lable at ScienceDirect

Environmental Pollution 158 (2010) 737–748

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

Assessing pesticide concentrations and fluxes in the stream of a smallvineyard catchment – Effect of sampling frequency

M. Rabiet*, C. Margoum, V. Gouy, N. Carluer, M. CoqueryCemagref, UR QELY, 3bis quai Chauveau, CP 220, F-69336 Lyon, France

This work brings new insights about the fluxes of pesticides in surface

water of a vineyard catchment, notably during flood events.

a r t i c l e i n f o

Article history:Received 6 April 2009Received in revised form14 September 2009Accepted 5 October 2009

Keywords:DiuronFloodFluxesPesticideSurface waterVineyard catchment

* Corresponding author. Present address: GroupeEnvironnement, University of Limoges, 123 AvenuLimoges, France. Tel.: þ33 (0) 5 55 45 74 84; fax: þ3

E-mail address: marion.rabiet@unilim.fr (M. Rabie

0269-7491/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.envpol.2009.10.014

a b s t r a c t

This study reports on the occurrence and behaviour of six pesticides and one metabolite in a smallstream draining a vineyard catchment. Base flow and flood events were monitored in order to assess thevariability of pesticide concentrations according to the season and to evaluate the role of samplingfrequency on the evaluation of fluxes estimates. Results showed that dissolved pesticide concentrationsdisplayed a strong temporal and spatial variability. A large mobilisation of pesticides was observedduring floods, with total dissolved pesticide fluxes per event ranging from 5.7 � 10�3 g/Ha to 0.34 g/Ha.These results highlight the major role of floods in the transport of pesticides in this small stream whichcontributed to more than 89% of the total load of diuron during August 2007. The evaluation of pesticideloads using different sampling strategies and method calculation, showed that grab sampling largelyunderestimated pesticide concentrations and fluxes transiting through the stream.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The intensive use of pesticides for crop protection againstdiseases led to the widespread presence of these compounds in allthe compartments of the environment (soil, water, air). Thecontamination of surface waters by these biologically activechemicals is an important unwanted side effect of their use. Suchcontaminations represent a potential risk to aquatic life particularlyin rivers draining agricultural areas. Numerous authors have eval-uated the transport of pesticides from agricultural lands to rivers(e.g., Clark and Goolsby, 2000; Du Preez et al., 2005; Claver et al.,2006; Blanchoud et al., 2007). Large-scale studies on pesticidelosses to surface waters showed that the transfer of pesticides islargely influenced by the intrinsic properties of each compound,but also by the weather conditions, the soil types and land use(Kreuger, 1998; Capel et al., 2001; Riise et al., 2004). The role ofhydrology in non-point diffuse pesticide pollution is documented,but hardly quantified. Several authors emphasized the significanceof rainfall-induced surface runoff as a major source of pesticides instreams (Richards and Baker, 1993; Kuivila and Foe, 1995; Ng and

ment de Recherche Eau Sole Albert Thomas, F-87060

3 (0) 5 55 45 72 03.t).

All rights reserved.

Clegg, 1997; Kreuger, 1998; Capel et al., 2001; Bach et al., 2001;Schulz, 2001a,b; Holvoet et al., 2007; Vryzas et al., 2009). Rapidflow processes drastically reduce the time available for reactions,such as sorption or degradation, and can lead to a direct transfer ofpesticides towards surface waters (Muller et al., 2003). It is widelyrecognised that during floods, fluxes of many pollutants (e.g.,nutrients, metals, pesticides) can vary over several orders ofmagnitude and may account for the majority of the annual load ofpollutants in large river systems (Meybeck, 2005; Zonta et al., 2005;Eyre and Pont, 2003; House and Warwick, 1998). Much less isknown about pesticide transfer dynamics and loads in small streamsystems, where hydrological conditions are subject to fast andconsiderable variation with time (Gargouma et al., 1997; Ng andClegg, 1997; Kreuger, 1998; Louchart et al., 2004; Leu et al., 2004;Domange, 2005). Surface waters draining vineyard areas are ofspecific concern due to the large quantities and diversity of pesti-cides applied and the vulnerability of cultivated land (i.e., shallowsoil depth, steep slopes). Indeed, several studies reported thepresence of pesticide residues in surface waters in vineyard areas(Lennartz et al., 1997; Louchart et al., 2001, 2004; Ribolzi et al.,2002; Rabiet et al., 2008; Hildebrandt et al., 2008). In spite of thishighlighted vulnerability of vineyard areas, very few studiesreported on the spatio-temporal variability of pesticide concen-trations nor on the evaluation of corresponding fluxes. Theassessment of pesticide fluxes is essential to evaluate the envi-ronmental performance of agricultural practices and to improve

Fig. 1. Location of the Morcille watershed and the sampling sites.

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748738

water quality at a basin scale. In small stream, Preston et al. (1992)and Cohn (1995) showed that flux estimates of nutrients or metalswere often strongly biased when water discharge varied greatlywith time.

One of the objectives of the European Water FrameworkDirective (WFD, European Commission, 2000) is to reach a ‘‘goodstatus’’ for European rivers by 2015. The implementation of theWFD implies the intensification of the monitoring of contaminants,the identification of the causes of degradation and the imple-mentation of corrective actions to obtain a good chemical andbiological status. For this purpose, Member States have to establisha comprehensive monitoring strategy to establish contaminantexposure levels in surface waters and to evaluate the water qualityimprovement linked to various management programmes at thecatchment scale. Within the WFD, operational monitoring isrequired for the 33 priority substances; it should be performed viagrab sampling 12 times per year. Pesticide transfer during floodevents are not specifically considered whereas many studiesshowed their major role by mobilising great quantity of contami-nants. Moreover, several authors (e.g., Schleppi et al., 2006; Johnes,2007; Horowitz, 2008) have shown for river systems and also smallstreams that an infrequent monitoring programme, with samplingperformed only for a representative range of flow conditions mayyield better estimates of solute loads (nutrients, sediment-associ-ated trace elements or major elements) than a routine water qualitymonitoring programme with sampling at fixed time intervals,

regardless of flow conditions The objectives of this work was on theone hand to evaluate the spatio-temporal variability of pesticidesconcentrations and fluxes in an small-sized vineyard catchment(inferior to 10 km2) and on the other hand to assess the outcome ofdifferent sampling strategies on the estimation of pesticide fluxes.In particular, the effect of sampling frequency on the evaluation ofpesticide loads and concentrations using several sampling strate-gies was estimated and discussed.

The Morcille catchment was selected for this study as it isa vineyard’s area and its small size (8 km2) corresponds to the scalewhere novel agricultural practices are implemented and evaluatedfor their environmental performance towards the protection ofwater quality. One site of the Morcille stream was instrumented inorder to monitor both flood and between-flood (base flow) eventsduring one hydrological year. We used various set of monitoreddata (weekly or monthly grab, weekly automatic fractionatedsampling) to compare concentrations and fluxes obtained usingthese different sampling strategies.

2. Material and methods

2.1. Study area

The investigation site is the Morcille catchment, which has been studied fora long-time by the Cemagref, French Agricultural and Environmental EngineeringResearch Institute. It is a representative catchment of the Beaujolais area for thestudy of the transport of pesticides from agricultural parcels to surface water (Gouy

St Joseph

(Site 1)

Les Versauds

(Site 2)

St Ennemond

(Site 3)

Industrial and urban

Vineyard

Forest

Cultures

Grasslands and heathlands

Fig. 2. Land use at the three sampling sites on the Morcille watershed.

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748 739

and Nivon, 2007; Rabiet et al., 2008). The Morcille catchment is located about 70 kmnorth of the city of Lyon in France (Fig. 1). Its small size (8 km2) allows to definea study area where transfers are more easily comprehensible. The head of the basinis essentially covered by forest, whereas downstream it is mainly dedicated tovineyard with the presence of about 70 wine-growers (Gouy and Nivon, 2007). Thevine covers about 70% of the catchment’s total area (Fig. 2). No industrial activitiesare present on the site. The hydrology regime of the Morcille stream variesconsiderably over the year, with instantaneous water flow ranging from 0.005 m3/s(low-water mark) to more than 0.5 m3/s (during storms). It is characterized bysignificant hydrological events (strong and fast floods). The substratum of thecatchment is an altered crystalline basement of mainly granite, with some ratherhomogeneous soils varying from sandy loam on the upper slopes to more clayeysoils in the bottom of the valley (Gouy and Nivon, 2007). Considering the soiltexture, the rather steep slopes (varying from 2% to more than 20%) and the presenceof a shallow crystalline basement, it is likely that infiltration and sub-surface lateralflow are major processes. Surface runoff only appears during high intensity rain-storms exceeding the soil infiltration capacity. In this area, climatic, geological,topographical and agricultural conditions concur to potentially create very dynamichydrological conditions and a high erosion rate. Thus, pesticide transfer to surfacewaters may be favoured. In vineyards, herbicides are applied from the end of Marchto April, whereas fungicides and insecticides are used from May to August.Considering both hydrologic conditions and pesticide application dates, the morecritical period as regards to surface water contamination extends from April to theend of August.

2.2. Water sampling

A monitoring survey was performed from March 2007 to March 2008 duringa complete vine growing season. Three sampling sites along the Morcille streamfrom upstream (1-St-Joseph), intermediate (2-Les Versauds) to downstream (3-St-Ennemond) were monitored in order to approach the spatial variability (Figs. 1 and2). As for the temporal variability, a sampling strategy was defined so as to inves-tigate the three sites during one year, focusing on the pesticide application periodsand the following months. Hence, a more intensive monitoring, with weekly grabsamples, was performed from March to September 2007, which included thepesticide application period; a monthly grab monitoring was done thereafter (fromOctober 2007 to March 2008).

In order to better identify pesticide concentration amplitude and fluxes duringstorm events, a specific monitoring was performed at the intermediate site (2-LesVersauds) during spring and summer 2007. Sampling was performed using tworefrigerated automated samplers (Buhler, model 4010) with a capacity of 24 poly-ethylene (PE) bottles. The high capacity of each samplers as well as the use of twosamplers allowed to minimize the risk of missing a portion of the flood event.It permitted a fractionated flood sampling at a rate proportional to the flow, allowingto better estimate fluxes than a time dependant sampling. Automatic samplingparameterization was designed using knowledge and experience from the previoustwo years of flow measurements (2005 and 2006).

In addition, in order to compare various sampling strategies to assess pesticidefluxes and mean concentrations in this stream, a weekly composite sampling was

Table 1Properties of the studied pesticides (Tomlin, 2000; AFSSA, 2008).

Compound (abbreviation) Category Water solubility(mg/L)

Koc (L/kg) Log Kow Half life(days)

Azoxystrobin (AZS) Fungicide 6.0 423 2.50 8.7–14Carbendazim (CBZ) Fungicide 8.0 200 1.56 30–150Dimetomorph (DMM) Fungicide 18.0 290–566 2.70 23–35Diuron (DIU) Herbicide 36.4 480 2.85 90–180Procymidone (PCM) Fungicide 4.5 1514 3.14 28–84Tebuconazole (TBZ) Fungicide 32.0 803–1251 3.70 43–170

0

20

40

60

80

100

0 2 4 6 8Time (days)

)%(

oC/

C

DIU DCPMU AZSTBZ DMM PCM

Fig. 3. Study of pesticide stability during storage in PE bottles over a seven days period.Y-axis is the ratio expressed in percent of the concentration of pesticide at time t (C)over the concentration at t0 (C0).

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748740

performed in July and August 2007 with an additional refrigerated automaticsampler (Bamo, model Aquacell) equipped with a single PE bottle of 10 L. Thissampler was also programmed to be proportional to the flow. Thus five samplingmethods were compared at site 2: grab sampling (weekly, bimonthly or monthly),automatic fractionated sampling during storms and automatic weekly compositesampling.

In parallel, the water depth was continuously recorded at site 2 with a pressuresensor and converted into water flow values using a sampling-site specific calibra-tion function. Rainfall was measured using two rain gauges (i.e., tipping bucketsystem) located upstream and in the middle of the Morcille catchment (Fig. 2). Dataobtained with each gauge were similar and didn’t show any significative spatialvariability of rainfall. In this study, we reported only the data obtained with theclosest rain gauge to site 2.

2.3. Pesticides analysis

2.3.1. Pesticides selectionIt was not intended to analyse all the pesticides present in surface water of the

Morcille stream, but rather to select a range of compounds widely used in vineyardcatchment and with various physico-chemical properties. Pesticides were selectedamong a list resulting from previous screening analysis in surface water at the samesite (Gouy and Nivon, 2007; Rabiet et al., 2008). Six pesticides and a main metabolitewere chosen from different families of herbicides and fungicides currently used forvineyard treatments: azoxystrobin (AZS), carbendazim (CBZ), dimetomorph (DMM),procymidone (PCM), tebuconazole (TBZ), diuron (DIU) and one of its metabolites:3-(3,4-dichlorophenyl)-1-methyl urea (DCPMU). Their main physico-chemicalproperties are presented in Table 1. Diuron, an herbicide largely used on vineyard, isincluded in the list of the priority substances of the WFD (European Commission,2000, 2008). Insecticides, although widely used in vineyard to protect crop againpest, were not monitored in this study since occurrence as well as concentrationswere found to be negligible in the water bodies of the Morcille watershed (Gouy andNivon, 2007).

2.3.2. ChemicalsPure pesticides used for standard solutions or quality controls were of analytical

grade and supplied by Cluzeau Info Labo (Sainte Foy la Grande, France) or SigmaAldrich (Saint Quentin Fallavier, France). Organic solvents used were pesticide orHPLC grade (SDS, Peypin, France).

The individual stock solutions were prepared in acetone at a concentration of100 mg/L and stored at 4 �C. These stock solutions were stable for 2 months. Stan-dard working solutions at various concentrations were prepared weekly by appro-priate dilutions of the stock solutions in ultrapure water/acetonitrile (80/20, V/V).

2.3.3. Sample pre-treatmentExcept for weekly composite sampling, samples were collected within

a maximum of 24 h after sampling and returned to the laboratory for immediatefiltration. Water samples were filtered on glass fiber filters (0.7 mm, B ¼ 47 mm,GF-F Whatman, Sodipro, Echirolles, France). Extraction was carried out on solidphase extraction (SPE) cartridges (Oasis HLB, 60 mg, 3 mL, Waters, France) afterspiking with linuron used as analytical control standard. Ten microliters of internalstandard control (deutered diuron) were added to the final extract of 240 mL water/acetonitrile (80/20, V/V). The concentration factor was about 1000. The organicextracts and corresponding calibration standards were kept at �20 �C prior toanalysis.

2.3.4. AnalysisWater sample extracts were quantified by ESI-LC-MSMS (LC 1100 Agilent, API

4000 Applied Biosystems, Courtaboeuf, France). The compounds were separated ona 250 � 2 mm Purospher Star RP-18 column (Merck). The mobile phase was water/acetonitrile gradient at 0.3 mL/min. The limit of quantification (LQ) ranged between0.020 and 0.080 mg/L according to the pesticides, and recoveries varied from 80 to107%, depending on the concentration level and the compound (Margoum et al.,2007). All the concentration results presented in this study were then corrected withthe corresponding recovery rate.

2.4. Quality assurance

Possible losses of compounds by sorption or degradation during samplecollection, storage and transport was evaluated in two separate laboratory studies.First, pesticide stability during storage in PE bottles was assessed. Mineral Evianwater (3 replicates) was spiked with the studied pesticides (5 mg/L) and then storedin the same conditions as during monitoring (at 4 �C in darkness). Pesticidesanalyses were performed at time 0, 1 h, 24 h, 48 h and 7 days in order to simulatedifferent time of storage inside the automatic samplers. Results are presented inFig. 3. Diuron is the most stable compound in PE bottle with losses below 3%, evenover a seven days period. For dimetomorph and DCPMU, losses are below 10% duringstorage. It is also the case for procymidone over a 2 days period. However, overa seven days period, procymidone showed a significant decrease of concentration,with about 40% of losses. This behaviour could be due to its relatively high Koc(Table 1). Despite their respectively low half life and high Koc, azoxystrobin andtebuconazol showed a good stability, with maximum losses reaching a maximum of15% over 7 days. In conclusion, stability of all compounds is satisfying over 7 days,expect for procymidone; thus average weekly concentrations of procymidone mightbe somewhat underestimated.

In the second study, spiked and unspiked mineral water (Evian) were runthrough the automatic sampler (i.e., passing through the tubes and collected in thereceiving bottles of the sampling equipment) in order to verify the potential loss ofpesticides and the absence of contamination, respectively. No significant adsorptionor release of pesticides was observed (i.e., the relative difference of pesticideconcentrations before and after the sampler was below 5%).

Field and analytical blanks (using ultrapure water in PE bottles) were alsoperformed for each monitoring campaign.

2.5. Flux calculation

Pesticide fluxes were estimated at site 2 over various periods and using differentdata and calculation methods.

The weekly load of pesticides (WL) was calculated for the March–Septemberperiod, using the manual weekly sampling data, by multiplying the average weeklystream flow by the total pesticide concentration (i.e., sum of the concentrations ofthe 8 studied pesticides) measured on grab samples as follows:

WLðmgÞ ¼ Q � Ci � ti (1)

where Q is the average weekly stream flow during the considered period ti (L/s); Ci isthe concentration of total pesticides analysed in grab samples collected once a week

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748 741

(mg/L); and ti is the considered time period (s) (seven days surrounding the samplingday).

For the monthly monitoring, over the October–March period, monthly pesticidefluxes were estimated in the same way by multiplying the average monthly streamflow by the total pesticide concentration measured on the samples collectedmanually once a month. Concentrations below the LQ were set to half of the LQ forthese calculation. Compounds not detected were set to 0.

The total pesticide load during a flood (FL) was calculated using the followingequation:

FLðmgÞ ¼X

qi � Ci � ti (2)

with qi, the instant flow at the time i (L/s); Ci , the concentration value linearlyinterpolated between both nearest measured pesticide concentrations (mg/L); and ti,the time interval between the flow measurements (s). This time varied from severalseconds to minutes.

3. Results and discussion

3.1. Hydrological context

Daily rainfall (mm) and water flow (L/s) of the Morcille stream atsite 2 for the study period (from 01/03/2007 to 31/03/2008) arepresented in Fig. 4. Total annual precipitation during year 2007 wasabout 909 mm. This value is not significantly different from theaverage annual precipitation during the past fifteen years(860 mm). However, summer was particularly wet in 2007, witha total precipitation of 426 mm from May to August (Fig. 4),compared to 290 mm in average (for the past fifteen years) (MeteoFrance). Especially, the months of June and August were particu-larly wet (125 and 101 mm respectively, versus an average of 67 and78 mm, respectively). This specific context could contribute toincrease surface water contamination by pesticide: on the onehand, a higher occurrence of storms in summer is favorable tosurface runoff contributing to mobilise more pesticides, and, on theother hand, wet conditions often lead farmers to use more pesti-cides than a normal year, because of a higher risk of crop disease.

01/0

3/07

15/0

3/07

29/0

3/07

12/0

4/07

26/0

4/07

10/0

5/07

24/0

5/07

07/0

6/07

21/0

6/07

05/0

7/07

19/0

7/07

02/0

8/07

16/0

8/07

30/0

8/07

13/0

9/07

0

01

02

03

04

05

06

07

08

09

001

01/0

3/07

15/0

3/07

29/0

3/07

12/0

4/07

26/0

4/07

10/0

5/07

24/0

5/07

07/0

6/07

21/0

6/07

05/0

7/07

19/0

7/07

02/0

8/07

16/0

8/07

30/0

8/07

13/0

9/07

Dai

ly r

ainf

all (

mm

)

.tpeS70tsuguA70yluJ70enuJ70yaM70lirpA70hcraM

Fig. 4. Daily rainfall (mm) and water flow (L/s) at site 2 in th

3.2. Grab monitoring of pesticide concentrations along the Morcillestream

Dissolved herbicides and fungicides concentrations measuredfrom grab weekly and monthly samples along the Morcille streamfrom 6 March 2007 to 18 March 2008 are reported in Fig. 5.

The presence of pesticides was demonstrated throughout theinvestigation period with a large seasonal variation, correspondingto the pesticide application calendar. Indeed, total pesticideconcentrations range from below the LQ to 16.4 mg/L. Diuron andDCPMU concentrations ranged from below the limit of quantifica-tion (LQ) to 13.3 mg/L (Fig. 5a). Concentrations were below 0.2 mg/Lin March 2007 and 2008, corresponding to a residual backgroundlinked to previous vineyard treatments. DIU and DCPMU concen-trations increased in the stream at the beginning of April 2007,according to the onset of herbicide spreading on vineyard. Thehighest herbicide concentration was observed in May and June andat the beginning of July.

Total fungicide concentrations were very low in March and April2007 (below 0.3 mg/L). In May, fungicide concentrations increasedslightly with values ranging between 0.3 and 0.9 mg/L at thedownstream site. Starting in June, fungicide concentrations rosesharply. The maximum concentration was observed in July with8.3 mg/L at site 3 (Fig. 5b).

Herbicide as well as fungicide concentrations decreased inAugust, but remained at a detectable level during all year round:this was especially the case for diuron, mainly applied in late winterbut also locally until early summer due to the wet conditions. Thus,surface water contamination by pesticides largely overstepped theapplication period, as shown by the high pesticide concentrationsstill present in January (up to 2.0 mg/L at site 3).

Besides seasonal variations, we observed a large variability ofpesticide concentrations with time, over very short periods. Forinstance, between the 30 May 2007 and the 05 June 2007, diuron

0

05

001

051

002

052

003

27/0

9/07

11/1

0/07

25/1

0/07

08/1

1/07

22/1

1/07

06/1

2/07

20/1

2/07

03/0

1/08

17/0

1/08

31/0

1/08

14/0

2/08

28/0

2/08

13/0

3/08

27/0

3/08

Wat

er fl

ow (L

/s)

27/0

9/07

11/1

0/07

25/1

0/07

08/1

1/07

22/1

1/07

06/1

2/07

20/1

2/07

03/0

1/08

17/0

1/08

31/0

1/08

14/0

2/08

28/0

2/08

13/0

3/08

27/0

3/08

0.voN70.tcO70 80hcraM80.beF80.naJ70.ceD7

skroW

e Morcille catchment from March 2007 to March 2008.

0

1

2

3

4

5

6

7

8

9

01

060/ 3 30/31 2

/0 032

/7 0303

0/ 4 0/0141

40/740/52 0

/3 0510

0/ 515

0/ 5 0/2253

50/060/50 1

/2 0619

0/ 626

0/ 603

0/ 7 70/01 1/7 07

270/4 31

0/ 7 0/70814

0/ 8 80/12 2/8 08

090/4 11

0/ 918

0/ 924

0/ 9 01/61 2/0 11

111/ 2

2/4 01

190/ 2 30/81

Tot

al f

ungi

cide

s co

ncen

trat

ion

(µg/

L)

1etiS 2etiS 3etiS

2yaM7002lirpA7002hcraM botcO7002.tpeS7002tsuguA7002yluJ7002enuJ700 -7002re8002hcraM

0

2

4

6

8

01

21

41

0/603 0/31

3 /02302

30/7 040/3

40/01 7140/ 52

40/ 3050/ 0/01

5 0/515 /22

503

50/0 060/5

60/21 9160/ 62

60/ 3070/ 0/01

7 /7170 /42

703

70/180/70

80/41 1280/ 82

80/ 4090/ 0/11

9 /8190 /42

901

01/611/02

21/11 4210/ 91

20/ 8130/

Diu

ron

and

DC

PM

U c

once

ntra

tion

(µ

g/L

)

1etiS 2etiS 3etiS

botcO7002.tpeS7002tsuguA7002yluJ7002enuJ7002yaM7002lirpA7002hcraM -7002re8002hcraM

a

b

Fig. 5. Pesticides concentrations (in mg/L) along the Morcilles stream from March 2007 to March 2008; a: diuron and DCPMU; b: total fungicide (sum of the five fungicidesmonitored).

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748742

and metabolite concentrations at site 3 increase from 0.3 to 13.3 mg/L. This high reactivity may be related to the rapid pesticide transfervia runoff due to storm events, which can mobilise a large andvariable quantity of pesticides. Indeed, this period of the year wasmarked by large rainfall, with about 55 mm during one week (cf.Fig. 4); this led to a shorter delay between pesticide application andprecipitation and favouring the direct transfer to surface water.

We present in Fig. 6 the relationship observed between totaldissolved pesticide concentrations and water flow during grabmonitoring at site 2. The positive linear regression is statisticallysignificant (r2 ¼ 0.78; P < 0.0001). The low pesticide concentrationfor high flux corresponds to a sample collected during a stormevent at the end of summer, thus at the end of the applicationperiod.

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0 2 4 6 8 10 12 14

Pesticide concentrations (µg/L)

Wa

te

r flo

w (m

3/s

)

Fig. 6. Water flow (m3/s) versus total pesticide concentrations (mg/L) in the Morcillestream at site 2 during weekly monitoring (March 2007–September 2007).

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748 743

We noted that pesticide concentrations generally increasedwith water flow during the application period and the followingmonths as well, when it is likely that pesticide input to soil andtheir availability in soil is still significant. This underlined the majorrole of hydrological conditions on the mobilisation of pollutant.As for pesticide occurrence during base flow, it is assumed that itcan be related to the slow processes of water movement throughthe soil matrix such as sub-surface runoff, drainage and exchangeswith shallow water table (Muller et al., 2003).

On a spatial point of view, we measured a large increase ofpesticide concentrations from upstream to downstream, which iscoherent with the increasing proportion of vineyard surfacedownstream (Fig. 2). Indeed, at the upstream reference site (site 1),only diuron and its metabolite, as well as dimetomorph andprocymidone were quantified and only twice at low levels (fromthe LQ up to 2 mg/L). This reflects the small impact of vineyardtreatment at the upstream site, whereas at the outlet, pesticideswere systematically quantified with a mean concentration of totalpesticides of 3.2 � 4.0 mg/L at site 3.

0

2

4

6

8

01

21

41

60

/3

0 /3

1

03

02

/3

0

27/0

3

30

/4

0

10/0

4

17/0

4

25/0

4

03/0

5 50/

01 1

5/0

5/

22

05

30/0

5/

50

06

21

/6

0 /9

1

06

62

/6

0

03/0

7

1

Dis

so

lve

d p

es

tic

ide

co

nc

en

tra

tio

ns

at s

ite

2 (

µg

/L)

UID UMPCD SZA

7002enuJ7002yaM7002lirpA7002hcraM

Fig. 7. Dissolved pesticide concentrations (mg/L)

The distribution between the different compounds/moleculesat site 2 is reported in Fig. 7. Diuron predominated, withconcentrations ranging from 0.01 mg/L to 7.5 mg/L. Thus, theaverage concentration of diuron (1.2 � 2.0 mg/L) largely exceededthe European environmental quality standard (EQS) of 0.2 mg/Lexpressed as annual average value (European Commission,2008). Moreover, at site 3, the downstream site, more than onethird of the samples exhibited diuron concentrations higher thanthe maximum allowable concentration (MAC) of 1.8 mg/L (Euro-pean Commission, 2008). These results reflect the strong impactof diuron used in vineyard on the surface water quality all alongthe year. Moreover, DCPMU, a main metabolite of diuron, wassystematically quantified with concentrations ranging between0.01 and 0.5 mg/L at site 2. At site 3, DCPMU concentration rangedbetween 0.2 and 1.4 mg/L, with a mean concentration of0.3 � 0.3 mg/L. Dimetomorph and procymidone, two fungicidescommonly utilized on vines, were mostly quantified during themonth of July, with concentrations reaching up to 3.4 and 1.3 mg/L respectively at the intermediate site, reflecting their applicationperiod on the basin. Tebuconazol was detected all along theapplication period at site 2 with concentration ranging between0.02 and 1.9 mg/L, corresponding to an average concentration of0.2 � 0.3 mg/L. Carbendazim and azoxystrobin were less oftenquantified in the stream, with mean concentrations of0.1 � 0.2 mg/L and 0.08 � 0.09 mg/L, respectively at the down-stream site (site 3).

The weekly load of total dissolved pesticides at site 2, esti-mated from equation (1), ranged between 3.3 g and 337 g overthe study period. This lead to an annual load of pesticides of2200 g then 12.5 g/Ha taking into account the vine surfacedrained at site 2 (175 Ha). More than 50% of this amount (1088 g)was transferred during June and July only, with 623 and 465 g,respectively. Noteworthy, diuron and its metabolite represented68% (1502 g) of the annual total load. It represented about 8.6 gof diuron and metabolite per hectare. Considering the

0/

70

17/0

7/

42

07

31/0

7/

70

08

41

/8

0 /1

2

08

82

/8

0 /4

0

09 9

0/1

18

1

/9

09

0/4

26

1

/0

1

20/1

1

11

/2

1

24/0

1/

91

02

18/0

3

ZBT MMD ZBC MCP

7002yluJ tsuguA

7002

rebmetpeS

7002

-7002rebotcO

8002hcraM

at site 2 from March 2007 to March 2008.

Table 2Annual pesticide fluxes in g and g/Ha and percentage of exportation in the Morcillestream.

Pesticides Annualflux (g)

Annual specificflux (g/Ha)

Annual recommendedamount (g/Ha) (Couteuxand Lejeune, 2007)

Percentage ofexportation (%)

DIU 1335 7.60 1500 0.57DCPMU 167.7 0.96 –AZS 31.7 0.18 250 0.07TBZ 223.8 1.28 70 1.80DMM 164.0 0.94 250 0.38CBZ 52.0 0.30 105 0.28PCM 180.0 1.03 750 0.14

Total 2154 12.30

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748744

recommended amount of diuron to be applied per hectare andper year on this type of culture (i.e., 1500 g/Ha, Couteux andLejeune, 2007), about 0.57% of the total amount of applied diuronreached the Morcille stream (Table 2). Noteworthy, the load ofexported diuron should probably decrease in the future, as its useis no longer authorized. This value is in agreement withmeasured pesticide losses to surface water reported by otherauthors (for cereal cultures), ranging from 0.01 to 0.5% of appli-cation amounts (Brown et al., 1995; Kreuger, 1998). In fact, ourcalculation was based on pesticide concentrations measured inweekly grab samples, which mainly corresponded to base-flowconditions; thus it did not take into account the majority ofhigh-flow events, likely to contribute to annual pesticide loadtransiting through the Morcille stream. We might consider thatthe estimation of 0.57% of diuron losses probably largely under-estimates the actual quantity of diuron exported from vineparcels to the stream. However, in a Mediterranean vineyardcatchment, Louchart et al. (2001) estimated to 0.52% the totalseasonal losses for diuron, including losses by storm flow andbase flow. This lower value compared to our result is probablydue to different characteristics of the watershed studied, inparticular the slope; alternatively it might result from differencesin the monitoring and calculation methods. In this study, baseflow only contributed to 4% of total losses.

Table 3Maximal pesticide concentrations (mg/L) and fluxes (g) for each monitored flood at site 2

DIU DCPMU

29/04/2007 Cmax (mg/L) 11.9 1.09Flux (g) 0.477 0.023

26/05/2007 Cmax (mg/L) 5.30 0.30Flux (g) 0.68 0.064

27–28/05/2007 Cmax (mg/L) 9.20 0.60Flux (g) 6.91 0.59

01–02/07/2007 Cmax (mg/L) 134.0 0.71Flux (g) 45.5 0.98

02/08/2007 Cmax (mg/L) 7.50 –Flux (g) 0.89 –

06–07/08/2007 Cmax (mg/L) 6.40 0.79Flux (g) 7.21 1.08

21/08/2007 Cmax (mg/L) 21.1 –Flux (g) 4.80 –

17/09/2007 Cmax (mg/L) 13.0 –Flux (g) 7.33 –

27/09/2007 Cmax (mg/L) 4.80 –Flux (g) 1.01 –

3.3. Storm events monitoring during spring and summer 2007

We monitored 8 floods from April to September 2007. Amongthem, 5 floods were fully monitored and 3 floods could only besampled in the region of the water flow peak. Sampled flood eventswere characterized by a maximum water flow ranging between0.03 and 0.78 m3/s and a duration lasting between 2 and 11 h. Thesecharacteristics are typical of most events taking place in theMorcille catchment (Taillandier, 2008).

The maximum pesticide concentration recorded during eachflood event, as well as the total flux is presented in Table 3 for eachmonitored compound. Fig. 8 presents the distribution of diuronconcentrations during the 8 monitored floods, as it is the mainpesticide quantified during the study period.

As expected, we observe a large mobilisation of diuron in thestream during the high-flow events. Concentrations rose rapidly,then usually hit the highest point almost at the same time as waterflow peak (Fig. 8). However, a lag time (ranging between fewminutes and more than 1 h according to the event) was sometimesobserved before or after the maximal pesticide concentration, incomparison with the water flow peak. Concentrations declinedthereafter. Maximum pesticide concentration during the 8 moni-tored floods ranged from 4.8 mg/L (27 September 2007, outside thespreading period) to 140 mg/L (01–02 July 2007, considered asexceptional) with an average value of 26.7 � 42.8 mg/L. The 01–02July storm event was characterized by very high concentrations ofall monitored pesticides (about 10 times higher than for other floodevents). Thus, we considered this event as exceptional, probablydue to the occurrence of the flood just after pesticides application.Some authors (Louchart et al., 2001; Leu et al., 2004; Phillips andBode, 2004) have indeed shown that the major pesticide lossesoccurred during the first rain event after application. Excluding thisexceptional event, as well as the event taking place out of spreadingperiod (end of September), maximum concentrations during floodswere between 7.3 and 21.1 mg/L, with an average value of12.7�4.3 mg/L. Among all the monitored pesticides, diuron was themost mobilised compound during flood. Its contribution reachedup to 76% of the total pesticide load transiting during flood.Maximum diuron concentrations observed during the flood eventswere always higher than the MAC (European Commission, 2008).

.

AZS TBZ DMM CBZ PCM

0.13 0.60 0.10 0.10 0.270.0008 0.011 0.079 0.0005 0.027

0.43 0.32 0.75 0.10 0.140.057 0.057 0.084 0.015 0.040

0.54 1.01 0.82 0.12 0.230.46 1.02 0.55 0.13 0.26

0.14 6.50 14.4 0.16 1.300.14 5.00 6.80 0.24 1.50

– – – – –– – – – –

0.08 1.41 2.81 0.28 3.170.082 1.50 3.67 0.28 3.49

– – 5.98 – –– – 1.3 – –

– – – – 1.40– – – – 0.193

– – – – –– – – – –

0.00

0.01

0.02

0.03

0.04

0.05

19:40 20:52 22:04 23:16

m(

wol

fr

et

aW

3

)s/

0

2

4

6

8

10

12

14

Water flow

Diuron

29/04/2007

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

02:24 14:24 02:24 14:24 02:24 14:24

0

2

4

6

8

10

12

14

)L

/g

µ(

no

it

ar

tn

en

oc

no

ru

iD

Water flow

Diuron

26-27/05/2007

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

9:36 14:24 19:12 0:00 4:48 9:36 14:24 19:12 0:00

-10

10

30

50

70

90

110

130

150

01-02/07/2007

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

00:00 04:48 09:36 14:24 19:12 00:00

0

2

4

6

8

10

12

14

02/08/2007

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

18:00 22:48 03:36 08:24 13:12

0

2

4

6

8

10

12

14

06-07/08/2007

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

12:00 13:12 14:24 15:36 16:48

0

5

10

15

20

25

21/08/2007

0.00

0.05

0.10

0.15

0.20

0.25

0.30

19:12 20:24 21:36 22:48 00:00 01:12

0

2

4

6

8

10

12

14

17/09/2007

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

18:00 20:24 22:48 01:12 03:36

0

2

4

6

8

10

12

14

27/09/2007

Fig. 8. Diuron concentrations (mg/L) and water flow (m3/s) at site 2 during flood events for the April–September 2007 period.

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748 745

Depending on the flood (and excluding the extreme event of01–02 July 2007), pesticide maximum pesticide concentrationsvaried by a factor of 3 only, whereas peak water flows varied bya factor of up to 30. Thus, maximum pesticide concentrations did

not seem to be a simple function of water flow and it is likely thatother processes interfered, such as pesticide availability via runoff.

During floods, the total dissolved pesticide fluxes ranged from1.0 to 60 g. When occurring immediately after application, even

0

0.05

0.1

0.15

0.2

0.25

0.3

01/08 06/08 11/08 16/08 21/08 26/08 31/08

m(

eg

ra

hc

siD

3

)s/

0.78 m3

/s

× ××

××

× Grab monitoring with [DIU]

Floods monitored

0.16 µg/L0.97 µg/L

0.12 µg/L

0.25 µg/L

0.11 µg/L

Fig. 9. Water flow (m3/s) in the Morcille stream at site 2, with the sampling dates forthe monitoring of grab samples and floods in August 2007 (�: diuron concentrationmeasured using grab sampling).

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748746

a flood of moderate intensity could in fact deliver to the streama significant load of pesticides in a very short period.

3.4. Significance of the sampling strategy on the estimation ofdiuron fluxes and average concentration

3.4.1. Flux estimationIn order to assess the effect of the sampling strategy on the

estimation of diuron fluxes and average concentration, we useddifferent methods and various set of data monitored at site 2 inAugust 2007. This specific month was chosen for this exercise sincea majority of the storm events during this period could be collected(Fig. 9). Besides, pesticide concentrations during this period werestill very significant.

The monthly diuron flux was estimated using weekly, bimonthlyor monthly simulated grab monitoring by choosing either the first,second, third or fourth weekly sample as the reference concentra-tion for the calculation; or alternatively, by using only 1, 2 or 4weekly data. In addition, diuron monthly flux was calculated usingthe weekly average composite sample concentrations. Finally,diuron monthly flux was estimated for either base flow only, usinggrab samples collected during base-flow conditions; or storm flowconditions only, using automated sampling performed duringfloods. For the floods that could not be monitored during this period,we estimated the corresponding diuron fluxes using an empiriclinear model derived from observed data. Indeed, excluding theextreme event of 01–02 July 2007, a linear regression (r2¼ 0.91) wasobtained between diuron fluxes during flood events and the totalvolume of water conveyed during the event. Thus, diuron fluxduring a flood could be estimated with the following equation:

Diuron flux ðgÞ ¼hWater volume

�m3�� 3:6� 10�3

i� 0:1

(3)

y =3.6.10-3

x - 0.1

R2 = 0.87

0123456789

0 500 1000 1500 2000 2500Water volumes (m

3)

)g

(x

ul

fn

or

ui

D

Fig. 10. Diuron flux (g) versus water volumes (m3) during 7 floods (site 2).

Although only 7 data points were available to derive this rela-tionship (cf. Fig. 10), it allowed to estimate with less than 30% ofbias diuron load during a flood event taking place during theconsidered period (Table 4). Due to pesticide decay with time andto the limited range of water flow values used to build this equa-tion, the validity of this equation is clearly limited to the studyperiod and it would be incorrect to use it in another context andpurpose.

Results are presented in Fig. 11. According to the method used,diuron monthly fluxes ranged from 4.7 to 73.9 g. The highest valuecorresponded to the calculation using the weekly compositesampling results, which was considered as the most accurateevaluation. Hence, this value was used thereafter to estimate thebias due to grab sampling. Monthly diuron fluxes, estimated fromgrab monitoring (weekly, bimonthly or monthly) varied between4.7 and 41.2 g, which is significantly lower than the ‘‘most realistic’’flux value (73.9 g). Even the grab weekly surveys underestimatedpesticide flux by a factor of about 5 at most. As a matter of fact, inAugust 2007 around 89% of the diuron load were transportedduring floods, which in total lasted for less than 5 days (i.e., 15% ofthe time). These results are similar to those of Chen et al. (2005) andLouchart et al. (2001) showing that floods contributed to 90% and84% (at the field scale) respectively, of the total load (for floodseason and annual time scale respectively) of pesticides. Thisobservation shows the major role of storm events in the mobi-lisation of pesticides. Pesticide losses during base flow appeared tobe of minor importance in comparison with losses during stormflow. A grab monitoring could take into account only partially atbest the rapid pesticide transfers. Thus, it results in non-repre-sentative average concentrations and loads, and in a higherprobability of mis-interpretation of the water quality status, asshowed as well by Holvoet et al. (2007). A combined monitoringsurvey during base flow and storm flow allowed to evaluate thetemporal variability in pesticide concentrations. Moreover, the totaldiuron flux could be derived with a good precision: indeed ourestimation reached 53.6 g, that is 72.5% of the ‘‘reference’’ value(Fig. 11). However, it corresponded to a high technical, material andanalytical cost investment.

3.4.2. Estimation of annual average and monthly average diuronconcentration

The annual average diuron concentration ranged between1.2 � 2.0 mg/L using weekly monitoring (4 samples per monthduring the application period, then one sample per month, that isn ¼ 35 samples), and 1.4 to 1.6 � 2.5 mg/L with monthly monitoring(n¼ 12) (Table 3). Thus, both sampling methods lead to comparableaverage and standard deviation results, as previously showed forflux estimates. Moreover, monthly average diuron concentrationswere calculated for August, using the data obtained through thedifferent monitoring strategies (grab and automatic weeklycomposite samples). Results showed that average diuron concen-tration ranged from 0.4� 0.4 mg/L using weekly grab sampling data

Table 4Comparison of calculated and estimated diuron flux (g) during floods at site 2.

Flood eve Calculateddiuron flux (g)

Watervolumes (m3)

Estimateddiuron flux (g)

Bias (%)

26 May 2007 0.68 348 0.80 1327–28 May 2007 6.91 2106 7.1 302 August 2007 0.89 335 0.70 �1906–07 August 2007 7.21 2263 7.70 621 August 2007 5.88 1970 6.60 1217 September 2007 7.33 1619 5.30 �2727 September 2007 1.01 621 1.70 73

14.2

22.7

4.9

41.2

5.1

10.7

4.7 5.6

48

53.6

73.9

0

10

20

30

40

50

60

70

80

90

100

W

lk

ee

2 ,1

y

,3

na

d4

-iB

no

m

3 d

na

1 yl

ht

m-iB

o

4 d

na

2 yl

htn

htn

oM

l

1 y

no

M

2 yl

ht

Mo

3 yl

htn

htn

oM

l

4 y

B

sa

fe

wol

otS

r

wolf

m

wolf

mrot

S +

wolf

es

aB

A

pm

oc

eg

are

v

tis

o

e

)g

( x

ulf

no

rui

d yl

ht

no

M

Grab surveysAutomatic +

grab surveys

Fig. 11. Monthly diuron flux (g) at site 2 in August 2007, estimated using differentmethods and monitoring data.

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748 747

(n ¼ 4) to 1.6 � 1.2 mg/L using average composite sampling (n ¼ 4).Combining monitoring during base flow and storm flow (using theestimated diuron flux, 53.6 g, cf. Fig. 11, and total water volumesconveyed during the month, 42300 m3) lead to an average diuronconcentration of 1.3 mg/L in August, in good agreement with theaverage concentration obtained using the composite weeklysampling. Obviously, grab monitoring, even at a weekly rate,underestimated the average concentration in the Morcille stream.

4. Conclusion

The present study reported the great spatial and temporalvariability of some herbicide and fungicide concentrations moni-tored during one year in an agricultural catchment devoted tovineyard. Their occurrence was closely related to the applicationcalendar and the hydrological conditions. Pesticide concentrationswere higher during the application season, with maximumconcentrations occurring during June and July. Concentrationslargely exceeded water quality criteria for the protection of aquaticlife. Pesticides were measured in surface water for several monthsafter the application and in the case of diuron, throughout the year.It pointed out their potential to persist in soils and shallowgroundwater. The loss of diuron in the catchment outflow wasevaluated at 0.57% of the applied amount, which is in agreementwith literature data. However, as shown by our study on the effectof sampling strategy, this value is certainly largely underestimateddue to grab sampling. Our study of pesticide behaviour duringfloods showed the major role of hydrological conditions in thetransport of pesticides. Indeed, floods contributed to more than89% of the total load of diuron transiting through the streamduring August 2007. These results provided further evidence thatfloods are the dominant pathways for the transport of pesticides ina small agricultural stream characterized by shallow soils andpreponderant surface flow pathways. Using comprehensive dataobtained in August, we showed that the estimated diuron fluxesand average concentrations were comparable using either weeklycomposite sampling (proportional to the flow), or with a combinedgrab sampling during base flow and automated sampling duringfloods.

A major conclusion of this study was that grab monitoring ofsurface water is not adapted for an accurate assessment of thecontaminant exposure level, especially in dynamic hydro-systemssuch as the Morcille stream, where it largely underestimates the

pesticides concentration and fluxes. These observations haveimportant consequences for the monitoring requirements ofpesticides, in the view of the implementation of the WFD.

Acknowledgments

This study was partly financed by the French regionalprogrammes (Rhone-Alpes region- LTER ‘‘Zone atelier bassin duRhone’’ and Cluster Environment) and the national researchprogrammes ANR-EcoGER (Papier) and EC2O (Padyma). We wouldlike to thank N. Ball and B. Motte for technical field assistance andC. Guillemain for analytical optimization and analyses of pesticides.Thanks are due to A. Albinet for helpful correction.

References

AFSSA, 2008. Direction du vegetal et de l’environnement, AGRITOX – Base dedonnees sur les substances actives phytopharmaceutiques. http://www.dive.afssa.fr/agritox/index.php (October 2008, online).

Bach, M., Huber, A., Frede, H.G., 2001. Input pathways and river load of pesticides inGermany – a national scale modeling assessment. Water Science and Tech-nology 43, 261–268.

Blanchoud, H., Moreau-Guigon, E., Farrugia, F., Chevreuil, M., Mouchel, J.M., 2007.Contribution by urban and agricultural pesticide uses to water contamination atthe scale of the Marne watershed. Science of the Total Environment 375,168–179.

Brown, C.D., Hodgkinson, R.A., Rose, D.A., Syers, J.K., Wilcockson, S.J., 1995. Move-ment of pesticides to surface waters from a heavy clay soil. Pesticide Science 43,131–140.

Capel, P.D., Larson, S.J., Winterstein, T.A., 2001. The behaviour of 39 pesticides insurface waters as a function of scale. Hydrological Processes 15, 1251–1269.

Chen, B., Huang, G., Li, Y., Zang, B., Huang, Y., 2005. Pesticide-loss simulation andhealth risk assessment during the flood season in watershed systems. WaterInternational 30, 88–98.

Claver, A., Ormad, P., Rodriguez, L., Ovelleiro, J.L., 2006. Study of the presence ofpesticides in surface waters in the Ebro river basin (Spain). Chemosphere 64 (9),1437–1443.

Clark, G.M., Goolsby, D.A., 2000. Occurrence and load of selected herbicides andmetabolites in the lower Mississippi River. Science of the Total Environment248, 101–113.

Cohn, T.A., 1995. Recent advances in statistical methods for the estimation ofsediment and nutrient transport in rivers. Review of Geophysical Supplement,1117–1123.

Couteux, A., Lejeune, V., 2007. Index Phytosanitaire Acta, 43rd ed. Acta, Paris.Domange, N., 2005. Etude des transferts de produits phytosanitaires a l’echelle de la

parcelle et du bassin versant viticole (Rouffach, Haut-Rhin). These de doctorat,Universite Louis Pasteur, Strasbourg 1, 327 pp.

Du Preez, L.H., Jansen van Rensburg, P.J., Jooste, A.M., Carr, J.A., Giesy, J.P., Gross, T.S.,Kendall, R.J., Smith, E.E., Van Der Kraak, G., Solomon, K.R., 2005. Seasonalexposures to triazine and other pesticides in surface waters in the westernHighveld corn-production region in South Africa. Environmental Pollution 135,131–141.

European Commission, 2000. Directive 2000/60/EC of the European Parliament andof the Council of 23 October 2000 establishing a framework for Communityaction in the field of water policy. Official Journal of the European Union L327,1–72.

European Commission, 2008. Directive 2008/105/EC of the European Parliamentand of the Council of 16 December 2008 on environmental quality standards inthe field of water policy, amending and subsequently repealing CouncilDirectives 82/176/EEC, 83/513/EEC, 84/156/EEC, 84/491/EEC, 86/280/EEC andamending Directive 2000/60/EC of the European Parliament and of the Council,Official J of the European Union, L 348, pp. 84–97.

Eyre, B.D., Pont, D., 2003. Intra- and inter-annual variability in the different forms ofdiffuse nitrogen and phosphorus delivered to seven sub-tropical east Australianestuaries. Estuarine Coastal Shelf Science 57, 137–148.

Gargouma, M., Blanchard, M., Chesterikoff, A., Ansart, P., Chevreuil, M., 1997.Seasonal transport of herbicides (triazines and phenylureas) in a small streamdraining an agricultural basin: Melarchez (France). Water Research 31,1489–1503.

Gouy, V., Nivon, C., 2007. Caracterisation et suivi de la qualite des eaux sur le bassinversant de la Morcille sur la periode 2001–2006. Rapport d’etude Cemagref-Chambre d’agriculture du Rhone, 81 pp.

Hildebrandt, A., Guillamon, M., Lacorte, S., Tauler, R., Barcelo, D., 2008. Impact ofpesticides used in agriculture and vineyards to surface and groundwater quality(North Spain). Water Research 42, 3315–3326.

Holvoet, K.M.A., Seuntjens, P., Vanrolleghem, P.A., 2007. Monitoring and modelingpesticide fate in surface waters at the catchment scale. Ecological Modelling209, 53–64.

M. Rabiet et al. / Environmental Pollution 158 (2010) 737–748748

Horowitz, A.J., 2008. Determining annual suspended sediment and sediment-associated trace element and nutrient fluxes. Science of the Total Environment400, 315–343.

House, W.A., Warwick, M.S., 1998. Intensive measurements of nutrient dynamics inthe River Swale. Science of the Total Environment 210–211, 111–137.

Johnes, P.J., 2007. Uncertainties in annual riverine phosphorus load estimation:impact of load estimation methodology, sampling frequency, baseflow indexand catchment population density. Journal of Hydrology 332, 241–258.

Kreuger, J., 1998. Pesticides in stream water within an agricultural catchment insouthern Sweden, 1990–1996. Science of the Total Environment 216, 227–251.

Kuivila, K.M., Foe, C.G., 1995. Concentration, transport and biological effects ofdormant spray pesticides in the San Francisco Estuary, California. Environ-mental Toxicology and Chemistry 14, 1141–1150.

Lennartz, B., Louchart, P., Andrieux, P., Voltz, M.,1997. Diuronand simazine lossesto runoffwater in Mediterranean vineyards. Journal of Environmental Quality 26, 1493–1502.

Leu, C., Singer, H., Stamm, C., Muller, S.R., Schwarzenbach, R.P., 2004. Simultaneousassessment of sources, processes and factors influencing herbicides losses tosurface waters in a small agricultural catchment. Environmental Science andTechnology 38, 3827–3834.

Louchart, X., Voltz, M., Andrieux, P., Moussa, R., 2001. Herbicide transport to surfacewaters at field and watershed scales in a Mediterranean vineyard area. Journalof Environmental Quality 30, 982–991.

Louchart, X., Voltz, M., Coulouma, G., Andrieux, P., 2004. Oryzalin fate and transportin runoff water in Mediterranean vineyards. Chemosphere 57, 921–930.

Margoum, C., Guillemain, C., Rabiet, M., Gouy, V., Coquery, M., 2007. Dissipation ofpesticides in surface water and biofilms in a small agricultural catchment:development of a methodology for studying environmental impact of pesti-cides. In: Proceedings XIII Symposium Pesticide Chemistry, Environmental Fateand Human Health, 3–6 September 2007, Piacenza, Italy.

Meybeck, M., 2005. Looking for water quality. Hydrological Processes 19, 331–338.Muller, K., Deurer, M., Hartmann, H., Bach, M., Spiteller, M., Frede, H.G., 2003.

Hydrological characterisation of pesticide loads using hydrograph separation atdifferent scales in a Gerlan catchment. Journal of Hydrology 273, 1–17.

Ng, H.Y.F., Clegg, S.B., 1997. Atrazine and metolachlor losses in runoff events from anagricultural watershed the importance of runoff components. Science of theTotal Environment 193, 215–228.

Phillips, P.J., Bode, R.W., 2004. Pesticides in surface water runoff on south-easternNew York State, USA: seasonal and stormflow effects on concentrations. PestManagement Science 60, 531–543.

Preston, S.D., Bierman, V.J., Silliman, S.E., 1992. Impact of flow variability on error inestimation of tributary mass loads. Journal of Environmental Engineering ASCE118, 402–419.

Rabiet, M., Margoum, C., Gouy, V., Carluer, N., Coquery, M., 2008. Transfert despesticides et metaux dans un petit bassin versant viticole – Etude preliminairede l’influence des conditions hydrologiques sur le transport de ces contami-nants, Ingenierie EAT, Special issue «Azote, phosphore et pesticides: Strategieset perspectives de reduction des Flux», pp. 65–75.

Ribolzi, O., Valles, V., Gomez, L., Voltz, M., 2002. Speciation and origin of particulatecopper in runoff water from a Mediterranean vineyard catchment. Environ-mental Pollution 117, 261–271.

Richards, R.P., Baker, D.B., 1993. Pesticide concentration patterns in agriculturaldrainage networks in the Lake Erie basin. Environmental Toxicology andChemistry 12, 13–26.

Riise, G., Lundekvam, H., Wu, Q.L., Haugen, L.E., Mulder, J., 2004. Loss of pesticidesfrom agricultural fields in SE Norway-runoff through surface and drainagewater. Environmental Geochemistry and Health 26, 269–276.

Schleppi, P., Waldner, P.A., Stahli, M., 2006. Errors of flux integration methods forsolutes in grab samples of runoff water, as compared to flow-proportionalsampling. Journal of Hydrology 319, 266–281.

Schulz, R., 2001a. Rainfall-induced sediment and pesticide input from orchards intothe Lourens River, Western Cape, South Africa: importance of a single event.Water Research 35, 1869–1876.

Schulz, R., 2001b. Comparison of spreaydrift- and runoff-related input of azinphos-methyl and endosulfan from fruit onchards into the Lourens River, South Africa.Chemosphere 45, 543–551.

Taillandier, P., 2008. Data treatment and study of the hydrological mechanism of theMorcille’s catchment area. Rapport de stage, Institut Polytechnique deGrenoble, Ecole d’ingenieurs pour l’energie, l’eau et l’environnement, 85 pp (inFrench).

Tomlin, C.D.S., 2000. The Pesticide Manual, 12th ed. British Crop Protection Council,Farnham, UK.

Vryzas, Z., Vassiliou, G., Alexoudis, C., Papadopoulou-Mourkidou, E., 2009. Spatialand temporal distribution of pesticide residues in surface waters in north-eastern Greece. Water Research 43, 1–10.

Zonta, R., Collavini, F., Zaggia, L., Zuliani, A., 2005. The effect of floods on thetransport of suspended sediments and contaminants: a case study fromthe estuary of the Dese River (Venice Lagoon, Italy). Environment 31,948–958.