Science of the Total Environment 479–480 (2014) 1–10

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Sulfonylurea herbicides in an agricultural catchment basin and itsadjacent wetland in the St. Lawrence River basin

Yves de Lafontaine a,⁎, Conrad Beauvais a, Allan J. Cessna b,1, Pierre Gagnon a,Christiane Hudon a, Laurier Poissant a

a Environment Canada, Watershed Hydrology and Ecology Research Division, Centre Saint-Laurent, 105 McGill St., Montreal, QC H2Y 2E7, Canadab Agriculture and Agri-Food Canada, Science and Technology Branch, Saskatoon, SK S7N 0X2, Canada

H I G H L I G H T S

• Presence of sulfonylurea herbicides in streams and an adjacent wetland was assessed.• Herbicide presence depended on stream hydrology and water quality characteristics.• Maximum concentrations in streams were above level for aquatic plant toxicity.• Presence of herbicides in wetland poses less toxicological risk to wetland flora.

⁎ Corresponding author.1 Current address: Environment Canada, Watershed H

Division, National Hydrology Research Centre, 11 Innov3H5, Canada.

0048-9697/$ – see front matter. Crown Copyright © 2014http://dx.doi.org/10.1016/j.scitotenv.2014.01.094

a b s t r a c t

a r t i c l e i n f o

Article history:Received 12 December 2013Received in revised form 26 January 2014Accepted 27 January 2014Available online 15 February 2014

Keywords:Sulfonylurea herbicidesSurface watersDischargeWetlandPrecipitation

The use of sulfonylurea herbicides (SU) has increased greater than 100 times over the past 30 years in bothEurope andNorth America. Applied at low rates, their presence, persistence and potential impacts on aquatic eco-systems remain poorly studied. During late-spring to early fall in 2009–2011, concentrations of 9 SU wereassessed in two agricultural streams and their receiving wetland, an enlargement of the St. Lawrence River(Canada). Six SU in concentrations NLOQ (10 ng L−1) were detected in 10% or less of surface water samples.Rimsulfuron was detected each year, sulfosulfuron and nicosulfuron in two years and the others in one yearonly, suggesting that application of specific herbicides varied locally between years. Detection frequency and con-centrations of SU were not significantly associated with total precipitation which occurred 1 to 5 d before sam-pling. Concentrations and fate of SU differed among sites due to differences in stream dynamics and waterquality characteristics. The persistence of SU in catchment basin streams reflected the dissipation effects associ-atedwith streamdischarge.Maximum concentrations of some SU (223 and 148 ng L−1) were occasionally abovethe baseline level (100 ng L−1) for aquatic plant toxicity, implying potential toxic stress to flora in the streams.Substantially lower concentrations (max 55 ng L−1) of SU were noted at the downstream wetland site, likelyas a result from dilution and mixing with St. Lawrence River water, and represent less toxicological risk to thewetlandflora. Sporadic occurrence of SU at low concentrations in air and rain samples indicated that atmosphericdeposition was not an important source of herbicides to the study area.

Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.

1. Introduction

Sulfonylurea herbicides (SU) are selective systemic weed controlproducts most frequently applied postemergence to major crops suchas corn, wheat, barley, canola and potato. Registered uses of SU includearound 30 active ingredients (Russell et al., 2002). Unlike many otherherbicides, SU are phytoactive at low application rates ranging from 2to 40 g active ingredient (a.i.) per hectare (Beyer et al., 1988). The

ydrology and Ecology Researchation Blvd, Saskatoon, SK S7N

Published by Elsevier B.V. All rights

world-wide use of SU has increased over the past two decades from129 tonnes in 1992 to 2135 tonnes in 2011, with major increases inEurope and North America (Food and Agriculture Organization, 2013).In the province of Quebec (Canada), SU sales increased from 275 kga.i. in 1992 to 11976 kg a.i. in 2006 (MDDEP, 2012) but decreased to2928 kg a.i. in 2010, indicating variability in the use of these herbicides.

SU herbicides are weak acids (pKa values ranging between 3 and 5)and are highly water soluble (log Kow b 1). Their half-lives in soil varyfrom 5 to 70 d depending on the herbicide, soil pH and other soil char-acteristics (Cessna et al., 2006; Hollaway et al., 2006). In soil, they aredegraded either by hydrolysis or microbial activity and the degradationproducts of some SU may persist in soil for years (Rosenbom et al.,2010). Given their field half-lives and relatively high water solubility(Table 1), SU can leach to groundwater as well as enter surface waters

reserved.

Table 1Physical–chemical characteristics of the sulfonylurea herbicides analyzed in this study.

Herbicide Water solubility(mg L−1 @ pH 7,25 °C)Tomlin (2006)

Field half-life(day)University of Hertfordshire (2013)

Field half-life(day)Sanseman (2007)

Field half-life(day)Tomlin (2006)

Vapor pressure(mPa × 105

@ 25 °C)Sanseman (2007)

Henry's Lawconstant(Pa m3 mol−1

@pH 7, 25 °C)Tomlin (2006)

Soil sorptioncoefficientKoc (mL g−1)Sanseman (2007)

Ethametsulfuron-methyl 50 35–70 – 63 0.000077 6.34 × 10−12 (calc) –

Metsulfuron-methyl 2790 10 30 52 0.033 4.5 × 10−11 35Nicosulfuron 70 19–26 21 24–43 0.0000016 b1.48 × 10−11 29–79Primisulfuron-methyl 390 17–30 30 4–29 0.51 – 4–20Prosulfuron 4000 16 9–20 5–23 0.35 – 18–41Rimsulfuron 7300 10–24 2–5 – 150 – –

Sulfosulfuron 1627 24 14–75 11–47 8.8 8.83 × 10−9 5–89Thifensulfuron-methyl 2240 4–10 2–6 4–7 1.7 9.7 × 10−16 45Tribenuron-methyl 2040 10–14 10 3–5 5.2 1.08 × 10−8 46

2 Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

of streams and rivers via soil erosion and surface runoff (Almvik et al.,2011; Battaglin et al., 2000; Cessna et al., 2006; Kreuger and Adielson,2008; Struger et al., 2011). Cessna et al. (2006) reported dissipationhalf-lives of 16 to 84 d for 3 SU in weakly alkaline waters (pH = 8.0–8.7). Consequently, SU are of concern for their potential impacts onaquatic ecosystems, as suggested from laboratory exposure experi-mentswith duckweed (Lemnaminor) (Fairchild et al., 1997), sago pond-weed (Potamogeton pectinatus) (Coyner et al., 2001) and submergedEurasian water-milfoil (Myriophyllum spicatum) (Michael, 2003). SUcan also be toxic to micro-algae including cyanobacteria and dinoflagel-lates (Nyström et al., 1999) and reported EC50 values suggest a toxicitythreshold of around 100 ng L−1 for various algal species (Battaglin et al.,2000). By contrast, aquatic flora would exhibit toxic impact only whenexposed to very high SU concentrations (N100,000 ng L−1) (Michael,2003). Although widely used for nearly three decades, the presence,concentrations and distribution dynamics of SU remain poorly studiedfor surface waters of major agricultural watersheds.

One such watershed is the Yamaska River watershed (4898 km2) inthe province of Quebec, Canada which has the largest cultivated area(2230km2) and accounts for 20%of all agricultural activities in theprov-ince (Poissant et al., 2008a; Roy, 2002). This major agricultural basindrains into the southwestern sector of Lake Saint-Pierre, an enlarge-ment of the St. Lawrence River, which supports large emergentmarshesand extensive macrophyte beds covering 260 km2 (85%) of the lakesurface area (Hudon and Carignan, 2008 990/id). The objective of thepresent study was to assess over three growing seasons (2009 to2011) the presence and the spatio-temporal variation in SU concentra-tions in surface waters of the Baie Saint-François (BSF; 39.8 km2) catch-ment basin, a sub-watershed within the Yamaska River watershed.Surface water samples were collected from late spring through earlyfall (late April to September) incorporating the time of pesticide appli-cation. In addition, SU concentrations in air and precipitationweremea-sured to investigate the possibility of atmospheric deposition of SU tothe BSF catchment basin. The results from this study would providedata necessary to assess whether these herbicides occurred in surfacewaters at sufficiently high concentrations to exert detrimental effectson non-target aquatic organisms.

2. Materials and methods

2.1. Study area

Located between the Yamaska River and the Saint-François River,the Baie Saint-François (BSF) catchment basin (Fig. 1) has been exten-sively used in recent years for corn (59%), soya (13%) andhay (15%) pro-duction (Poissant et al., 2008a). The catchment basin streams drain intoa major fluvial wetland (9.5 km2 — adjacent to the mouths of theYamaska and Saint-François Rivers) which is part of the Lake Saint-Pierre wetland ecosystem (112 km2). Estimated pesticide loads appliedto the BSF catchment basin totaled 2.7 metric tons in 2006, of which

0.41% represented SU, due, in part, to the low SU application rates ratherthan to the extent of land application (Poissant et al., 2008b). Based onsales in 2006, nicosulfuron, rimsulfuron, primisulfuron and prosulfuronwere the top four SU used in corn culturewithin the entire Yamaskawa-tershed (Poissant et al., 2008b). However, details on application timesand applied rates are not compiled in Quebec; thus, quantifying localor regional sources of SU over time and space was not possible.

2.2. Study sampling sites and environmental conditions

Information on the top soil characteristics, vegetation cover and gen-eral environment of the BSF catchment basin was summarized in a pre-vious study (Poissant et al., 2008b). Surfacewater sampling was carriedout at three sites (Fig. 1). The sites on the “Bois-de-Maska” and“Castorerie” streams were situated in the agricultural sector of thecatchment basin used mainly for corn production. The Bois-de-Maskasite was near a forested area whereas the Castorerie site was proximalto large open fields. This latter site wasmoved upstream approximately1.2 km between 2010 and 2011 so that it was situated in closer proxim-ity to cultivated fields. Both streams flow directly into the BSF wetland.The “BSF” site was positioned at the wetland outlet near the St. Law-rence River and downstream of the catchment basin. This site wasmore than 5 kmaway from the nearest agriculturalfield andwas, there-fore, not directly influenced by pesticide applications. Except for theCastorerie site which was not sampled in 2009, sites were sampled forthree consecutive years: 2009, 2010 and 2011. In 2009, sampling wasconducted bi-weekly between 24 April and 12 June and then monthlyuntil 10 September (n = 8). In 2010, monthly samples were taken be-tween 30 April and 18 August (n = 5). In 2011, weekly sampling wasfrom 5May until 27 July (n= 13). At each sampling in each year, dupli-cate water samples were collected at 0.1- to 0.2-m depth in 1-L amberglass bottles rinsed three times with stream water and the bottlescapped with Teflon-lined caps. Each year, samples, preserved at 4 °Cin the dark until analysis (1 to 4 mo), were accumulated and then ana-lyzed as single batches.

In order to determine whether rainfall washout of the SU from theatmosphere may have contributed to concentrations detected in thesurface water samples, time-integrated samples of air and precipitationwere collected once in 2009 (11 June, air sample only) andon amonthlybasis in 2010 (n=4; 11 June, 9 July, 6 and 30 August) and 2011 (n=3;7 and 28 July and 2 September) at a weather station (WS) located in theBSF wetland (Fig. 1). Air samples were collected using a high-volumesampler (Model PS-1, General Metal Works, Village of Cleves, OH)installed at 1.5 m above the ground. Air was pumped continuously(flow rate of ~300 m3 d−1) for 1 mo through a glass fiber filter for par-ticle collection and subsequently through a polyurethane foam (PUF)/XAD-2 resin/PUF cartridge for vapor collection. Rain samples werecollected by an automated Meteorological Instruments of Canada (MICtype B) wet-only precipitation sampler in which water passed throughXAD-2 resin contained in a column made of Teflon. Each rain sample

Fig. 1.Map of the Baie Saint-François catchment basin and its adjacentwetlandwith location of the three sampling sites and theweather station (WS) (upper left) and photographs of thethree sampling sites: Castorerie (upper right), Bois-de-Maska (bottom left) and BSF (bottom right).

3Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

was an accumulation of all precipitation events that occurred during a1-mo period. The glass fiber filters, PUF/XAD-2 resin/PUF cartridgesand the XAD-2 columns were preserved at 4 °C until analysis.

2.3. Laboratory analyses

The nine SU monitored in the water, air and precipitation sampleswere selected because they show a range in physical–chemical proper-ties (Table 1) and could have been used in the BSF catchment basinduring the current study (Poissant et al., 2008b). Analytical standards(98 to 99% purity) of all SU were obtained from Chem Service, WestChester, PA.

All analyses were performed at the National Hydrology Research Cen-ter in Saskatoon (Canada). Water samples were analyzed following ex-traction and analytical procedures described elsewhere (Donald et al.,2007). Eachwater sample (500mL)was passed through a preconditioned[methanol (10 mL) followed by de-ionized water (10 mL)] Oasis hydro-philic–lipophilic resin cartridge (Waters Corporation, Milford, MA)under vacuum. After drying under vacuum, the cartridge was elutedwith methanol (10 mL), the eluate evaporated to dryness using a streamof dry nitrogen gas, and the resulting extract residue dissolved in deion-ized water (1 mL) and transferred to a 2-mL high performance liquidchromatography (HPLC) vial. All replicate water samples were analyzed.

The PUF/XAD-2 resin/PUF cartridges, glass fiber filters, and XAD-2resin from the Teflon columns were extracted by immersing overnightin methanol (250, 150 and 150 mL, respectively) and then filteringinto a 500-mL evaporating flask. Each matrix was re-immersed inmethanol (100 mL) and filtered into its respective evaporating flask.Each combined extract was then concentrated to ~20mL using a rotaryevaporator. The concentrated extracts were transferred to a 15-mL cen-trifuge tube, taken to dryness using a gentle steam of nitrogen gas, and

the extract residue dissolved in deionized water (1mL) and transferredto a 2-mL HPLC vial.

Separation and quantification of the SUwas achieved using aWaters2695 Alliance HPLC system using a Waters Xterra Mass C18 analyticalcolumn interfaced with a Waters Micromass Quattro Ultima triplequadrupole mass spectrometer equipped with an electrospray ioniza-tion interface set to positive ion mode (Waters Limited, Mississauga,ON). None of the SU was available as its isotopically labeled analog soenhancement/suppression of analyte ionization due to matrix effectswithin the ion source of themass spectrometerwere notmonitored. Re-coveries of herbicides fromwater samples fortified at 100 ng L−1 variedbetween 81 and 91% except for tribenuron-methyl for which the aver-age recovery was 43%. The method detection limit (MDL) of each SUwas of the order of 1 ng L−1 and the limit of quantification (LOQ) was10 ng L−1. Values between MDL and LOQ were defined as trace quanti-ties (b10 ng L−1). SU concentrations in air (PUF and filter) and rainsamples were expressed in ng m−3 and ng L−1, respectively. Based ona limit of quantification of 2.5 ng per PUF and filter sample extract, thelimits of quantification for PUF and filter samples were ~0.3 pg m−3.

In addition, temperature, specific conductivity, pH, dissolved oxygen(DO) and percent oxygen saturation (%O2) of surface waters weremeasured during each sample collection using an Hydrolab DS-5Xmultisensor probe. Water samples were also taken for laboratoryanalyses of major ions (Na+, K+, Mg++ and Ca++), dissolved inorganiccarbon (DIC), dissolved organic carbon (DOC), total phosphorus (TP),nitrite–nitrate ions (NO2–NO3), ammonium ion (NH4

+), apparent colorand chlorophyll-a (Chl). Due to equipment failure, precipitation recordsat the weather recording station (WS) were incomplete for 2009 andpart of 2010, so records of daily precipitation and wind direction foreach sampling year were obtained from the nearest weather station(Nicolet, Environment Canada #7025442) located 25 km downwind

4 Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

(in alignmentwith the prevailingwind direction) of the study area. Sta-tistical comparison of precipitation records between these two sites for2010 and 2011 revealed no significant difference in temporal distribu-tion series (K-S test, p= 0.15, n= 361) and the difference in daily pre-cipitation was unbiased (Wilcoxon test, p = 0.9) with a mediandifference of 1.3 mm. The Nicolet station was therefore taken as provid-ing weather conditions representative of our study location.

2.4. Statistical analyses

Chemical concentrations were averaged for the duplicate samples.Trace concentrations (b10 ng L−1) of SU were replaced with half theLOQ for cases in which quantifiable concentrations were detected inthe other replicate. Surface water properties were compared at thethree sites with univariate and multivariate tests. Untransformed mea-sures were compared with Kruskal–Wallis tests while mixed modelswith sampling dates as random effects were fitted to log-transformedvariables (when needed to improve normality). Both stepwise discrim-inant analyses and partition analyses of surface water properties wereused to test differences between sites and identify themost discriminat-ing factors. Spearman correlationswere calculated to reveal associationsamongwater quality variables and also betweenwater quality variablesand precipitation accumulated over 1 to 5 d before sampling. Parametervariability at the three sites was compared with Brown & Forsythe testson transformed variables. All tests were done at the 0.05 significancelevel and statistical analyses were performed with SAS, version 9.3(SAS Institute Inc., Cary, NC, USA) and JMP, version 9 (SAS Institute Inc.).

3. Results and discussion

3.1. Environmental conditions

Cumulative precipitation summed from 15 April to 15 Septembereach year showed that 2010 was drier than the two other years, mostlydue to low precipitation in May (Fig. 2). Summer 2009 was particularlywet with frequent rain events (up to 45 mm d−1) from mid-June untilAugust. Precipitation in 2011 was characterized by frequent but low(max 20 mm d−1) rainfall throughout spring and summer except fortwo major rain events (N40 mm d−1) at the end of August, whichmade cumulative precipitation highest during that year.

Surface water properties varied between sites (Fig. 3) as well asseasonally at each site (Fig. 4) with temporal variation being morepronounced at the two catchment basin sites (Bois-de-Maska andCastorerie) than at the wetland site (BSF). Several variables exhibitedvery similar patterns of variation across sites as revealed by the correla-tion analyses (not shown). Significant positive associations were found

Month

May June July August September

Cu

mu

lati

ve p

reci

pit

atio

n (

mm

)

0

100

200

300

400

500

600

700200920102011

Fig. 2. Cumulative precipitation (mm) at the Nicoletweather station between 15 April and15 September during each sampling year.

between conductivity, Na+, Ca++, Mg++ and DIC (rs = 0.7–0.9,p b 0.001) and K+ (rs = 0.4–0.5, p b 0.01), and between DOC andTP (rs = 0.45, p b 0.01) which were both negatively correlated to %O2

(rs = −0.45, p b 0.01). Apparent color, NH4+ and TP were significantly

associated with each other (rs = 0.55–0.68, p b 0.001). DIC wasweaklycorrelated to NH4

+ (rs = 0.37, p b 0.01), but not to DOC (rs =−0.10, pN 0.05). Chlorophyll and %O2 were positively associated (rs = 0.39, pb 0.01) with each other and were both negatively related to tempera-ture (rs = −0.40 and −0.30 respectively, p b 0.01), mostly due to aninverse seasonal pattern of variation between these variables. FinallypH was not significantly correlated to any variable, except for Ca++

and K+ (rs = 0.49 and 0.39 respectively, p b 0.01).Globally, results from discriminant analysis and partition analyses

(not shown) revealed that the three sites were significantly distinctdue to differences in major ions (Na+, Mg++, Ca++, K+), conductivity,DOC, TP, %O2, NH4

+ and apparent color. Water quality parameters [lowspecific conductivity, lowDIC, lowNH4

+ and slightly higher temperature(mixed model F-test, p b 0.001); Figs. 3 and 4] indicated that surfacewater at the BSF site was markedly influenced by and more typicallyrepresentative of the St. Lawrence River water mass. The BSF wetlandsite was characterized by a progressive seasonal drop in water depth(from ~2 m to 0.5 m) related to water level fluctuations in the St.Lawrence River. By contrast, water quality measurements at theshallower Castorerie and Bois-de-Maska streams were indicative ofstrong agricultural influence: much higher specific conductivity andDIC, and concentrations of associated major ions (Na+, Mg++, Ca++)(mixed model F-test, p b 0.001). Although TP concentrations werehigh at all locations and did not significantly differ between sites(mixed model F-test, p = 0.23), the two streams exhibited clear differ-ences. The Castorerie site was characterized by slightly deeper andconsistently oxygen supersaturated water and by significantly higherNO2–NO3 concentrations (mixed model F-test, p b 0.001) (Figs. 3 & 4)than the Bois-de-Maska site. The Castorerie site was also subject torapid fluctuations in water depth following important rainfall events,which induced decreased specific conductivity and DIC, and increasedNO2–NO3 and DOC levels (Fig. 4). By contrast, the Bois-de-Maska sitewas hydrologically less dynamic with a relatively constant waterdepth (10 to 15 cm) with very frequent oxygen non-saturated condi-tions and high concentrations of NH4

+ andDOC, all indicative of relative-ly stagnant waters. The pH of surface water varied between 6.5 and 8.0,and was significantly (p b 0.001) more alkaline at the Castorerie sitethan at the two other sites which were slightly acidic. Except for highervalues at the BSF site in spring 2009 and 2010, chlorophyll levelswere not systematically different between sites (mixed model F-test,p= 0.68), indicating no consistent spatial differences in phytoplanktonbiomass during the summer months.

Spearman correlation analyses between water properties andcumulative precipitation for up to 5 d before sampling revealed thatmajor rainfall events during the 2 d prior to sampling coincided with asignificant (p b 0.005) decrease in water conductivity, associatedmajor ions and inorganic carbon content, but with a rise in apparentcolor (p b 0.05). Overall, water quality properties clearly showed thatthe BSF site was more representative of St. Lawrence River water andmarkedly different from the two other sites where differences inwater characteristicswere due to local variation in agricultural activitiesand stream hydrological dynamics.

3.2. Sulfonylurea herbicides in atmospheric samples

All of the nine monitored SU are characterized by relatively lowvapor pressures (with the exception of rimsulfuron), weak sorption tosoil and high water solubility (Table 1). Low vapor pressure togetherwith high water solubility results in relatively low Henry's Lawconstants (Table 1) which would indicate minimal vapor loss to the at-mosphere. Thus, when present in the atmosphere, SUwouldmost likely

Wat

er d

epth

(m

)0.0

0.5

1.0

1.5

2.0

2.5

Co

nd

uct

ivit

y (m

S c

m-1

)

0

200

400

600

800

1000

1200D

IC (

mg

-C L

-1)

10

20

30

40

50

60

70

NO

2-N

O3

(mg

-N L

-1)

0

2

4

6

8

10

BSF Bois-Maska Castorerie

NH

4+ (m

g-N

L-1

)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

O2

satu

rati

on

(%

)

0

40

80

120

160

200

pH

5.5

6.0

6.5

7.0

7.5

8.0

8.5

DO

C (

mg

-C L

-1)

4

6

8

10

12

14

16

BSF Bois-Maska Castorerie

Ch

loro

ph

yll-

a (µ

g L

-1)

0

10

20304050

Tem

per

atu

re (

oC

)

5

10

15

20

25

30

Fig. 3. Physical and chemical characteristics of surfacewater at the three sampling sites in the Baie Saint-François catchment basin during 2009–2011. Box andwhiskers plots represent the25th percentile, the median and the 75th percentile and the error bars indicate the 10th and 90th percentiles with outlying data shown as single data points.

5Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

be associatedwith particles, due either to desiccation of application driftdroplets or to wind erosion of treated soil.

No SU were detected in air (PUF, filter) or rain samples collected in2009 and 2010. In 2011, SUwere detected in only one of three rain sam-ples; that collected on July 28 (Table 2). Of the six SU detected, the threewith relatively higher vapor pressures (sulfosulfuron, rimsulfuron andthifensulfuron-methyl; Table 1) were also detected in all three air sam-ples. Sulfosulfuron was present in the atmosphere only as a vapor,thifensulfuron-methyl as a particulate, and rimsulfuron as vapor andparticulate. The concentrations of SU detected over the BSF wetland inthe air samples were b2 pg m−3 whereas those in rain were b2 ng L−1

(Table 2). The low and infrequently detected SU concentrations mea-sured in the air and rain samples strongly led us to conclude that atmo-spheric input did not represent an important source of SU into the BSFcatchment basin and its adjacent wetland.

3.3. Sulfonylurea herbicides in surface water samples

This study confirmed that, at least once over the three years of study,six of the nine monitored SU were present in two streams draining

agricultural fields within the Baie Saint-François catchment basin(Table 3). In addition to rimsulfuron, nicosulfuron and prosulfuron,which were previously reported to be present in the Yamaska River(Struger et al., 2011) and in the Chibouet River, a tributary of the upperYamaska River (Giroux and Pelletier, 2012) (Table 4), the occurrence ofethametsulfuron-methyl, metsulfuron-methyl and sulfosulfuron wasquantified for the first time. Rimsulfuron was detected in the streamsevery year, sulfosulfuron and nicosulfuron in two years, and the othersin one year only. When all stream samples containing quantifiableherbicide concentrations (NLOQ) were pooled, detection frequency was10% for rimsulfuron and metsulfuron-methyl and 8.6% and 7.1% forsulfosulfuron and nicosulfuron, respectively. When trace level concentra-tions were included, frequency of occurrence increased to 31.9% forsulfosulfuron and 24.3% for rimsulfuron but was b20% for the other fourherbicides. The other three monitored SU (primisulfuron, thifensulfuron-methyl and tribenuron-methyl) were not detected in any year.

Detection frequencies and concentrations of SU reported in Ontarioand Quebec streams (Struger et al., 2011), Quebec rivers including theChibouet River (Giroux and Pelletier, 2012), the midwestern UnitedStates (Battaglin et al., 2000) and surface waters in Sweden (Kreuger

Wat

er d

epth

(m

)0.0

0.5

1.0

1.5

2.0

2.5BSF Bois-de-Maska Castorie

Tem

per

atu

re (

oC

)

8

12

16

20

24

pH

6.0

6.5

7.0

7.5

8.0

NO

2- N

O3

(mg

-N L

-1)

0

2

4

6

8

10

Sampling Date

May 5

May 11

May19

May 25

June 2

June 8

June 1

5

June 2

0

June 2

9

July

4

July

13

July

20

July

27

NH

4+ (m

g-N

L-1

)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

DIC

(m

g-C

L-1

)

10

20

30

40

50

60

70

DO

C (

mg

-C L

-1)

4

6

8

10

12

14

16

Sampling Date

May 5

May 11

May19

May 25

June 2

June 8

June 1

5

June 2

0

June 2

9

July

4

July

13

July

20

July

27

Ch

loro

ph

yll-

a (µ

g L

-1)

0

5

10

15

20

Co

nd

uct

ivit

y (µ

S c

m-1

)

200

300

400

500

600

700

O2

satu

rati

on

(%

)

0

40

80

120

160

200

Fig. 4. Temporal variation in surface water characteristics at the three sampling sites in 2011.

6 Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

and Adielson, 2008) (Table 4), were similar to the low detection fre-quencies (b10%) and low concentrations (ng L−1) for the SU detectedin the current study. Low detection frequencies and concentrations in

Table 2Sulfonylurea herbicide concentrations in the air (PUF,filter; pgm−3) and rain (XAD resin; ng L−

in 2009 and 2010. Numbers in italics identify values below limit of detection either from air or

Date Sample Ethametsulfuron-methyl

Metsulfuron-methyl

Nicosulfuron Sulfosulfu

July 7, 2011 PUF – – – 1.8FILTER – – – –

XAD resin – – – –

July 28, 2011 PUF – – – b1FILTER – – – –

XAD resin 1.4 1.4 b0.2 0.6Sept 2, 2011 PUF – – – b1

FILTER – – – –

XAD resin – – – –

surfacewaters are generally characteristic of SU because they are usual-ly applied only once during the growing season and at low rates (12 to25 g a.i. ha−1).

1) samples during2011. Sulfonylureaherbicideswere not detected in any air or rain samplewater samples.

ron Primisulfuronn-methyl

Prosulfuron Rimsulfuron Thifensulfuron-methyl

Tribenuron-methyl

– – 1.4 – –

– – 1.5 – –

– – – – –

– – b1 – –

– – – – –

– – b0.2 1.2 –

– – – – –

– – b1 1.1 –

– – – – –

Table 3Detection frequency and maximum concentration of sulfonylurea herbicides at the 3 sampling sites. Detection frequency values for each herbicide indicate number of detects (NLOQ = 10 ng L−1) and values in parentheses are number ofobservations at trace levels (1–10 ng L−1). Detection frequencies for all sites are expressed in percentage. Italicized entries identfiy values at trace concentrations (i.e. values below 10 ng per liter).

Sampling sites Years # samples Ethametsulfuron-methyl Metsulfuron-methyl Nicosulfuron Primisulfuron-methyl Prosulfuron Rimsulfuron Sulfosulfuron Thifensulfuron-methyl Tribenuron-methyl

Castorerie 2010–2011 18 3/(3) 6/(2) 0 0 1/(0) 3/(2) 0/(3) 0 0Max concentration 148 48 20 27 7

Bois-de-Maska 2009–2011 26 0/(3) 1/(1) 4/(0) 0 0/(1) 4/(4) 4/(6) 0 0Max concentration 8 40 84 10 223 28

BSF 2009–2011 26 0/(3) 0/(3) 2/(1) 0 0 0/(4) 1/(8) 0 0Max concentration 9 5 55 6 17

All sites (NLOQ) 70 4.3% 10.0% 8.6% 0% 1.4% 10.0% 7.1% 0% 0%(NMDL + NLOQ) 70 17.1% 18.6% 10.0% 0% 2.8% 24.3% 31.4% 0% 0%

Table 4Comparison of maximum concentrations of sulfonylurea herbicides (ng L−1) measured in agricultural streams and rivers from various studies.

# samples Ethametsulfuron-methyl Metsulfuron-methyl Nicosulfuron Primisulfuron-methyl Prosulfuron Rimsulfuron Sulfosulfuron Thifensulfuron-methyl Tribenuron-methyl

Baie Saint-François (3 sites)[2009–2011] (this study)

70 148 48 84 – 20 223 28 – –

Quebec Rivers [2008–2010],Giroux and Pelletier (2012)Chibouet R. 114 78 100Des Hurons R. 114 96 61St. Regis R. 107 25 91St. Zephirin R. 118 35 160

SW Ontario streams [2006–2008],Struger et al. (2011)

57 525 157 4 145 47

Swedish streams [1998–2006],Kreuger and Adielson (2008)

536 100 210 180 120

US Midwest streams [1998],Battaglin et al. (2000)

130 b10 266 b10 36 15

7Y.de

Lafontaineetal./Science

oftheTotalEnvironm

ent479–480

(2014)1–10

8 Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

Generally, SU were detected in the stream and wetland (BSF) waterfrom May to August and were detected more frequently and at higherconcentrations at the Bois-de-Maska and Castorerie sites than at theBSF site near thewetland outlet (Figs. 5 and 6). The seasonal occurrenceand the concentrations of SU between siteswere not significantly corre-lated and no significant correlation was found between pairs of herbi-cides, indicating no strong association in local concentrations of thevarious herbicides. Although high concentrations of someSUwere occa-sionally reported after rainfall events (such as Aug. 18, 2010 — Fig. 5;May 5 & July 27, 2011 — Fig. 6), neither the frequency of detection northe concentrations of the SU in surface waters at each site was signifi-cantly associatedwith total precipitationwhich occurred 1 to 5 d beforesampling.

Herbicide concentrations at the two stream sites generally decreasedover time probably as a result of various dissipation processes includingboth destructive (hydrolysis, microbial degradation, photodegradation)and non-destructive (water transport) mechanisms. Hydrolysis mayhave contributedmore to the decreased SU concentrations in the streamwater passing the Bois-de-Maska sampling site due tomore acidic water(pH= 6.8–7.4— Fig. 3), which would enhance hydrolysis (Dinelli et al.,1997), versus that at the Castorerie site (pH= 7.1 to 7.6). However, thedecrease in SU concentrations with time at these sites may be related todifferences in local hydrology and flushing times of stream water. Thespatial differences in water quality characteristics between the twostreams (Figs. 3 & 4) revealed that the Castorerie streamwas hydrologi-cally more dynamic with higher andmore variable flow relative to morestagnant conditions in the Bois-de-Maska stream. Thismay have contrib-uted to a pattern of decreased SU concentrations with time at theCastorerie site versus the more static concentrations observed at theBois-de-Maska site (Figs. 5 & 6). The discharge of these streams didvary over time in response to rainfall events during the season, assuggested by the significant correlation between total precipitation andsurface water characteristics (conductivity, major ions and DIC). The ob-servation thatmaximumherbicide concentrationswere observed during

Sampling date

4/24/2009

5/8/2009

5/22/2009

5/29/2009

6/12/2009

7/17/2009

8/14/2009

9/10/2009

Pre

cip

itat

ion

(m

m)

0

10

20

30

40

Rim

sulf

uro

n (

ng

L-1

)

1

10

100

Nic

osu

lfu

ron

(n

g L

-1)

1

10

100BSFBois-de-Maska

2009

Fig. 5. Temporal variation in the occurrence and concentrations (ng L−1) of sulfonylurea herbicisampling in each year. Horizontal dotted lines indicate the LOQ. The Castorerie site was not sa

the spraying season and after some rainfall events (Figs. 5 & 6) suggeststhat the interval between herbicide application and significant rainfallscan contribute to important temporal variation in the occurrence andconcentrations of herbicides within a catchment basin (Kreuger, 1998;Kreuger and Adielson, 2008).

3.4. Temporal variation in SU concentrations

The occurrence of herbicides in surfacewaters is primarily a functionof application rate, time of application, field half-life, the soil leachingproperties and the persistence of herbicides in water (Cessna et al.,2001, 2006). Because the application times and rates of the various SUused in Quebec are not compiled, the variation in the sources of SUover space and time within the BSF catchment basin are unknown forthe three years of our study. The difference in the occurrence and levelsof SU in the Bois-de-Maska and Castorerie streams strongly suggested,however, that therewas year-to-year variation in SUused on agricultur-al fields that drained into these two streams. For example, nicosulfuron,whichwas frequently detected at the Bois-de-Maska site but not detect-ed at the Castorerie site in 2009, was detected more frequently at theCastorerie site in 2010 but was virtually absent from both sites in2011 (Figs. 5 & 6). As expected, the herbicide was detected at the BSFsite in both 2009 and 2010 when it was detected at one or both of thestream sites, but not in 2011 when it was not detected at either streamsite. Similarly, ethametsulfuron-methyl which occurred at both sites in2011, but was never detected at either site during the two previousyears, was detected at the BSF site only in 2011. Such year-to-year var-iation in SU present in the surfacewaters of the BSF catchment basin hasalso been observed in Ontario streams (Struger et al., 2011) and theChibouet River (Giroux and Pelletier, 2012) and can be best explainedby year-to-year variation in the use of specific SU in the respectivecatchment basins of these streams and rivers. This variation impliesthat local sources of annually used SU, rather than long-term use of SU

Sampling date

4/30/2010

5/17/2010

6/7/2010

7/7/2010

8/18/2010

Pre

cip

itat

ion

(m

m)

0

10

20

30

40

Su

lfo

sulf

uro

n (

ng

L-1

)

1

10

100

BSFBois-de-MaskaCastorerie

Rim

sulf

uro

n (

ng

L-1

)

1

10

100

2010

des at each site in 2009 and 2010 and precipitation (mm) cumulated over two days beforempled in 2009.

Sampling date

5/5/2011

5/11/2011

5/19/2011

5/25/2011

6/2/2011

6/8/2011

6/15/2011

6/20/2011

6/29/2011

7/4/2011

7/13/2011

7/20/2011

7/27/2011

Pre

cip

itat

ion

0

10

20

30

40

50

Rim

sulf

uro

n

1

10

100

Su

lfo

sulf

uro

n

1

10

100

Met

sulf

uro

n

1

10

100

Eth

amet

sulf

uro

n

1

10

100 BSFBois-de-MaskaCastorerie

Pro

sulf

uro

n

1

10

100

Fig. 6. Temporal variation in the occurrence and concentrations (ng L−1) of sulfonylureaherbicides at each sampling site in 2011 and precipitation (mm) cumulated over twodays before sampling. Horizontal dotted lines indicate the LOQ. Nicosulfuronwas detectedonly once in 2011 as a trace concentration at the BSF site (July 4; not shown).

9Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

or their long-range atmospheric transport, would largely determinetheir presence and concentrations in surface stream waters.

The inter-annual variation in detection frequency and concentra-tions of SUmay also be related to climatic and hydrological variation be-tween years which can affect transport of herbicides in surface runoff.Detection frequency and concentrations of herbicides were lowest dur-ing the drier year 2010 and higher in the wetter years 2009 and 2011(Fig. 2). Such a positive relationship between inter-annual differencesin precipitation events and herbicide detection in stream waters haspreviously been reported in other areas (Almvik et al., 2011; Strugeret al., 2011). The relatively higher detection frequencies and higher con-centrations of several SU in May and June each year would correspondto the time of local application of herbicides in Quebec (Giroux andPelletier, 2012). Concentrations generally decreased during July andAugust. Almvik et al. (2011) reported that SU may persist in soil up toone year and be transferred to surface waters at different times. Thepossibility and the extent to which herbicides used in past years were

present in our surface water samples could not be precisely assessedhere. Year-to-year variation in herbicide profiles at our sites as well asthe virtual absence of any SU in late-April samples in 2009 and 2010led us to infer that between-year contamination was low and did notcontribute significantly to herbicide concentrations in surface waterduring springtime. Presumably, snowmelt and spring runoff resultedin thedilution of herbicide concentrations in streamwaters and the sub-sequent dissipation of herbicides by water discharge.

3.5. Spatial variation in SU concentrations and toxicological risk

MaximumSU concentrationsmeasured in our studywerewithin therange of values reported for other agricultural streams and surface wa-ters both inNorthAmerica andEurope, butwere higher for nicosulfuronand rimsulfuron than in four Quebec river sites monitored over thesame time period (Giroux and Pelletier, 2012) (Table 4). Concentrationsof SU exceeding the baseline concentration for non-target aquatic planttoxicity (100 ng L−1) either from a single compound or from the sum ofall SU present in one sample were measured at the catchment basinsites during spring 2009 and spring 2011 (Figs. 5 & 6) and may haveposed some toxic risk to local stream flora (Battaglin et al., 2000;Nyström et al., 1999). Although these high concentrations were ofshort duration, the spatial variation in SU concentrations implied thatthe risk of toxic stress to aquatic plants would be higher at the twocatchment basin sites (Bois-de-Maska and Castorerie) than at the wet-land outlet (BSF) adjacent to the St. Lawrence River during spring andsummer months.

Water properties (Figs. 3 & 4) indicated that surfacewater of the BSFwetland (BSF site)wasmore typically representative of the St. LawrenceRiver water mass. The lower concentrations of SU measured at the BSFsite may be explained by rapid degradation during their transit to andthrough the wetland and/or by dilution resulting from the intrusion ofSt. Lawrence River water into the wetland. Kreuger (1998) reportedthat decreased pesticide concentrations in drainage water at the outletof a small agricultural catchment basin resulted primarily due to dilu-tion from intrusion of groundwater. No data exist on SU levels in theSt. Lawrence River but SU concentrations in its nearest tributary(Yamaska River) were bLOQ (Struger et al., 2011) suggesting that theSt. Lawrence River was probably also characterized by very low SU con-centrations, which would support the dilution hypothesis for lower SUconcentrations at the BSF site.

It has been argued that the alternate occurrence of samples with nodetectable residues interspersed among samples with quantifiable resi-dueswould be indicative thatmovement of herbicides occurred inmoreor less discrete pulses (Michael, 2003). Thus, the intermittent presenceof SU at the Bois-de-Maska and Castorerie sites may indicate occasionalinputs of rainfall runoff to these streams. At the BSF site, where the tem-poral variation inwater properties, particularly pH specific conductivityandwater depth (Fig. 4), indicated the presence of water of different or-igin over time, the sporadic occurrence and very low concentrations ofSU most likely resulted from intrusion of St. Lawrence River water intothe wetland. In consequence, all measured SU concentrations in thedownstream wetland at the BSF site were below criteria for aquaticlife protection andwould therefore presumably represent a low ecotox-icological risk to the flora at the wetland outlet and in the St. LawrenceRiver nearshore environment.

In conclusion, this study confirmed that the presence and concentra-tions of SU in the BSF catchment basin varied extensively between yearsaswell aswithin a given year, most probably as a result of hydroclimaticvariability in stream discharges and variability in herbicide use. Spatialvariation in SU concentrations were largely related to differences in hy-drology of local streams in the catchment basin. The transport of SUfrom the two catchment basin streams into the downstream wetlandresulted mainly in trace level SU concentrations which would not posemajor toxic risk to the wetland flora and fauna.

10 Y. de Lafontaine et al. / Science of the Total Environment 479–480 (2014) 1–10

Conflict of interest

The authors confirm that there are no known conflicts of interest as-sociated with this publication and there has been no significant supportfor this work that could have influenced its outcome.

Acknowledgments

We thankMartin Pilote for his help duringfield sampling and projectplanning and Jonathan Bailey for laboratory analyses. This study wasfunded by the St. Lawrence Action Plan and the Pesticide Science Fundto Environment Canada.

References

Almvik M, Riise G, Bolli R, Borresen T, Christiansen A, Odenmarck S, et al. Multi-yeartransport studies of sulfonylurea herbicides from a barley field in Norway,2007–2010— including development of LC–MS/MSmethods for quantitative analysisof sulfonylurea herbicides and degradation products. 6 (10) Høgskoleveien, Norway:Bioforsk Plant Health and Plant Protection; 20111–45 [BIOFORSK Report].

Battaglin WA, Furlong ET, Burkhardt MR, Peter CJ. Occurrence of sulfonylurea, sulfon-amide, imidazolinone, and other herbicides in rivers, reservoirs and ground waterin the Midwestern United States, 1998. Sci Total Environ 2000;248:123–33.

Beyer EM, Duffy MJ, Hay JV, Schlueter DD. Herbicides: chemistry, degradation and modeof action. Sulfonylureas. New York, NY: Marcel Dekker; 1988117–89.

Cessna AJ, Elliott JA, Tollefson L, Nicholaichuk W. Herbicide and nutrient transport froman irrigation district into the South Saskatchewan River. J Environ Qual 2001;30:1796–807.

Cessna AJ, Donald DB, Bailey J, Waiser M, Headley JV. Persistence of the sulfonylurea-herbicides thifensulfuron-methyl, ethametsulfuron-methyl and metsulfuron-methylin farm dugouts (ponds). J Environ Qual 2006;35:2395–401.

Coyner A, Gupta G, Jones T. Effect of chlorsulfuron on growth of submerged aquatic mac-rophyte Potamogeton pectinatus (sago pondweed). Environ Pollut 2001;111:453–5.

Dinelli G, Vicari A, Bonetti A, Catizone P. Hydrolytic dissipation of four sulfonylurea herbi-cides. J Agric Food Chem 1997;45:1940–5.

Donald DB, Cessna AJ, Sverko E, Glozier N. Pesticides in drinking water supplies of theNorthern Great Plains. Environ Health Perspect 2007;115:1183–91.

Fairchild JF, Ruessler DS, Haverland PS, Carlson AR. Comparative sensitivity of Selenastrumcapricornutum and Lemna minor to sixteen herbicides. Arch Environ Contam Toxicol1997;32:353–7.

Food and Agriculture Organization. FAOSTAT — FAO Statistics; 2013 [http://faostat.fao.org/DesktopDefault.aspx?PageID=424&lang=en#ancor. 10-9-2013].

Giroux I, Pelletier L. Présence de pesticides dans l'eau au Québec: bilan dans quatre coursd'eau de zones en culture de maïs et de soya en 2008, 2009 et 2010. Québec, Canada:Ministère du Développement durable, de l'Environnement et des Parcs; 20121–46.

Hollaway KL, Kookana RS, Noy DM, Smith JG, Wilhelm N. Persistence and leaching of sul-fonylurea herbicides over a 4-year period in the highly alkaline soils of south-easternAustralia. Aust J Exp Agric 2006;46:1069–76.

Hudon C, Carignan R. Cumulative impacts of hydrology and human activities on waterquality in the St. Lawrence River (Lake Saint-Pierre, Quebec, Canada). Can J FishAquat Sci 2008;65:1165–80.

Kreuger J. Pesticides in stream water within an agricultural catchment in southernSweden, 1990–1996. Sci Total Environ 1998;216:227–51.

Kreuger J, Adielson S. Monitoring of sulfonylurea herbicides in stream water drainingintensively cultivated areas in southwestern Sweden during a 9-year period(1998–2006). 1–21 + appendicesUppsala, Sweden: Swedish University of Agricul-tural Sciences, Division of Water Quality Management; 2008103 [Ekohydrologi].

MDDEP. Bilan des ventes de pesticides au Québec pour l'année 2009. Québec: Ministèredu Développement Durable, de l'Environnement et des Parcs; 20121–85.

Michael JL. Environmental fate and impacts of sulfometuron onwatersheds in the SouthernUnited States. J Environ Qual 2003;32:456–65.

Nyström B, Bjömäster B, Blanck H. Effects of sulfonylurea herbicides on non-target aquaticmicro-organisms — growth inhibition of micro-algae and short-term inhibition ofadenine and thymidine incorporation in periphyton communities. Aquat Toxicol1999;47:9–22.

Poissant L, Beauvais C, Lafrance P, Deblois C. Pesticides in fluvial wetlands catchmentsunder intensive agricultural activities. Sci Total Environ 2008a;404:182–95.

Poissant L, Beauvais C, Conruyt M, Copin N. Initiative nationale d'élaboration de normesagroenvironnementales (INÉNA). Démonstration et validation (2006–2008) - Bassinde la rivière Yamaska. Montréal, QC, Canada: Environnement Canada; 2008b1–101[Rapport INÉNA].

Rosenbom AE, Kjaer J, Olsen P. Long-term leaching of rimsulfuron degradation productsthrough sandy agricultural soils. Chemosphere 2010;79:830–8.

Roy L. Les impacts environnementaux de l'agriculture sur le Saint-Laurent. Nat Can2002;126:67–77.

Russell MH, Saladini JL, Lichtner F. Sulfonylurea herbicides. Pestic Outlook 2002;13:166–73.

Sanseman S. Herbicide handbook. In: Lawrence KS, editor. 9th ed. Weed Science Societyof America; 2007.

Struger J, Grabuski J, Cagampan S, Rondeau M, Sverko E, Marvin C. Occurrence and distri-bution of sulfonylurea and related herbicides in central Canadian surface waters2006–2008. Bull Environ Contam Toxicol 2011;87:420–5.

Tomlin CDS. The pesticide manual: a world compendium of pesticides. 14th revised ed.Alton, Hampshire, UK: British Crop Protection Council; 2006.

University of Hertfordshire. Pesticide properties database. Agriculture & EnvironmentResearch Unit (AERU); 2013 [4-1-2013].