areas (Fig. 3). In samples from rural streams,
low Springs, OH, USA) (Appendix C, Fig. C1). Samples were refrig- glyphosate and AMPA frequency of detection (FOD) was 4 and
erated immediately after sampling and kept in a cooler for a 10.2%, respectively (Fig. 3). Glyphosate concentrations were
maximum of 6 h until arrival to the laboratory. Then, samples were 0.3 mg L1 and AMPA levels were 0.8 mg L1. On the other hand,
stored at 18 C until analysis. streams and wetlands situated within urban land use had a much
higher FOD of both glyphosate and AMPA compared to that of the
rural landuse water samples. Glyphosate was present in 79% of the
2.3. Glyphosate and AMPA analysis urban stream samples (mean: 1.1 mg L1 range: 0.25e4.8 mg L1)
and in 77% of the stormwater wetland samples (mean: 1.1 mg L1
2.3.1. Chemicals range: 0.25e14.2 mg L1). AMPA was detected in 91% of urban
Glyphosate (>98%), AMPA (99%), isotope-labelled glyphosate streams (mean: 1.6 mg L1 range: 0.25e4.3 mg L1) and 97% of the
(1,2e13C,15N) (>99%), ammonium acetate (NH4Ac; 99.99%), and sampled wetlands (mean: 1.3 mg L1 range: 0.25e10 mg L1). For
ammonium hydroxide (NH4OH; >99%) were purchased from more details on the spatial and temporal variations of glyphosate
Sigma-Aldrich. Ultrapure deionized water (resistivity and AMPA levels see Fig. E1 (Appendix E).
16.7 mU cm1) was obtained from a Milli-Q® water purification The streams in the upper, forested, mountainous area of the
system (Millipore Ltd., Bedford, MA, USA). A standard stock solution Yarra River catchment feed water storages that are used for
of 1000 mg L1 was prepared for each analyte. Standard working drinking water. The sample taken at site S14 is located upstream of
solutions for LC-MS/MS analysis and for samples fortification were the agricultural area in the Maroondah Reservoir Park, a protected
prepared by dilution with ultrapure water. water supply catchment. This sample had no detectable levels of
glyphosate or AMPA throughout the studied period. The Yarra River
2.3.2. LC-MS/MS analysis then flows through the Yarra Valley, characterized by horticultural
Glyphosate and AMPA analyses were carried out using the direct and viticulture production and livestock. Previous studies have
injection method of Okada et al. (2019). The chromatographic shown that the application of glyphosate in agricultural soils can
analysis was performed on an Agilent Technologies 1290 Infinity II cause the non-point source pollution of groundwater, surface wa-
liquid chromatograph (LC) using a Metrosep A Supp 4® 9 mm, ter, and sediments (Okada et al., 2018). However, in this study the
250 4.0 mm ion exchange column (Metrohm, Gladesville, NSW, detection of glyphosate and AMPA in the rural streams was low,
Australia). The target molecules were detected with an Agilent® despite glyphosate being widely used throughout the growing
6495 B Triple Quadrupole mass spectrometer (MS/MS). Target season of intensive cultures such as vine and fruit orchards
compounds were monitored according to retention time and two (Preston, 2018). In this case, glyphosate dispersion into aquatic
ion transitions and quantified using Agilent® MassHunter® environments may have been attenuated by the soil capacity to
Quantitative Analysis software (v. B.08.00). strongly sorb both glyphosate and AMPA (Okada et al., 2016; Sidoli
Procedural blanks that consisted of ultrapure water spiked with et al., 2016). Also, the presence of riparian vegetation (Syversen and
internal standard were used to monitor any background concen- Bechmann, 2004) reduces the probability of glyphosate reaching
tration. In addition, blank injections of ultrapure water were run surface waters. For instance, several studies have found that
after the highest level of the calibration curve and every six samples
to ensure the lack of carryover. For quality assurance, spiked matrix
samples were checked for recovery at the end of every run. The
limit of detection (LOD) was set as the lowest concentration that
could be reliably differentiated from the background levels with a
signal-to-noise ratio 3, was 0.25 mg L1. The method limit of
quantification (LOQ) was 0.5 mg L1. Samples with concentra-
tions < LOD were considered non-detectable and were set to zero.
Samples that had concentrations < LOQ were set to the LOD value in
figures and statistical analyses (Struger et al., 2015). Selected
samples were also analysed by the enzyme linked immunosorbent
assay (ELISA) using a commercial ELISA kit for glyphosate to
compare the results obtained by the LC-MS/MS method (Appendix
D, Fig. D1).
Table 1
Glyphosate and AMPA occurrence in urban surface waters around the world and in this study.
Reference Country Source n Compound FOD (%) Average (mg L1) Max. (mg L1)
Gly: glyphosate, AMPA: aminomethylphosphonic acid, n: total number of samples; FOD: frequency of detection.
6 E. Okada et al. / Water Research 168 (2020) 115139
of wetlands, the highest concentrations were detected in values towards the end of the study (Fig. 5). AMPA levels tended to
November, whereas in urban streams AMPA levels tended to be be higher than glyphosate concentrations, and remained steady
higher in January (Fig. 4). throughout the sampled period (Fig. 5). This can be a result of
No significant differences were found in glyphosate concentra- several factors. First, there are fewer weed controls at the end of the
tions between urban streams and wetlands using the mixed linear dry Victorian summer. Secondly, during the warmer months,
model and there was no interaction between source of sample and glyphosate concentration decreases because biological degradation
month (p > 0.05, data not shown). When glyphosate data is ana- is favoured at higher environmental temperatures (Bento et al.,
lysed altogether there is a significant seasonal variation in which 2016). At the same time, the AMPA that is produced as a degrada-
glyphosate has its highest value at the beginning of spring tion product is more persistent than glyphosate (Wang et al., 2016;
(November) (p < 0.05; Fig. 5). In the case of AMPA, there were no la Cecilia et al., 2018; Sun et al., 2019). Third, glyphosate may also be
significant differences between urban streams and wetlands (data removed from the water column because it is strongly sorbed to
not shown) or between time of sampling (p > 0.05; Fig. 5). -It is sediments (Coupe et al., 2012) and can also undergo fast chemical
important to note that samples were taken during base flow, except degradation (Paudel et al., 2015; la Cecilia and Maggi, 2018). Lastly,
for the samples taken in December, which were taken after a heavy household and industrial detergents, such as diamine tetra-
rainfall (see Appendix A; Fig. A1). In December, most of the streams methylene phosphonate (EDTMP) or diethylenetriamine-penta-
had bankfull flow, bank over flow, or evidence that the bank had methylene phosphonic acid (DTPMP), also degrade into AMPA
overflowed as a consequence of the heavy rainstorm. However, we (Nowack, 2003). In this study, the detection of AMPA in urban
did not find a clear association between levels of glyphosate and streams cannot be attributed to the discharge of wastewater
AMPA and precipitation events. treatment plant effluents since in Melbourne, sewerage system and
In average, the highest glyphosate concentrations were detected stormwater are separated except on rare occasions. However, it is
in November (springtime), which may result from recent herbicide possible that phosphonate detergents end up in the stormwater
applications due to spring weed control in amenity areas, pave- from general household use, e.g. washing down of patios, decks,
ments and gardens. We observed that the urban councils per- cars in driveways etc.
formed herbicide sprayings during spring and summer in parks and
near creeks and riverbanks during our sampling trips. This type of 3.3. Risks
weed management is relevant in the studied area since there are 87
species of declared invasive plants in the State of Victoria, which are Urban wetlands not only act as water retention and depuration
considered noxious weeds (CaLP Act, 1994), of which each Council areas but also offer an appealing environment for diverse biota,
has the responsibility to control and manage. From December to such as endemic birds, frogs, and fish. These constructed aquatic
February, glyphosate concentration decreased, reaching its lowest ecosystems are mainly quiescent water bodies that only have a
major water input during storm events. They are, therefore, a sink
for any chemicals that may be washed away from houses or in-
dustrial developments (Allinson et al., 2015) and can be affected by
the accumulation of pollutants, such as glyphosate. There are
guidelines for the construction of these stormwater wetlands in
Victoria (Melbourne Water, 2017a,b), but not for chemicals in
entrained stormwater other than for total nitrogen and phos-
phorus. The water in and/or extracted from stormwater retention
ponds and wetlands may pose a (real or perceived) risk to soil and
aquatic organisms from chemicals contained therein. In this
context, the concentrations of glyphosate detected in this study
were well below the strictest level of protection for freshwater
species (99% in pristine ecosystems; 370 mg L1) established by the
Australian and New Zealand Guidelines for Fresh and Marine Water
Quality (ANZECC and ARMCANZ, 2000). The Canadian Council of
Ministers of the Environment Water Quality Guidelines has set a
value of 800 mg L1 as the benchmark for protection of aquatic life
from glyphosate (Canadian Council of Ministers of the
Environment, 2012). In the USA, the chronic Aquatic Life Bench-
marks for glyphosate are 25,700 and 49,900 mg L1 for fish and for
aquatic invertebrates, respectively (EPA, 2016). In the EU, there are
no defined environmental quality standards for glyphosate or
AMPA in surface water (Sze k
acs and Darvas, 2018). However, the EU
has a very strict maximum allowed level of glyphosate in drinking
water of 0.1 mg L1 (European Parliament and Council, 1998).
Currently, there is no agreement among countries’ regulation
agencies on the human health-based limits for glyphosate in
drinking water. For example in Australia, the guideline value for
glyphosate in drinking water is 1000 mg L1 (HMRC and NRMMC,
2011), while in the USA the limit is 700 mg L1 (US EPA 2002) and
in Canada the maximum allowed concentration is 280 mg L1
(Guidelines for Canadian Drinking Water Quality, 2014). Although
Fig. 5. Average glyphosate and AMPA variations in urban streams and wetlands, only
the sampled wetlands and streams are not used as a source of
taking into account data from November to February. Error bars represent standard drinking water, none of the samples exceeded the guideline value
error. Different letters indicate significant differences (P < 0.05). established by the Australian Drinking Water Guidelines (HMRC
E. Okada et al. / Water Research 168 (2020) 115139 7