- Email: [email protected]
Comparison of different sampling techniques for the evaluation of pesticide spray drift in apple orchards
The Science of the Total Environment 288 Ž2002. 199᎐213
Comparison of different sampling techniques for the evaluation of pesticide spray drift in apple orchards Olivier Briand a , Florence Bertrand a , Rene ´ Seux a,U , Maurice Millet a,b a
Ecole Nationale de la Sante´ Publique, Laboratoire d’Etude et de Recherche en En¨ ironnement et Sante´ (LERES), A¨ enue Professeur Leon ´ Bernard, 35043 Rennes cedex, France b Laboratoire de Physico-Chimie de l’Atmosphere de la Surface, et Departement de ` (UMR 7517), Centre de Geochimie ´ ´ Chimie de l’Uni¨ ersite ´ Louis Pasteur, 1 rue Blessig, 67084 Strasbourg cedex, France Received 27 March 2001; accepted 3 July 2001
Abstract An analytical method using gas chromatography᎐mass spectrometry has been developed for the evaluation of different sampling techniques to characterise spray drift in a commercial apple orchard. Eleven pesticides were studied Žfungicides, insecticides and herbicides.. A collection of airborne spray-drift pesticides released from a low-profile air-blast orchard sprayer was investigated using six types of samplers: Ž1. a Perkin-Elmer low volume automatic air sampler using with glass tube packed with Supelpak-2; Ž2. a high volume air sampler; Ž3, 4. an impinger containing cyclohexane that could be preceded by a glass fibre filter; and Ž5, 6. glass cartridges packed with Supelpak-2 that could be preceded by a glass fibre filter. Retention efficiencies of the different sampling techniques are compared, and physical forms of the retained pesticides are discussed. These techniques have allowed us to evaluate pesticide spray-drift in the orchard. Results have shown that the molecules’ properties Ž k H and vapour pressure. and weather conditions Žtemperature and relative humidity. strongly influence pesticide gas and particles distribution. However, in the studied orchard, it is difficult to differentiate pesticide spray-drift and post-application transfers since treatment duration was ) 2 days. 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Pesticides; Atmosphere; Sampling techniques; Spray Drift; GC-MS; Orchard
U
Corresponding author. E-mail addresses: [email protected] ŽR. Seux., [email protected] ŽM. Millet..
0048-9697r02r$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 1 . 0 0 9 6 1 - 5
200
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
1. Introduction The intensive use of pesticides has led to a ubiquitous contamination of remote environments because synthetic chlorinated hydrocarbon pesticides and industrial chemicals have been identified in the Arctic ecosystem ŽHargrave et al., 1987; Patton et al., 1989., the Antarctic environment ŽGiam et al., 1974; Risebrough and Carmignani, 1972., deep-sea organisms ŽBarber and Warlen, 1979; Ballschmiter et al., 1981., and a variety of other remote areas. The atmosphere is known to be a good route for the dissemination of semi-volatile organic compounds such as pesticides and PCBs ŽEisenrich et al., 1981; Bidleman and Leonard, 1982; Billings and Bidleman, 1983; Pacyna and Oehme, 1988; Oehme, 1991.. Pesticides enter the atmosphere through many processes such as application drift during spraying operations ŽPayne and Thompson, 1992; Watanabe, 1998., wind erosion of soil and volatilisation ŽGlotfelty et al., 1989; Nash, 1989; Foster et al., 1995; Kloppel and Kordel, 1997; Cherif and ¨ ¨ Wortham, 1997.. After entry into the atmosphere, pesticides are transported and distributed over long distances, sometimes far from their emission sites ŽReisinger and Robinson, 1976; Oehme and Mano, 1984; Larsson and Olka, 1989; Riley et al., 1989; Oehme, 1991; Hoff et al., 1992a,b., deposited through wet and dry processes ŽSeiber and Woodrow, 1995; Waite et al., 1995. andror degraded ŽAtkinson et al., 1999.. Spray drift corresponds to the fraction of pesticides which never reach the crop during treatment and are locally transported by the wind before deposition. Many factors influence this phenomenon: droplet diameter, spraying technique, land topography, climatic conditions, etc. Depending on literature values ŽVan der Werf, 1996., losses due to spray drift can vary from 1 to 30% of applied quantities. To determine the atmospheric contamination of the atmosphere by pesticides, many analytical methods have been developed ŽSherma and Shafik, 1975; Wehner et al., 1984; Seiber et al., 1990; Millet et al., 1996; Sanusi et al., 1997.. These methods employ filters and solid adsor-
bents for the sampling of particles and gas respectively by using high volume samplers. After sampling, a solvent extraction followed by a concentration step was made. Solvent extraction is accurate, but is generally long and increases detection limits due to losses induced by the different steps Žextraction, clean-up, concentration, etc.. An alternative is to use temperature as an extraction procedure. With thermal-desorption, not many steps are needed, which permits lower detection limits, the elimination of parasite peaks caused by solvents and opens up the possibility of automation ŽVan der Hoed and Halmans, 1987; McCaffrey et al., 1994.. Nevertheless, this method cannot be applied for trace analysis in ambient air, but can be successfully used for exposed area such as crops during treatment periods for assessing spray drift and volatilisation of post-application processes ŽClement et al., 2000; Briand et al., ´ 2001.. In a previous work ŽClement et al., 2000., an ´ analytical method using thermal desorptionrgas chromatography᎐mass spectrometry for the evaluation of the atmospheric level of pesticides caused by spray-drift during applications and volatilisation of post-application was developed. From experiments performed in this study, it appeared that some compounds, especially captan, used in apple orchards in Brittany at present, are sensitive to high temperatures since they seem to be degraded during the thermal-desorption process. Another factor to take into account was the low vapour pressure of these pesticides, which does not facilitate thermal extraction. From this observation, the aim of this work was to develop an analytical method using solvent extraction and gas chromatography᎐mass spectrometry for an evaluation of different sampling techniques in order to characterise spray drift in an orchard. The pesticides chosen were representative of fruit farming: captan, cyproconazole, cyprodinyl, formothion, iprodione, nuarimol, phosalone, simazine, tebufenpyrad and vamidothion. Advantages, limitations and validation of this method are presented and results of the different campaigns on a Britanny apple orchard are also presented and discussed.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
2. Materials and methods 2.1. Sampling sites The field experiment took place in a commercial apple and pear orchard at Domagne ´ in Brittany, France Ž30 km south-east of Rennes. from 17 March to 29 May, 1999. The field surface area is 60 ha: 45 ha of apples, and 15 ha of pears. The orchard is located in a zone of intensive corn and rape crops. Mature apple trees are planted in 3.50-m rows with 1.00-m tree spacing. Individual trees were 2.50-m high and 1.50-m wide. One type of sprayer and different nozzle configurations were used to produce a range of applied volumes, between 250 and 1000 lrha. The Holder NI 1000 used was a standard low-profile air-blast orchard sprayer which delivers droplets larger than 1 mm in diameter. Weather conditions were recorded for each experiment Žtemperature, relative humidity, speed and direction of wind. using a complete solution by Young ŽGroupe Leader, France. The environmental conditions during spraying were very different between March 1999; the air temperature ranged from 6 to 14⬚C, with a high relative humidity Ž) 85%.; and May 1999, air temperature ranged from 22 to 28⬚C, with a lower relative humidity Ž- 75%.; wind speeds ranged from 2 to 9 m sy1 , in different directions. During a final campaign, which was performed in September 2000, the temperature ranged from 25 to 41⬚C, with a relative humidity - 60% and a wind speed lower than 1 m sy1 . 2.2. Sampling procedure Many kinds of sampling methods were employed to evaluate pesticide losses by spray-drift immediately during and after applications. The first method used a Perkin-Elmer low volume automatic air sampler ŽNorwalk, CT, USA., which was modified to permit a flow rate between 1 and 2 l miny1 Žagainst 0.05᎐0.5 l miny1 initially .. This kind of collector allowed the sequential automatic sampling of 24 tubes. Tube sampling time can be programmed, and was generally fixed to 2 h. A
201
pump SKC 224 44 ŽArelco, France., which permits controlled lower flow rate between 0.01 and 5 l miny1 , was used with this collector. A second method using an ‘Explorer high volume automatic air sampler’ ŽZambelli, Italy., was used at a flow rate of 40 l miny1 Ž70 l miny1 maximum permitted.. It allowed sequential automatic sampling of eight samples. Flow rate, volume and temperature were recorded for every sample. Two other methods using an impinger with cyclohexane RS ŽCarlo Erba, Val de Reuil, France. and low volume glass tubes, associated with a NMP 50 pump ŽNeuberger, Germany. at 1.45 l miny1 flow rates, were used in a specific way during the last three campaigns. These samplers could be preceded by glass fibre filters ŽMillipore AP40, 25 mm in diameter. mounted on IOM inhalable dust sampler ŽSKC, Dorset, USA.. All samplers were placed, depending on the campaigns, at a height between 1.00 and 1.50 m. 2.3. Preparation of adsorbent tubes Field experiments were carried out by using glass tubes Ž4 mm i.d.=89 mm. and packed with Supelpack 䊛 2 resin obtained from Perkin-Elmer Corp. ŽNorwalk, CT, USA. and Supelco ŽBellefonte, PA, USA. respectively. Each cartridge was hand-made and contained the same amount of resin Ž125 mg., kept in position with glass wool, Pesticide grade ŽSupelco.. Before use, the resin was washed for 24 h in a Soxhlet, with a mixture of n-hexanerdiethyl ether Ž90:10.. Before use, the tubes were closed with Teflon 䊛 caps and stored in the dark. For the Explorer, which can simultaneously collect the particulate and gaseous phase, using glass Ž ⌽ s 47 mm GF-Filters. fibre filters ŽMillipore, France. and 2 g Supelpack-2 resin, respectively; both were washed for 24 h in a Soxhlet with a mixture of n-hexanerdiethyl ether Ž90:10., and stored before use in the dark in clean capped glass bottles. IOM glass fibre filters were cleaned and stored in the same manner as those used for the Explorer.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
202
Table 1 Physical-chemical properties of the studied pesticides ŽIn: Tomlin, 1997 and Index Phytosanitaire 1999.. Molecules
Chemical family
Action
Molecular weight Žg moly1 .
KH ŽPa m3 moly1 .
Vapour pressure ŽmPa.
Solubility Žwater. Žmg ly1 .
Kow log P
Captan Cyproconazole Cyprodinyl Formothion Iprodione Nuarimol Phosalone Simazine Tebufenpyrad Vamidothion
N-trihalomethylthio Triazole Anilinopyrimidine Organophosphorous Dicarboximide Pyrimidinyl carbinol Organophosphorous 1,3,5-triazine Pyrazole Organophosphorous
Fungicide Fungicide Fungicide Insecticide Fungicide Fungicide Insecticide Herbicide Acaricide Insecticide
300.6 291.8 225.3 257.3 330.2 314.7 367.8 201.7 333.8 287.3
1.07= 10y3 7.2= 10y5 6.6 at 7.2= 10y3 1.1= 10y5 1.27= 10y4 3.3= 10y5 7.4= 10y3 5.6= 10y3 - 1.28= 10y3 Negligible
1.17= 10y2 3.46= 10y2 5.1= 10y1 0.113 5 = 10y4 - 0.0027 - 0.06 2.94= 10y3 - 10y2 Negligible
3.3 140 13 2600 13 26 3.05 6.2 2.6 4.1= 10 6
2.8 2.91 3.9 ? 3.1 3.18 4.01 2.5 4.61 1.18
2.4. Standard solutions of pesticides Guaranteed pesticides: captan Ž98.4%.; cyproconazole Ž99.4%.; 3,5 dichloroaniline Ž98.0%.; formothion Ž74.5%.; iprodione Ž99.9%.; nuarimol Ž96.5%.; phosalone Ž98.5%.; simazine Ž96.2%.; and vamidothion Ž96.5%. were obtained from Cluzeau Info Labo ŽBordeaux, France.. Stock solutions Ž1 g ly1 . were prepared in toluene Pestipur 䊛 ŽSDS, Peypin, France. or in acetone ŽSDS, Peypin, France. for simazine. Guaranteed cyprodinyl and tebufenpyrad standard were obtained in iso-octane at 10 ng ly1 from Cluzeau Infos Labo ŽBordeaux, France.. All dilutions were made using n-hexane Pestanal TM ŽRiedel de Haen, ¨ Seelze, Germany.. Due to their conditioning, cyprodinyl and
tebufenpyrad were added directly to the final mixture. Physical᎐chemical properties of the pesticides studied are presented in Table 1. 2.5. Analytical method A Hewlett-Packard 5890 gas chromatograph fitted with a Hewlett-Packard 5970B quadruple mass spectrometer was used. The analytical column was a dimethylpolysiloxane DB-1 Ž30 m= 0.249 mm i.d.; film thickness s 0.25 m. ŽJ & W Scientific. and the carrier gas was Helium N55 at a flow rate of 1 ml miny1 , which induced a head pressure of 9 Psi. The GC temperature program began at 60⬚C for 1 min, increased at 20⬚C miny1 to 150⬚C, increased at 2⬚C miny1 to 200⬚C and finally increased at 5⬚C miny1 to 260⬚C Žheld for
Table 2 Ions Ž mrz ., relative abundance and retention time for each pesticide studied Molecules
Ions Ž mrz .
Relative abundance Ž%.
Retention time
3,5 Dichloroaniline Simazine Formothion Captan Cyprodinyl Vamidothion Cyproconazole Nuarimol Iprodione Tebufenpyrad Phosalone
161r163 201r186r173 125r93 79r149 224r225 87r145 222r139 314r235r203 314r316r187 318r333r276 182r367r184
100r68 100r58r41 100r100 100r22 100r61 100r43 100r46 100r100r81 100r67r21 100r78r46 100r56r33
8.1 min 14.7 min 17.6 min 25.0 min 25.4 min 28.7 min 31.9 min 36.3 min 38.1 min 39.6 min 40.0 min
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
5 min.. Detection was accomplished by using electron impact ionisation, source temperature s 250⬚C, energy s 70 eV, in SIM mode Žselected ion monitoring.. Two or three ions were chosen for each pesticide and their mrz ratios are presented in Table 2. Identification was based on the relative abundance of these 2᎐3 ions for each molecule ŽTable 2.. Quantification was done on the total ion chromatogram. Calibration curves were obtained by the analysis of five mixtures of the 11 pesticides with respective concentrations ranging from 0.1 to 2.0 g mly1 . Five replicate measurements were made for 2 g mly1 solution, in order to determine relative uncertainty. The sensibility and accuracy of the method were determined and verified. Values of detection limits ŽDL. were determined by the injection of decreasing concentrations of the pesticide mixture. DL were first expressed in ngrl, and are then converted to ngrm3, after considering an average volume pumped during sampling of 500 l and a final volume after extraction of 500 l. 2.6. Resin extraction procedure Resin and glass wool, which have retained pesticides during the sampling procedure, were transferred in a 10-ml vial, and the tube was rinsed with 5 ml with a mixture of n-hexanerdiethyl ether Ž90r10., which was used for extraction. They were subsequently extracted for 15 min in a Branson 2000 ultrasonic bath ŽFisher, Elancourt, France.. Tubes extractions were undertaken during the 24 h following sampling, and stored in the dark at 4⬚C until analysis. Extracts were filtered on green filter, No 111 ŽDurieu, France., the vial was then rinsed with another 1 ml of the mixture and this phase was added to the first one. Extracts were then evaporated until droplet by a N-Evaporator thermostated at 50⬚C, readjusted to 0.5 ml with n-hexane and analysed with GCrMS. To determine recovery, five tubes were spiked with 0.5 ml of the pesticides mixture at 2 mg ly1 and extracted with the same procedure as the one previously described. A second extraction was re-
203
alised on the same vial content to confirm extraction efficiency. This extraction procedure was also applied to filters, but no recovery was determined since no pesticides were detected during the second extraction step. 2.7. Resin retention efficiency An impinger head was adapted to a 250-ml round glass flask and a mixture of pesticides was transferred in the flask. Two sorbent tubes in tandem were connected to the exit of the impinger. The flask was then heated in a GC oven Ž190⬚C for 2 h.. Simultaneously, pesticides were carried under a flow of clean air to the first sorbent tube Žefficiency tube.. In the effort to collect pesticides not retained by the efficiency tube andror those which could be released during sampling after first retention, we used a second tube Žbreakthrough tube. at the end of the first one. For each experiment, the same amount of air was delivered through tubes. After heating, the flask was rinsed in order to determine the volatilisation percentage, expressed as a ratio between the initial quantity of pesticides introduced in the flask Žnoted QP. minus those which were in the flask at end of the experiment on QP. Retention percentage was expressed as the ratio of the quantity of pesticides in the efficiency tube Žconsidering extraction recovery as determined previously. on the quantity of volatilised pesticides. Breakthrough percentage was expressed in the same way considering the ratio of the quantity of pesticides in the second tube on the quantity of volatilised pesticides.
3. Results and discussion 3.1. Validation of the analytical method A calibration curve was determined with accuracy for each compound, with determination coefficients R 2 ranging from 0.96 to 0.99 ŽTable 3.. Difficulties were found with vamidothion since LD was higher than those obtained for the other
204
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
Table 3 Characteristics of the analytical method Molecules
Concentrations range Žmg ly1 .
R2
D.L. Žmg ly1 . in vials
D.L. Žng my3 . in air
3,5 dichloroaniline Captan Cyproconazole Cyprodinyl Formothion Iprodione Nuarimol Phosalone Simazine Tebufenpyrad Vamidothion
1.0᎐0.05 2.5᎐0.3 2.0᎐0.5 2.0᎐0.1 2.0᎐0.1 2.0᎐0.25 2.0᎐0.5 2.0᎐0.25 2.0᎐0.1 2.0᎐0.1 3.0᎐0.5
0.999 0.987 0.999 0.997 0.999 0.995 0.958 0.994 0.999 0.996 0.900
0.025 0.15 0.25 0.05 0.05 0.125 0.10 0.125 0.05 0.05 0.50
50 300 500 100 100 250 200 250 100 100 1000
compounds. That can be explained by the mass spectrum of vamidothion since this molecule is broken down into a large number of ions with neighbouring relative abundance which complicate its detection in SIM mode. DL values of 25 and 250 ngrm3 Žexcept for vamidothion. were obtained, which permit monitoring the levels of pesticides in the case of spray-drift releases. The different ranges used during the study did not show any significant variation of signal responses obtained in GC-MS, except for iprodione as its signal rapidly decreases. Fresh solutions were made each week. This problem was solved by storing solutions at y18⬚C, rather than q4⬚C. 3.2. Extraction reco¨ eries Extraction efficiencies were determined after considering all the steps of the extraction procedure Žextraction and N-evaporation.. The results are presented in Table 4. The recovery percentage ranged between 11 and 103% with a relatively high R.S.D. Ži.e. formothion, 79% - 103% 132%.. Recoveries obtained by Martinez-Vidal et al. Ž1997. for organochlorine pesticides on Supelpack and sonication extraction are in the same order of magnitude. For 3,5 dichloroaniline, iprodione and vamidothion, poor recoveries can be explained by their lower stability and by the analytical difficulties previously described. Losses observed for other
pesticides are not occurred during extraction step since no pesticide was detected after a second extraction of the resin. Nevertheless, high R.S.D. are probably due to the N-evaporation step. On account of the high R.S.D. obtained for the recovery percentage, no correction has been made to field results. 3.3. Resin retention efficiency The first part of the experiment consisted in optimising the volatilisation step. Shape of the impinger, temperature and time for volatilisation were chosen to obtain a volatilisation percentage approaching 100%. We retained a round flask of Table 4 Extraction recoveries Ž%. for the 11 pesticides studied Ž n s number of experiments. Pesticides
Mean recoveries Ž%.
Lowest recoveries Ž%.
R.S.D. Ž%.
3,5 Dichloroaniline Ž n s 3. Simazine Ž n s 5. Formothion Ž n s 5. Captan Ž n s 5. Cyprodinyl Ž n s 5. Vamidothion Ž n s 4. Cyproconazole Ž n s 5. Nuarimol Ž n s 5. Iprodione Ž n s 4. Tebufenpyrad Ž n s 5. Phosalone Žn s 5.
51 98 103 66 104 11 73 102 54 100 74
43 87 79 55 90 5 63 89 44 85 66
15 10 22 16 10 38 16 11 21 13 11
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
100 ml for impinger, and 190⬚C for 80 min at a flow rate of 3 l miny1 . Through the different temperatures investigated, it appeared that responses obtained for vamidothion decreased significantly with an increase of temperature, degradation probably occurred in the flask during the heating step. Different solvents, for reference mixed standard, were also investigated Ž n-Hexane, propan-2ol and acetone.; no significant variation was observed in volatilisation percentage. A system to maintain the efficiency tube and breakthrough tube between 25 and 30⬚C was added to the initial device, using a stream of compressed air Ž7 bars, 70 l miny1 . to sweep the external surface of the tubes. Without this system, tube temperature can reach 70᎐80⬚C and a breakthrough percentage ) 10% was observed. Resin efficiency percentage ranged between 60 and 100% according to recovery efficiencies previously determined. Lower resin efficiency was obtained for pesticides with higher recovery R.S.D. 3.4. Field experiments All sampling campaigns were performed in the ‘la Claye orchard’, east of Rennes Žwest of France.. Both meteorological conditions and the farmer’s pesticide application program were taken
205
into account to determine the sampling campaigns. The kind and quantities of pesticide applied to the orchard in 1999 is presented in Table 5. 3.4.1. Sampling campaigns on 17 March and 7 April Sampling on 17 March corresponds to the first herbicide treatment. This treatment was undertaken in order to remove grass at the bottom of apple trees. For this treatment, a mixture of simazine Ž1000 g hay1 ., glyfosinate Ž800 g hay1 . and diuron Ž1500 g hay1 . was sprayed at a height of 30 cm on approximately 20% Ž12 ha. of the total surface of the orchard. The collector was placed in the wind at a 80-cm height and 7 m from the end of one line of apple trees. Sampling was made every 2 h between 11.30 and 17.30 h at a flow rate of 0.5 l miny1 . After extraction and analysis of each tube, simazine was not detected. In order to verify that failure to simazine has no relation to the detection limit of the method, samples were also analysed by using a large volume injection GC-NPD, which allows a greater sensitivity. Once again, no simazine was detected; hence, the absence of this compound is not a detection limit problem. However, in the second tube Ž13.30᎐15.30 h., iprodione was detected and quantified Ž890 ng my3 ., although this fungicide was not applied in the orchard before or during the sampling campaign ŽFig. 1.. The origin of the fungicide has been
Table 5 Program extract of pesticides treatment in the apple orchard in 1999 Commercial name
Pesticide
Dose applied Žg hay1 .
Frequency of use
Topaze effervit
Captan Penconazole Iprodione Captan Cyproconazole Tebufenpyrad Simazine Glyfosinate Diuron Cyprodinyl Vamidothion Cyproconazole
12.5 50 1000 900 12 500 1000 800 1500 56.25 200 50
4 applicationsryear
Rovral Atemi C Masaı¨ Baral
Chorus Kilval Alto
1 applicationryear 2 applicationsryear 1 applicationryear 2 applicationsryear
2 applicationsryear 1 applicationryear 2 applicationsryear
206
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
Fig. 1. Atmospheric concentrations of iprodione measured during campaigns from 17 March and 7 April 1999.
identified as an application performed in a neighbouring rape parcel situated at 3 m from the collector. The application was made at a 1 m height at 1000 g hay1 . The strong, but punctual contamination observed is the consequence of spray drift during application. A second sampling campaign was made on 7 April when captan Ž900 g hay1 . and cyproconazole Ž12 g hay1 . were applied in the orchard using a Holder NI 1000 sprayer. The collector was placed in the wind at a 80-cm height and sampling was at 1 l miny1 every 2 h between 10.50 h, 04r07 and 10.50 h, 04r08. No pesticide applied in the orchard was detected in tubes, but in the second one Ž12.50᎐15.50 h. iprodione was detected and quantified at 310 ng my3 ŽFig. 1.. The origin of the presence of iprodione in the second tube was explained after discussion with the rape farmer. He confirmed that iprodione was applied to the rape parcel during the period where sampling was done. This application corresponds to a complementary treatment. In the first sampling campaign, spray-drift contamination was identified. Lower concentrations than in the experiment of 04r07 can be explained by the localisation of the collector. It was placed 300 m from the rape parcel, so deposition of pesticide aerosols and dilution of the pollution was higher than in the first campaign. 3.4.2. Sampling campaigns on 8 and 9 April These campaigns were made during a treat-
ment program which started on 7 April Žsee previous paragraph. until 9 April. Indeed, 3 days or more are necessary for the pesticide application on the total orchard Ž60 ha.. Collectors were not moved during the totality of the sampling campaign Žheight, emplacement and flow rate. and 12 tubes were used in order to perform sampling over 24 h Ž2 h per tube.. After extraction and analysis of the tubes, no captan or cyproconazole was detected in any tube, although these pesticides were applied in the orchard. However, simazine was detected in two tubes Ž1.50᎐3.50 h and 5.50᎐7.50 h. at concentrations of 3.96 and 0.10 g my3 , respectively ŽFig. 2.. Two hypotheses concerning this result can be made: 1. Simazine could come from the orchard. Indeed, an application of simazine was made in mid-March and strong wind during the night of sampling could re-suspended soil dust containing simazine which was collected. A short rainy event occurred during the night sampling and could have laid contaminated dust onto the sampler; or 2. Simazine could come from an unidentified agri- or non-agricultural Žroads or railways. treatment performed outside the orchard. The agricultural origin of simazine seems more probable since Chevreuil et al. Ž1996. have shown rainwater contamination by simazine Žup to 0.65 g ly1 . in the Seine basin during
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
207
Fig. 2. Concentrations of simazine Žexpressed in ng.my3 . measured during campaign on 8᎐9 April 1999.
the same period ŽApril. in 1996. However, it is difficult to attribute concentrations found in the two tubes to spray-drift during application, since simazine was measured very early in the morning of 9 April. Whatever the origin of simazine, post-application volatilisation processes correlated with an increase of humidity at the beginning of the morning ŽJury et al., 1984. and followed by wind transport can be envisaged in order to explain the simazine concentration found in the second tube, but not in the first one. Indeed, this concentration is too high to be explained by volatilisation processes. 3.4.3. Sampling campaigns on 22 and 27᎐28 April On 22 April, a fungicide treatment using captan Ž900 g hay1 ., cyproconazole Ž12 g hay1 . and iprodione Ž1000 g hay1 . was undertaken. Sampling conditions were the same as for the previous campaign. The collector was placed at a 1.50m height in the wind and sampling was performed at 14.25 h for 2.5 h Žflow rate: 0.8 l miny1 .. On 27 and 28 April, an insecticide and fungicide treatment with phosalone Ž135.125 g hay1 .,
cyprodinyl Ž56.25 g hay1 . and iprodione Ž1000 g hay1 ., respectively, was also made. The collector was placed in the wind 100 m from the orchard at a 1.50-m height. For this campaign, the collector was situated in an area where it could receive a representative part of the wind which sweeps the orchard. After extraction and analysis of tubes corresponding to these two campaigns, no pesticide was found. 3.5. Discussion From all the campaigns presented here, it appears that the sampling method used seems not to be suitable for the evaluation of spray drift during pesticide treatment in the orchard. From results obtained, two hypotheses can be advanced: 1. Pesticides are present in the atmosphere at very low concentrations. The method, which was developed, is not sensitive enough to detect these pesticides. Two phenomena can explain this hypothesis. ` Pesticides are applied at low doses and
208
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
sampled volumes are too low with respect to detection limits. ` Spray-drift in the orchard is extremely low. The spraying technology, especially droplet size spectrum, which was used in the orchard can explain this. Indeed, the Holder NI 1000, which was used, delivers droplets more than 1 mm in diameter, which impact rapidly on the apple trees’ leaves or on the soil. Droplet size does not favour spray-drift, unlike droplets produced by sprayers used in large crops Žrape, maize, wheat, etc... With this kind of sprayer, droplets diameters are generally - 300 m, and so favour aerosol production which sediments slowly and can be then transported by the wind out of treatment areas. 2. The physical form, under which pesticides are emitted to non-target areas through the atmosphere Žaerosols, particles, gas phase., is not efficiently collected. The low application dose hypothesis, even if it is possible, cannot totally explain why no pesticides were detected, since compounds like captan and iprodione, which are used at higher rates, were also never detected. Furthermore, iprodione was detected in the case of an application on a large crop Žsee sampling campaign of 17 March and 7 April.. From the results obtained during the different campaigns, the second part of the first hypothesis seems to be a more suitable explanation. Indeed, pesticides detected in tubes came from spray drift from large crops treated with sprayers delivering small droplets. Concerning the last hypothesis, the sampling method is not efficient enough to collect pesticides present in the atmosphere in particulate phase. In order to check these hypotheses, another sampling campaign was undertaken in the orchard on 27 May, during two simultaneous pesticide treatments. These treatments were: 䢇
tebufenpyrad Ž500 g hay1 ., 1000 l hay1 were applied;
䢇
vamidothion Ž200 g hay1 . and cyprodinyl Ž56.25 g hay1 ., 250 l hay1 were applied.
For this campaign, three different sampling techniques were simultaneously performed. 䢇
䢇
䢇
The first consisted of an impinger containing 50 ml of cyclohexane, flow rate was fixed at 2 l miny1 and this device was used for the sampling of pesticides in the gas and aerosol phases. The second apparatus constituted a HighVolume Sampler which can separately collect the particulate Žon ⌽ s 47-mm GF-Filters. and gaseous Žwith 2 g Supelpack-2 resin. phases at a flow rate of 40 l miny1 . The third sampler, used for the collection of the gas phase, consisted of two sampling tubes containing 125 mg of Supelpak-2 resin placed in series Žthe second tube was used in order to underline an eventual loss from the first one.. The flow rate was 0.25 l miny1 .
Each sampler was placed at a 80-cm height in the wind and sampling was done over 6 h Žstarting at 11.30 h., which corresponds to the duration of the application of pesticides. Results obtained are illustrated in Fig. 3. For the Hi-Vol. sampler, the two pesticides are detected on the filter, which is assumed to be the particulate phase at concentrations of 87, 1.4 and 0.2 ng my3 for cyprodinyl, vamidothion and tebufenpyrad, respectively. In addition, 9.4 ng my3 of captan were also detected. This concentration is probably the result of a previous application in the orchard, and resuspended with soil particles. Pesticides, which are detected in the Supelpack-2 resin, and assumed to be the gas phase, are lower than on filter. Indeed, only 3.4 ng my3 were measured for cyprodinyl, while no trace of vamidothion and tebufenpyrad was found. Concerning sampling tubes, no loss was observed in the second one Ž90 l of air collected.. The first tube contains 4 ng my3 of cyprodinyl while no other pesticide was detected. Finally, for the impinger, all pesticides used during the sampling campaign were found in vari-
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
209
Fig. 3. Comparison of sampling methods Žconcentrations expressed in ng my3 ..
able concentrations in the solvent Ž21 ng my3 of cyprodinyl, 4.7 ng my3 for tebufenpyrad and traces of vamidothion.. The low concentrations underlined in this last campaign and the predominance of pesticides in the particulate phase, seem to demonstrate that spray-drift is limited in the orchard and the droplet size produced during pesticide application probably explains this observation. Large droplets induce a fast sedimentation onto leaves or soil and so minimise spray-drift in the orchard. Pesticides detected on the filters, associated with wind eroded dust, probably come from eolian erosion of treated trees or soils. This phenomenon is increased by agricultural engines’ passage between the trees. These observations explain why no pesticides were detected during the first campaigns, as the method was not sensitive enough and was not adapted to collect a particulate phase.
Low volume samplers or impingers are not suitable for the evaluation of pesticide atmospheric transfers, associated with wind eroded dust. Flow rates are too low to permit efficient particle sampling. The low concentrations observed in the Supelpak-2 tubes can be explained by the evaporation of spraying droplets containing pesticides during application or by volatilisation from treated surfaces, since high temperature Ž33⬚C. occurred during this campaign. The concentrations found are in accordance with the potential volatility of pesticides since cyprodinyl Ž H s 6.6᎐7.2.10y3 Pa m3 moly1 ., the more volatile, is detected at higher concentration than tebufenpyrad Žapprox. 10 times less volatile; H - 1.28= 10y3 . and vamidothion which has a negligible Henry’s law constant. It is important to note that concentration measured with Supelpak-2 tube is in accordance with
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
210
those obtained with the Hi-Vol. sampler Ž3.4 vs. 4.0 ng my3 .. The measurement of atmospheric concentration of cyprodinyl in the gas phase seems to be effective since the same concentrations were found using two different sampling techniques. Concentrations obtained with the impinger are in accordance with the volatility of pesticides, but differences observed with values from Hi-Vol. could be explained by partial dissolution of particulate pesticide in the solvent. This phenomenon of dissolution could overestimate concentration in the gas phase and underestimate concentration in the particulate phase since adhesion of the aerosols can occur in the bubbling tube of the impinger. 3.6. Next experiments In order to confirm previous observations, a sampling campaign was undertaken in the orchard on 14 September, 2000 Ž11.00᎐15.00 h. during phosalone treatment Ž125 grha, 250 lrha were applied.. During this campaign, four different collectors were placed at 1.50 m from ground at approximately 100 m in the wind of the orchard, which was low during sampling Ž- 1 m sy1 .. These collectors were: 1. An impinger with cyclohexane preceded by a
filter maintained at 15⬚C in a water bath Ž1.45 l miny1 .; 2. a sampling tube with Supelpak-2 preceded by a filter Ž1.45 l miny1 .; 3. a sampling tube with Supelpak-2 without filter Ž1.45 l miny1 .; and 4. Hi-Vol. sampler Zambelli equipped with a filter and resin Supelpak-2 Ž40 l miny1 ; 9.7 m3 .. The results obtained for the three first collectors are illustrated Fig. 4, and it appears that total pesticides concentrations observed Žparticulate and gaseous phases. are comparable for collectors Ž1. and Ž2.. Measurements with the Hi-Vol. sampler Ž4. are not totally available since a loss of resin occurred in the laboratory during extraction. However, the filter was analysed and the results obtained show concentrations of 19, 2 and 100 ng my3 for captan, cyprodynil and phosalone, respectively. Low concentrations obtained for phosalone vs. collectors Ž1. and Ž2. could be explained by the association of high flow rate and temperature, which could transfer pesticides from particles collected on filters to the gas-phase. The analysis of resin should demonstrate underestimation of the particulate phase. This phenomenon is well known when using high volume samplers ŽSanusi et al., 1999.. It appears that tubes are effective only for the
Fig. 4. Atmospheric concentrations of phosalone Žng my3 . in the different samplers used on 14 September, 2000.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
collection of the gas phase since concentrations detected in the tube without a filter are comparable to the concentrations obtained in the tube following a filter. This observation can be explained by the position of the entrance of the tube, which is vertically positioned, whereas the dust sampler is mounted horizontally with respect to the ground. Indeed, for this last device, particle sedimentation occurred on the filter and is favoured by flow rate, whereas in the first tube, sedimentation is against active pumping. Gas-particle partitioning, as it has been described previously, must be considered with many precautions. Indeed, with collectors Ž1. and Ž2., it appears that, even if total concentrations are comparable, strong partitioning differences between the two devices were underlined. However, the explanation of high concentration detected in cyclohexane could be due to the solubilisation of pesticides from the diffusion of solvent vapours through the filter. Concentrations measured, especially in the gas phase, are high in comparison to those obtained in previous campaigns. This difference could be explained by the relatively high temperature, which occurred during the campaign Ž41⬚C.. This campaign cannot confirm previous hypotheses in which spray-drift is negligible in the orchard. Nevertheless, relatively high concentrations obtained in the gas phase could be explained by sampling of fine aerosols ᎏ not retained on filters ᎏ coming from evaporation of droplets consecutively at high air temperature.
4. Conclusions The collection of airborne spray-drift pesticides released from a low-profile air-blast orchard sprayer and analysis using gas chromatography᎐mass spectrometry has been developed for the evaluation of different sampling techniques to characterise spray drift in a commercial apple orchard. Five types of sampler: Ž1. a Perkin-Elmer low volume automatic air sampler used with glass tube packed with Supelpak-2; Ž2. a high-volume air sampler; Ž3, 4. an impinger containing cy-
211
clohexane that could be preceded by a glass fibre filter; and Ž5, 6. glass cartridges packed with Supelpak-2 that could be preceded by a glass fibre filter were investigated for: 䢇
䢇
䢇
their capacity to be used for evaluation of spray-drift; their efficiency to collect pesticides in the atmosphere; and the influence of factors such as ambient humidity, temperature, amount of solid sorbents and sampling flow rate.
From experiments performed, spray-drift in the orchard does not support horizontal transfer of pesticides droplets, by the wind during treatment, but rather by post-application transfer of pesticides associated to wind-eroded dust. This phenomenon seems to be influenced by the spraying technique, as droplets produced in orchards are larger than those delivered by conventional sprayers Žgreat crops or mosquito control.. These droplets can rapidly sediment onto leaves or soils and act to reduce spray-drift. However, it has been observed that climatic conditions could invert this phenomenon, especially air temperature. Increase of temperature is in favour of the evaporation of droplets, and consequently, reduces their size. These smaller droplets can be more easily transported outside the orchard by the wind.
Acknowledgements This work was supported by the French Minis tr y o f E n v ir o n m e n t Ž c o n tr a c t n o . DGADrSRAEr97143.. The authors also wish to thank Mr. Olivier Ferron from SRPV-Rennes ŽFrance. for his help and Mr. Albert Navarre from la Claye orchard to have permitted us to perform air sampling on his domain. References Atkinson R, Guicherit R, Hites RA, Palm W-U, Seiber JN, de Voogt P. Transformation of pesticides in the atmosphere: a state of the art. Water Air Soil Poll 1999;115:219᎐243.
212
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
Barber RT, Warlen SM. Organochlorine insecticide residues in sea fish from 2500 m in the Atlantic Ocean. Environ Sci Technol 1979;13:1146᎐1148. Ballschmiter K, Buchert H, Bihler S, Zell M. Baseline studies of global pollution: IV The pattern of pollution by organochlorine compounds in the North Atlantic as accumulated by fish. Fresenius Z Anal Chem 1981;306:323᎐339. Bidleman TF, Leonard R. Aerial transport of pesticides over the Northern Indian Ocean and adjacent seas. Atmos Environ 1982;16:1099᎐1107. Billings WN, Bidleman TF. High volume collection of chlorinated hydrocarbons in urban air using three solid adsorbents. Atmos Environ 1983;17:383᎐391. Briand, O, Clement, M, Seux, R, Millet, M, 2001. Field experi´ ment for assessing atmospheric transfer of pesticides during and after application. Part 1: Analysis of pesticides by Adsorptionrthermal desorption-GCrMS. Environ Sci Technol, submitted. Chevreuil M, Gargouma M, Teil MJ, Chesterikoff A. Occurrence of organochlorines ŽPCSs, pesticides. and herbicides Žtriazines, phenylureas. in the atmosphere and in the fallout from urban and rural stations of the Paris area. Sci Total Environ 1996;182Ž1᎐3.:25᎐37. Cherif S, Wortham H. A new laboratory protocol to monitor the volatilization of pesticides from soil. Inter J Environ Anal Chem 1997;68:199᎐212. Clement M, Arzel S, Le Bot B, Seux R, Millet M. Adsorp´ tionrthermal desorption-GCrMS for the analysis of pesticides in the atmosphere. Chemosphere 2000;40Ž1.:49᎐56. Eisenrich SJ, Looney BB, Thornton JD. Airborne organic contaminants in the Great Lakes Ecosystem. Environ Sci Technol 1981;15:30᎐38. Foster P, Ferrari C, Turloni S. Environmental behaviour of herbicides. Atrazine volatilisation study. Fresenius Environ Bull 1995;4:256᎐261. Giam CS, Richardson RL, Wong MK, Sackett WM. Polychlorinated biphenyls in Antartica Biota. Antartica J US 1974; 8:303᎐305. Glotfelty DE, Leech MM, Jersey J, Taylor AW. Volatilization and wind erosion of soil surface applied atrazine, simazine, alachlor, and toxaphene. J Agric Food Chem 1989; 37:546᎐551. Hargrave BT, Vas WP, Erickson PE, Fowler BR. Supply of atmospheric organochlorines to food webs in the Arctic ocean. Sixth Int. Symposium of the Comm. for Atmos. Chem. and Global Poll. on Global Atmospheric Chemistry, Peterborough, Ontario, August, 1987:23᎐29. Hoff RM, Muir DCG, Grift NP. Annual cycle of polychlorinated biphenyls and organohalogen pesticides in air in Southern Ontario. 1. Air concentration data. Environ Sci Technol 1992a;26:266᎐275. Hoff RM, Muir DCG, Grift NP. Annual cycle of polychlorinated biphenyls and organohalogen pesticides in air in Southern Ontario. 2. Atmospheric transport and sources. Environ Sci Technol 1992b;26:276᎐283. Index Phytosanitaire, 35th edition, 1999. ACTA Editions, Paris.
Jury WA, Spencer WF, Farmer WJ. Behaviour assessment model for trace organics in soil: IV. Review of experimental evidence. J Environ Qual 1984;13Ž4.:580᎐586. Kloppel H, Kordel W. Pesticide volatilization and exposure of ¨ ¨ terrestrial ecosystems. Chemosphere 1997;35Ž6.:1271᎐1289. Larsson P, Olka L. Atmospheric transport of chlorinated hydrocarbons to Sweden in 1985 compared to 1973. Atmos Environ 1989;23:1699᎐1711. Martinez-Vidal JL, Egea Gonzalez FJ, Glass CR, Martinez Galera M, Castro Cano MI. Analysis of lindane, alpha- and beta-endosulfan and endosulfan sulfate in greenhouse air by gas chromatography. J Chrom A 1997;765:99᎐108. McCaffrey CA, MacLachlan J, Brookes BI. Adsorbent tube evaluation for the preconcentration of volatile organic compounds in air for analysis by gas chromatography᎐mass spectroscopy. Analyst 1994;119:897᎐902. Millet M, Wortham H, Sanusi A, Mirabel Ph. A multiresidue method for determination of trace levels pesticides in air and water. Arch Environ Contam Toxicol 1996;31:543᎐556. Nash RG. Volatilization and dissipation of acidic herbicides from soil under controlled conditions. Chemosphere 1989;18:2363᎐2373. Oehme M. Further evidence for long-range air transport of polychlorinated aromates and pesticides: North America to the Arctic. Ambio 1991;20:293᎐297. Oehme M, Mano S. The long-range transport of organic pollutants to the Arctic. Fresenius Z Anal Chem 1984; 319:141᎐146. Pacyna JM, Oehme M. Long-range transport of some organic compounds to the Norwegian artic. Atmos Environ 1988;22:243᎐257. Patton GW, Hinckley DA, Walla MD, Bidleman TF, Hargrave BT. Airborne organochlorines in the Canadian High Arctic. Tellus 1989;41B:243᎐255. Payne NJ, Thompson DG. Off target glyphosate deposits from aerial silvicultural applications under various meteorological conditions. Pestic Sci 1992;53᎐59:1992. Reisinger LM, Robinson E. Long-range transport of 2,4-D. J Appl Meteorol 1976;15:836᎐845. Riley CM, Wiesner CJ, Ecobichon DJ. Measurement of aminocarb in the long-range distance drift following aerial application to forests. Bull Environ Contam Toxicol 1989;42:891᎐898. Risebrough RW, Carmignani GM. In: Parker BC, Lawrence KS, editors. Chlorinated hydrocarbons in Antarctica. Allen Press, 1972:63᎐78. Sanusi A, Millet M, Wortham H, Mirabel Ph. A multiresidue for determination of trace levels of pesticides in atmosphere. Analusis 1997;25:302᎐308. Sanusi A, Millet M, Mirabel Ph, Wortham H. Gas-particles partitioning of pesticides in atmospheric samples. Atmos Environ 1999;33:4941᎐4951. Sherma J, Shafik TM. A multiclass, multiresidue analytical method for determining pesticide residues in air. Arch Environ Contam Toxicol 1975;3:57᎐71.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213 Seiber JN, Glotfelty DE, Lucas AD, McChesnay MM, Sagiebel JC, Wehner TA. A multiresidue method by high performance liquid chromatography-based fractionation and gas chromatographic determination of trace levels pesticides in the air and water. Arch Environ Contam Toxicol 1990; 19:583᎐592. Seiber JN, Woodrow JE. In: Ragsdale NN, Kearney PhC, Plimmer JR, editors. Origin and fate of pesticide in air. 8th International Congress of Pesticide Chemistry, Washington DC, 1995:157᎐172. Tomlin CDS, editor. The pesticide manual, 11th edition. Farnham, Surrey, UK: British Crop Protection Council, 1997. Van der Hoed N, Halmans MTH. Sampling and thermal desorption efficiency of tube type diffusive samplers: selection and performance of adsorbents. J Am Ind Hyg Assoc 1987;48:364᎐373.
213
Van der Werf HMG. Assessing the impact of pesticides in the environment. Agric Ecosys Environ 1996;60:81᎐96. Waite DT, Grover R, Westcott ND, Irvine DG, Kerr LA, Sommerstad H. Atmospheric deposition of pesticides in a small southern Saskatchewan watershed. Environ Toxicol Chem 1995;14Ž7.:1171᎐1175. Watanabe T. Determination of the concentration of pesticides in atmosphere at high altitudes after aerial application. Bull Environ Contam Toxicol 1998;60:669᎐676. Wehner TA, Woodrow JE, Kim Y-H, Seiber JE. Multiresidue analysis of trace organic pesticides in air. In: Keith LH, editor. Identification and analysis of organic compounds in air. Woburn, MA: Butterworths Publishers, 1984:273᎐290.
Comparison of different sampling techniques for the evaluation of pesticide spray drift in apple orchards Olivier Briand a , Florence Bertrand a , Rene ´ Seux a,U , Maurice Millet a,b a
Ecole Nationale de la Sante´ Publique, Laboratoire d’Etude et de Recherche en En¨ ironnement et Sante´ (LERES), A¨ enue Professeur Leon ´ Bernard, 35043 Rennes cedex, France b Laboratoire de Physico-Chimie de l’Atmosphere de la Surface, et Departement de ` (UMR 7517), Centre de Geochimie ´ ´ Chimie de l’Uni¨ ersite ´ Louis Pasteur, 1 rue Blessig, 67084 Strasbourg cedex, France Received 27 March 2001; accepted 3 July 2001
Abstract An analytical method using gas chromatography᎐mass spectrometry has been developed for the evaluation of different sampling techniques to characterise spray drift in a commercial apple orchard. Eleven pesticides were studied Žfungicides, insecticides and herbicides.. A collection of airborne spray-drift pesticides released from a low-profile air-blast orchard sprayer was investigated using six types of samplers: Ž1. a Perkin-Elmer low volume automatic air sampler using with glass tube packed with Supelpak-2; Ž2. a high volume air sampler; Ž3, 4. an impinger containing cyclohexane that could be preceded by a glass fibre filter; and Ž5, 6. glass cartridges packed with Supelpak-2 that could be preceded by a glass fibre filter. Retention efficiencies of the different sampling techniques are compared, and physical forms of the retained pesticides are discussed. These techniques have allowed us to evaluate pesticide spray-drift in the orchard. Results have shown that the molecules’ properties Ž k H and vapour pressure. and weather conditions Žtemperature and relative humidity. strongly influence pesticide gas and particles distribution. However, in the studied orchard, it is difficult to differentiate pesticide spray-drift and post-application transfers since treatment duration was ) 2 days. 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Pesticides; Atmosphere; Sampling techniques; Spray Drift; GC-MS; Orchard
U
Corresponding author. E-mail addresses: [email protected] ŽR. Seux., [email protected] ŽM. Millet..
0048-9697r02r$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 1 . 0 0 9 6 1 - 5
200
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
1. Introduction The intensive use of pesticides has led to a ubiquitous contamination of remote environments because synthetic chlorinated hydrocarbon pesticides and industrial chemicals have been identified in the Arctic ecosystem ŽHargrave et al., 1987; Patton et al., 1989., the Antarctic environment ŽGiam et al., 1974; Risebrough and Carmignani, 1972., deep-sea organisms ŽBarber and Warlen, 1979; Ballschmiter et al., 1981., and a variety of other remote areas. The atmosphere is known to be a good route for the dissemination of semi-volatile organic compounds such as pesticides and PCBs ŽEisenrich et al., 1981; Bidleman and Leonard, 1982; Billings and Bidleman, 1983; Pacyna and Oehme, 1988; Oehme, 1991.. Pesticides enter the atmosphere through many processes such as application drift during spraying operations ŽPayne and Thompson, 1992; Watanabe, 1998., wind erosion of soil and volatilisation ŽGlotfelty et al., 1989; Nash, 1989; Foster et al., 1995; Kloppel and Kordel, 1997; Cherif and ¨ ¨ Wortham, 1997.. After entry into the atmosphere, pesticides are transported and distributed over long distances, sometimes far from their emission sites ŽReisinger and Robinson, 1976; Oehme and Mano, 1984; Larsson and Olka, 1989; Riley et al., 1989; Oehme, 1991; Hoff et al., 1992a,b., deposited through wet and dry processes ŽSeiber and Woodrow, 1995; Waite et al., 1995. andror degraded ŽAtkinson et al., 1999.. Spray drift corresponds to the fraction of pesticides which never reach the crop during treatment and are locally transported by the wind before deposition. Many factors influence this phenomenon: droplet diameter, spraying technique, land topography, climatic conditions, etc. Depending on literature values ŽVan der Werf, 1996., losses due to spray drift can vary from 1 to 30% of applied quantities. To determine the atmospheric contamination of the atmosphere by pesticides, many analytical methods have been developed ŽSherma and Shafik, 1975; Wehner et al., 1984; Seiber et al., 1990; Millet et al., 1996; Sanusi et al., 1997.. These methods employ filters and solid adsor-
bents for the sampling of particles and gas respectively by using high volume samplers. After sampling, a solvent extraction followed by a concentration step was made. Solvent extraction is accurate, but is generally long and increases detection limits due to losses induced by the different steps Žextraction, clean-up, concentration, etc.. An alternative is to use temperature as an extraction procedure. With thermal-desorption, not many steps are needed, which permits lower detection limits, the elimination of parasite peaks caused by solvents and opens up the possibility of automation ŽVan der Hoed and Halmans, 1987; McCaffrey et al., 1994.. Nevertheless, this method cannot be applied for trace analysis in ambient air, but can be successfully used for exposed area such as crops during treatment periods for assessing spray drift and volatilisation of post-application processes ŽClement et al., 2000; Briand et al., ´ 2001.. In a previous work ŽClement et al., 2000., an ´ analytical method using thermal desorptionrgas chromatography᎐mass spectrometry for the evaluation of the atmospheric level of pesticides caused by spray-drift during applications and volatilisation of post-application was developed. From experiments performed in this study, it appeared that some compounds, especially captan, used in apple orchards in Brittany at present, are sensitive to high temperatures since they seem to be degraded during the thermal-desorption process. Another factor to take into account was the low vapour pressure of these pesticides, which does not facilitate thermal extraction. From this observation, the aim of this work was to develop an analytical method using solvent extraction and gas chromatography᎐mass spectrometry for an evaluation of different sampling techniques in order to characterise spray drift in an orchard. The pesticides chosen were representative of fruit farming: captan, cyproconazole, cyprodinyl, formothion, iprodione, nuarimol, phosalone, simazine, tebufenpyrad and vamidothion. Advantages, limitations and validation of this method are presented and results of the different campaigns on a Britanny apple orchard are also presented and discussed.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
2. Materials and methods 2.1. Sampling sites The field experiment took place in a commercial apple and pear orchard at Domagne ´ in Brittany, France Ž30 km south-east of Rennes. from 17 March to 29 May, 1999. The field surface area is 60 ha: 45 ha of apples, and 15 ha of pears. The orchard is located in a zone of intensive corn and rape crops. Mature apple trees are planted in 3.50-m rows with 1.00-m tree spacing. Individual trees were 2.50-m high and 1.50-m wide. One type of sprayer and different nozzle configurations were used to produce a range of applied volumes, between 250 and 1000 lrha. The Holder NI 1000 used was a standard low-profile air-blast orchard sprayer which delivers droplets larger than 1 mm in diameter. Weather conditions were recorded for each experiment Žtemperature, relative humidity, speed and direction of wind. using a complete solution by Young ŽGroupe Leader, France. The environmental conditions during spraying were very different between March 1999; the air temperature ranged from 6 to 14⬚C, with a high relative humidity Ž) 85%.; and May 1999, air temperature ranged from 22 to 28⬚C, with a lower relative humidity Ž- 75%.; wind speeds ranged from 2 to 9 m sy1 , in different directions. During a final campaign, which was performed in September 2000, the temperature ranged from 25 to 41⬚C, with a relative humidity - 60% and a wind speed lower than 1 m sy1 . 2.2. Sampling procedure Many kinds of sampling methods were employed to evaluate pesticide losses by spray-drift immediately during and after applications. The first method used a Perkin-Elmer low volume automatic air sampler ŽNorwalk, CT, USA., which was modified to permit a flow rate between 1 and 2 l miny1 Žagainst 0.05᎐0.5 l miny1 initially .. This kind of collector allowed the sequential automatic sampling of 24 tubes. Tube sampling time can be programmed, and was generally fixed to 2 h. A
201
pump SKC 224 44 ŽArelco, France., which permits controlled lower flow rate between 0.01 and 5 l miny1 , was used with this collector. A second method using an ‘Explorer high volume automatic air sampler’ ŽZambelli, Italy., was used at a flow rate of 40 l miny1 Ž70 l miny1 maximum permitted.. It allowed sequential automatic sampling of eight samples. Flow rate, volume and temperature were recorded for every sample. Two other methods using an impinger with cyclohexane RS ŽCarlo Erba, Val de Reuil, France. and low volume glass tubes, associated with a NMP 50 pump ŽNeuberger, Germany. at 1.45 l miny1 flow rates, were used in a specific way during the last three campaigns. These samplers could be preceded by glass fibre filters ŽMillipore AP40, 25 mm in diameter. mounted on IOM inhalable dust sampler ŽSKC, Dorset, USA.. All samplers were placed, depending on the campaigns, at a height between 1.00 and 1.50 m. 2.3. Preparation of adsorbent tubes Field experiments were carried out by using glass tubes Ž4 mm i.d.=89 mm. and packed with Supelpack 䊛 2 resin obtained from Perkin-Elmer Corp. ŽNorwalk, CT, USA. and Supelco ŽBellefonte, PA, USA. respectively. Each cartridge was hand-made and contained the same amount of resin Ž125 mg., kept in position with glass wool, Pesticide grade ŽSupelco.. Before use, the resin was washed for 24 h in a Soxhlet, with a mixture of n-hexanerdiethyl ether Ž90:10.. Before use, the tubes were closed with Teflon 䊛 caps and stored in the dark. For the Explorer, which can simultaneously collect the particulate and gaseous phase, using glass Ž ⌽ s 47 mm GF-Filters. fibre filters ŽMillipore, France. and 2 g Supelpack-2 resin, respectively; both were washed for 24 h in a Soxhlet with a mixture of n-hexanerdiethyl ether Ž90:10., and stored before use in the dark in clean capped glass bottles. IOM glass fibre filters were cleaned and stored in the same manner as those used for the Explorer.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
202
Table 1 Physical-chemical properties of the studied pesticides ŽIn: Tomlin, 1997 and Index Phytosanitaire 1999.. Molecules
Chemical family
Action
Molecular weight Žg moly1 .
KH ŽPa m3 moly1 .
Vapour pressure ŽmPa.
Solubility Žwater. Žmg ly1 .
Kow log P
Captan Cyproconazole Cyprodinyl Formothion Iprodione Nuarimol Phosalone Simazine Tebufenpyrad Vamidothion
N-trihalomethylthio Triazole Anilinopyrimidine Organophosphorous Dicarboximide Pyrimidinyl carbinol Organophosphorous 1,3,5-triazine Pyrazole Organophosphorous
Fungicide Fungicide Fungicide Insecticide Fungicide Fungicide Insecticide Herbicide Acaricide Insecticide
300.6 291.8 225.3 257.3 330.2 314.7 367.8 201.7 333.8 287.3
1.07= 10y3 7.2= 10y5 6.6 at 7.2= 10y3 1.1= 10y5 1.27= 10y4 3.3= 10y5 7.4= 10y3 5.6= 10y3 - 1.28= 10y3 Negligible
1.17= 10y2 3.46= 10y2 5.1= 10y1 0.113 5 = 10y4 - 0.0027 - 0.06 2.94= 10y3 - 10y2 Negligible
3.3 140 13 2600 13 26 3.05 6.2 2.6 4.1= 10 6
2.8 2.91 3.9 ? 3.1 3.18 4.01 2.5 4.61 1.18
2.4. Standard solutions of pesticides Guaranteed pesticides: captan Ž98.4%.; cyproconazole Ž99.4%.; 3,5 dichloroaniline Ž98.0%.; formothion Ž74.5%.; iprodione Ž99.9%.; nuarimol Ž96.5%.; phosalone Ž98.5%.; simazine Ž96.2%.; and vamidothion Ž96.5%. were obtained from Cluzeau Info Labo ŽBordeaux, France.. Stock solutions Ž1 g ly1 . were prepared in toluene Pestipur 䊛 ŽSDS, Peypin, France. or in acetone ŽSDS, Peypin, France. for simazine. Guaranteed cyprodinyl and tebufenpyrad standard were obtained in iso-octane at 10 ng ly1 from Cluzeau Infos Labo ŽBordeaux, France.. All dilutions were made using n-hexane Pestanal TM ŽRiedel de Haen, ¨ Seelze, Germany.. Due to their conditioning, cyprodinyl and
tebufenpyrad were added directly to the final mixture. Physical᎐chemical properties of the pesticides studied are presented in Table 1. 2.5. Analytical method A Hewlett-Packard 5890 gas chromatograph fitted with a Hewlett-Packard 5970B quadruple mass spectrometer was used. The analytical column was a dimethylpolysiloxane DB-1 Ž30 m= 0.249 mm i.d.; film thickness s 0.25 m. ŽJ & W Scientific. and the carrier gas was Helium N55 at a flow rate of 1 ml miny1 , which induced a head pressure of 9 Psi. The GC temperature program began at 60⬚C for 1 min, increased at 20⬚C miny1 to 150⬚C, increased at 2⬚C miny1 to 200⬚C and finally increased at 5⬚C miny1 to 260⬚C Žheld for
Table 2 Ions Ž mrz ., relative abundance and retention time for each pesticide studied Molecules
Ions Ž mrz .
Relative abundance Ž%.
Retention time
3,5 Dichloroaniline Simazine Formothion Captan Cyprodinyl Vamidothion Cyproconazole Nuarimol Iprodione Tebufenpyrad Phosalone
161r163 201r186r173 125r93 79r149 224r225 87r145 222r139 314r235r203 314r316r187 318r333r276 182r367r184
100r68 100r58r41 100r100 100r22 100r61 100r43 100r46 100r100r81 100r67r21 100r78r46 100r56r33
8.1 min 14.7 min 17.6 min 25.0 min 25.4 min 28.7 min 31.9 min 36.3 min 38.1 min 39.6 min 40.0 min
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
5 min.. Detection was accomplished by using electron impact ionisation, source temperature s 250⬚C, energy s 70 eV, in SIM mode Žselected ion monitoring.. Two or three ions were chosen for each pesticide and their mrz ratios are presented in Table 2. Identification was based on the relative abundance of these 2᎐3 ions for each molecule ŽTable 2.. Quantification was done on the total ion chromatogram. Calibration curves were obtained by the analysis of five mixtures of the 11 pesticides with respective concentrations ranging from 0.1 to 2.0 g mly1 . Five replicate measurements were made for 2 g mly1 solution, in order to determine relative uncertainty. The sensibility and accuracy of the method were determined and verified. Values of detection limits ŽDL. were determined by the injection of decreasing concentrations of the pesticide mixture. DL were first expressed in ngrl, and are then converted to ngrm3, after considering an average volume pumped during sampling of 500 l and a final volume after extraction of 500 l. 2.6. Resin extraction procedure Resin and glass wool, which have retained pesticides during the sampling procedure, were transferred in a 10-ml vial, and the tube was rinsed with 5 ml with a mixture of n-hexanerdiethyl ether Ž90r10., which was used for extraction. They were subsequently extracted for 15 min in a Branson 2000 ultrasonic bath ŽFisher, Elancourt, France.. Tubes extractions were undertaken during the 24 h following sampling, and stored in the dark at 4⬚C until analysis. Extracts were filtered on green filter, No 111 ŽDurieu, France., the vial was then rinsed with another 1 ml of the mixture and this phase was added to the first one. Extracts were then evaporated until droplet by a N-Evaporator thermostated at 50⬚C, readjusted to 0.5 ml with n-hexane and analysed with GCrMS. To determine recovery, five tubes were spiked with 0.5 ml of the pesticides mixture at 2 mg ly1 and extracted with the same procedure as the one previously described. A second extraction was re-
203
alised on the same vial content to confirm extraction efficiency. This extraction procedure was also applied to filters, but no recovery was determined since no pesticides were detected during the second extraction step. 2.7. Resin retention efficiency An impinger head was adapted to a 250-ml round glass flask and a mixture of pesticides was transferred in the flask. Two sorbent tubes in tandem were connected to the exit of the impinger. The flask was then heated in a GC oven Ž190⬚C for 2 h.. Simultaneously, pesticides were carried under a flow of clean air to the first sorbent tube Žefficiency tube.. In the effort to collect pesticides not retained by the efficiency tube andror those which could be released during sampling after first retention, we used a second tube Žbreakthrough tube. at the end of the first one. For each experiment, the same amount of air was delivered through tubes. After heating, the flask was rinsed in order to determine the volatilisation percentage, expressed as a ratio between the initial quantity of pesticides introduced in the flask Žnoted QP. minus those which were in the flask at end of the experiment on QP. Retention percentage was expressed as the ratio of the quantity of pesticides in the efficiency tube Žconsidering extraction recovery as determined previously. on the quantity of volatilised pesticides. Breakthrough percentage was expressed in the same way considering the ratio of the quantity of pesticides in the second tube on the quantity of volatilised pesticides.
3. Results and discussion 3.1. Validation of the analytical method A calibration curve was determined with accuracy for each compound, with determination coefficients R 2 ranging from 0.96 to 0.99 ŽTable 3.. Difficulties were found with vamidothion since LD was higher than those obtained for the other
204
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
Table 3 Characteristics of the analytical method Molecules
Concentrations range Žmg ly1 .
R2
D.L. Žmg ly1 . in vials
D.L. Žng my3 . in air
3,5 dichloroaniline Captan Cyproconazole Cyprodinyl Formothion Iprodione Nuarimol Phosalone Simazine Tebufenpyrad Vamidothion
1.0᎐0.05 2.5᎐0.3 2.0᎐0.5 2.0᎐0.1 2.0᎐0.1 2.0᎐0.25 2.0᎐0.5 2.0᎐0.25 2.0᎐0.1 2.0᎐0.1 3.0᎐0.5
0.999 0.987 0.999 0.997 0.999 0.995 0.958 0.994 0.999 0.996 0.900
0.025 0.15 0.25 0.05 0.05 0.125 0.10 0.125 0.05 0.05 0.50
50 300 500 100 100 250 200 250 100 100 1000
compounds. That can be explained by the mass spectrum of vamidothion since this molecule is broken down into a large number of ions with neighbouring relative abundance which complicate its detection in SIM mode. DL values of 25 and 250 ngrm3 Žexcept for vamidothion. were obtained, which permit monitoring the levels of pesticides in the case of spray-drift releases. The different ranges used during the study did not show any significant variation of signal responses obtained in GC-MS, except for iprodione as its signal rapidly decreases. Fresh solutions were made each week. This problem was solved by storing solutions at y18⬚C, rather than q4⬚C. 3.2. Extraction reco¨ eries Extraction efficiencies were determined after considering all the steps of the extraction procedure Žextraction and N-evaporation.. The results are presented in Table 4. The recovery percentage ranged between 11 and 103% with a relatively high R.S.D. Ži.e. formothion, 79% - 103% 132%.. Recoveries obtained by Martinez-Vidal et al. Ž1997. for organochlorine pesticides on Supelpack and sonication extraction are in the same order of magnitude. For 3,5 dichloroaniline, iprodione and vamidothion, poor recoveries can be explained by their lower stability and by the analytical difficulties previously described. Losses observed for other
pesticides are not occurred during extraction step since no pesticide was detected after a second extraction of the resin. Nevertheless, high R.S.D. are probably due to the N-evaporation step. On account of the high R.S.D. obtained for the recovery percentage, no correction has been made to field results. 3.3. Resin retention efficiency The first part of the experiment consisted in optimising the volatilisation step. Shape of the impinger, temperature and time for volatilisation were chosen to obtain a volatilisation percentage approaching 100%. We retained a round flask of Table 4 Extraction recoveries Ž%. for the 11 pesticides studied Ž n s number of experiments. Pesticides
Mean recoveries Ž%.
Lowest recoveries Ž%.
R.S.D. Ž%.
3,5 Dichloroaniline Ž n s 3. Simazine Ž n s 5. Formothion Ž n s 5. Captan Ž n s 5. Cyprodinyl Ž n s 5. Vamidothion Ž n s 4. Cyproconazole Ž n s 5. Nuarimol Ž n s 5. Iprodione Ž n s 4. Tebufenpyrad Ž n s 5. Phosalone Žn s 5.
51 98 103 66 104 11 73 102 54 100 74
43 87 79 55 90 5 63 89 44 85 66
15 10 22 16 10 38 16 11 21 13 11
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
100 ml for impinger, and 190⬚C for 80 min at a flow rate of 3 l miny1 . Through the different temperatures investigated, it appeared that responses obtained for vamidothion decreased significantly with an increase of temperature, degradation probably occurred in the flask during the heating step. Different solvents, for reference mixed standard, were also investigated Ž n-Hexane, propan-2ol and acetone.; no significant variation was observed in volatilisation percentage. A system to maintain the efficiency tube and breakthrough tube between 25 and 30⬚C was added to the initial device, using a stream of compressed air Ž7 bars, 70 l miny1 . to sweep the external surface of the tubes. Without this system, tube temperature can reach 70᎐80⬚C and a breakthrough percentage ) 10% was observed. Resin efficiency percentage ranged between 60 and 100% according to recovery efficiencies previously determined. Lower resin efficiency was obtained for pesticides with higher recovery R.S.D. 3.4. Field experiments All sampling campaigns were performed in the ‘la Claye orchard’, east of Rennes Žwest of France.. Both meteorological conditions and the farmer’s pesticide application program were taken
205
into account to determine the sampling campaigns. The kind and quantities of pesticide applied to the orchard in 1999 is presented in Table 5. 3.4.1. Sampling campaigns on 17 March and 7 April Sampling on 17 March corresponds to the first herbicide treatment. This treatment was undertaken in order to remove grass at the bottom of apple trees. For this treatment, a mixture of simazine Ž1000 g hay1 ., glyfosinate Ž800 g hay1 . and diuron Ž1500 g hay1 . was sprayed at a height of 30 cm on approximately 20% Ž12 ha. of the total surface of the orchard. The collector was placed in the wind at a 80-cm height and 7 m from the end of one line of apple trees. Sampling was made every 2 h between 11.30 and 17.30 h at a flow rate of 0.5 l miny1 . After extraction and analysis of each tube, simazine was not detected. In order to verify that failure to simazine has no relation to the detection limit of the method, samples were also analysed by using a large volume injection GC-NPD, which allows a greater sensitivity. Once again, no simazine was detected; hence, the absence of this compound is not a detection limit problem. However, in the second tube Ž13.30᎐15.30 h., iprodione was detected and quantified Ž890 ng my3 ., although this fungicide was not applied in the orchard before or during the sampling campaign ŽFig. 1.. The origin of the fungicide has been
Table 5 Program extract of pesticides treatment in the apple orchard in 1999 Commercial name
Pesticide
Dose applied Žg hay1 .
Frequency of use
Topaze effervit
Captan Penconazole Iprodione Captan Cyproconazole Tebufenpyrad Simazine Glyfosinate Diuron Cyprodinyl Vamidothion Cyproconazole
12.5 50 1000 900 12 500 1000 800 1500 56.25 200 50
4 applicationsryear
Rovral Atemi C Masaı¨ Baral
Chorus Kilval Alto
1 applicationryear 2 applicationsryear 1 applicationryear 2 applicationsryear
2 applicationsryear 1 applicationryear 2 applicationsryear
206
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
Fig. 1. Atmospheric concentrations of iprodione measured during campaigns from 17 March and 7 April 1999.
identified as an application performed in a neighbouring rape parcel situated at 3 m from the collector. The application was made at a 1 m height at 1000 g hay1 . The strong, but punctual contamination observed is the consequence of spray drift during application. A second sampling campaign was made on 7 April when captan Ž900 g hay1 . and cyproconazole Ž12 g hay1 . were applied in the orchard using a Holder NI 1000 sprayer. The collector was placed in the wind at a 80-cm height and sampling was at 1 l miny1 every 2 h between 10.50 h, 04r07 and 10.50 h, 04r08. No pesticide applied in the orchard was detected in tubes, but in the second one Ž12.50᎐15.50 h. iprodione was detected and quantified at 310 ng my3 ŽFig. 1.. The origin of the presence of iprodione in the second tube was explained after discussion with the rape farmer. He confirmed that iprodione was applied to the rape parcel during the period where sampling was done. This application corresponds to a complementary treatment. In the first sampling campaign, spray-drift contamination was identified. Lower concentrations than in the experiment of 04r07 can be explained by the localisation of the collector. It was placed 300 m from the rape parcel, so deposition of pesticide aerosols and dilution of the pollution was higher than in the first campaign. 3.4.2. Sampling campaigns on 8 and 9 April These campaigns were made during a treat-
ment program which started on 7 April Žsee previous paragraph. until 9 April. Indeed, 3 days or more are necessary for the pesticide application on the total orchard Ž60 ha.. Collectors were not moved during the totality of the sampling campaign Žheight, emplacement and flow rate. and 12 tubes were used in order to perform sampling over 24 h Ž2 h per tube.. After extraction and analysis of the tubes, no captan or cyproconazole was detected in any tube, although these pesticides were applied in the orchard. However, simazine was detected in two tubes Ž1.50᎐3.50 h and 5.50᎐7.50 h. at concentrations of 3.96 and 0.10 g my3 , respectively ŽFig. 2.. Two hypotheses concerning this result can be made: 1. Simazine could come from the orchard. Indeed, an application of simazine was made in mid-March and strong wind during the night of sampling could re-suspended soil dust containing simazine which was collected. A short rainy event occurred during the night sampling and could have laid contaminated dust onto the sampler; or 2. Simazine could come from an unidentified agri- or non-agricultural Žroads or railways. treatment performed outside the orchard. The agricultural origin of simazine seems more probable since Chevreuil et al. Ž1996. have shown rainwater contamination by simazine Žup to 0.65 g ly1 . in the Seine basin during
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
207
Fig. 2. Concentrations of simazine Žexpressed in ng.my3 . measured during campaign on 8᎐9 April 1999.
the same period ŽApril. in 1996. However, it is difficult to attribute concentrations found in the two tubes to spray-drift during application, since simazine was measured very early in the morning of 9 April. Whatever the origin of simazine, post-application volatilisation processes correlated with an increase of humidity at the beginning of the morning ŽJury et al., 1984. and followed by wind transport can be envisaged in order to explain the simazine concentration found in the second tube, but not in the first one. Indeed, this concentration is too high to be explained by volatilisation processes. 3.4.3. Sampling campaigns on 22 and 27᎐28 April On 22 April, a fungicide treatment using captan Ž900 g hay1 ., cyproconazole Ž12 g hay1 . and iprodione Ž1000 g hay1 . was undertaken. Sampling conditions were the same as for the previous campaign. The collector was placed at a 1.50m height in the wind and sampling was performed at 14.25 h for 2.5 h Žflow rate: 0.8 l miny1 .. On 27 and 28 April, an insecticide and fungicide treatment with phosalone Ž135.125 g hay1 .,
cyprodinyl Ž56.25 g hay1 . and iprodione Ž1000 g hay1 ., respectively, was also made. The collector was placed in the wind 100 m from the orchard at a 1.50-m height. For this campaign, the collector was situated in an area where it could receive a representative part of the wind which sweeps the orchard. After extraction and analysis of tubes corresponding to these two campaigns, no pesticide was found. 3.5. Discussion From all the campaigns presented here, it appears that the sampling method used seems not to be suitable for the evaluation of spray drift during pesticide treatment in the orchard. From results obtained, two hypotheses can be advanced: 1. Pesticides are present in the atmosphere at very low concentrations. The method, which was developed, is not sensitive enough to detect these pesticides. Two phenomena can explain this hypothesis. ` Pesticides are applied at low doses and
208
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
sampled volumes are too low with respect to detection limits. ` Spray-drift in the orchard is extremely low. The spraying technology, especially droplet size spectrum, which was used in the orchard can explain this. Indeed, the Holder NI 1000, which was used, delivers droplets more than 1 mm in diameter, which impact rapidly on the apple trees’ leaves or on the soil. Droplet size does not favour spray-drift, unlike droplets produced by sprayers used in large crops Žrape, maize, wheat, etc... With this kind of sprayer, droplets diameters are generally - 300 m, and so favour aerosol production which sediments slowly and can be then transported by the wind out of treatment areas. 2. The physical form, under which pesticides are emitted to non-target areas through the atmosphere Žaerosols, particles, gas phase., is not efficiently collected. The low application dose hypothesis, even if it is possible, cannot totally explain why no pesticides were detected, since compounds like captan and iprodione, which are used at higher rates, were also never detected. Furthermore, iprodione was detected in the case of an application on a large crop Žsee sampling campaign of 17 March and 7 April.. From the results obtained during the different campaigns, the second part of the first hypothesis seems to be a more suitable explanation. Indeed, pesticides detected in tubes came from spray drift from large crops treated with sprayers delivering small droplets. Concerning the last hypothesis, the sampling method is not efficient enough to collect pesticides present in the atmosphere in particulate phase. In order to check these hypotheses, another sampling campaign was undertaken in the orchard on 27 May, during two simultaneous pesticide treatments. These treatments were: 䢇
tebufenpyrad Ž500 g hay1 ., 1000 l hay1 were applied;
䢇
vamidothion Ž200 g hay1 . and cyprodinyl Ž56.25 g hay1 ., 250 l hay1 were applied.
For this campaign, three different sampling techniques were simultaneously performed. 䢇
䢇
䢇
The first consisted of an impinger containing 50 ml of cyclohexane, flow rate was fixed at 2 l miny1 and this device was used for the sampling of pesticides in the gas and aerosol phases. The second apparatus constituted a HighVolume Sampler which can separately collect the particulate Žon ⌽ s 47-mm GF-Filters. and gaseous Žwith 2 g Supelpack-2 resin. phases at a flow rate of 40 l miny1 . The third sampler, used for the collection of the gas phase, consisted of two sampling tubes containing 125 mg of Supelpak-2 resin placed in series Žthe second tube was used in order to underline an eventual loss from the first one.. The flow rate was 0.25 l miny1 .
Each sampler was placed at a 80-cm height in the wind and sampling was done over 6 h Žstarting at 11.30 h., which corresponds to the duration of the application of pesticides. Results obtained are illustrated in Fig. 3. For the Hi-Vol. sampler, the two pesticides are detected on the filter, which is assumed to be the particulate phase at concentrations of 87, 1.4 and 0.2 ng my3 for cyprodinyl, vamidothion and tebufenpyrad, respectively. In addition, 9.4 ng my3 of captan were also detected. This concentration is probably the result of a previous application in the orchard, and resuspended with soil particles. Pesticides, which are detected in the Supelpack-2 resin, and assumed to be the gas phase, are lower than on filter. Indeed, only 3.4 ng my3 were measured for cyprodinyl, while no trace of vamidothion and tebufenpyrad was found. Concerning sampling tubes, no loss was observed in the second one Ž90 l of air collected.. The first tube contains 4 ng my3 of cyprodinyl while no other pesticide was detected. Finally, for the impinger, all pesticides used during the sampling campaign were found in vari-
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
209
Fig. 3. Comparison of sampling methods Žconcentrations expressed in ng my3 ..
able concentrations in the solvent Ž21 ng my3 of cyprodinyl, 4.7 ng my3 for tebufenpyrad and traces of vamidothion.. The low concentrations underlined in this last campaign and the predominance of pesticides in the particulate phase, seem to demonstrate that spray-drift is limited in the orchard and the droplet size produced during pesticide application probably explains this observation. Large droplets induce a fast sedimentation onto leaves or soil and so minimise spray-drift in the orchard. Pesticides detected on the filters, associated with wind eroded dust, probably come from eolian erosion of treated trees or soils. This phenomenon is increased by agricultural engines’ passage between the trees. These observations explain why no pesticides were detected during the first campaigns, as the method was not sensitive enough and was not adapted to collect a particulate phase.
Low volume samplers or impingers are not suitable for the evaluation of pesticide atmospheric transfers, associated with wind eroded dust. Flow rates are too low to permit efficient particle sampling. The low concentrations observed in the Supelpak-2 tubes can be explained by the evaporation of spraying droplets containing pesticides during application or by volatilisation from treated surfaces, since high temperature Ž33⬚C. occurred during this campaign. The concentrations found are in accordance with the potential volatility of pesticides since cyprodinyl Ž H s 6.6᎐7.2.10y3 Pa m3 moly1 ., the more volatile, is detected at higher concentration than tebufenpyrad Žapprox. 10 times less volatile; H - 1.28= 10y3 . and vamidothion which has a negligible Henry’s law constant. It is important to note that concentration measured with Supelpak-2 tube is in accordance with
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
210
those obtained with the Hi-Vol. sampler Ž3.4 vs. 4.0 ng my3 .. The measurement of atmospheric concentration of cyprodinyl in the gas phase seems to be effective since the same concentrations were found using two different sampling techniques. Concentrations obtained with the impinger are in accordance with the volatility of pesticides, but differences observed with values from Hi-Vol. could be explained by partial dissolution of particulate pesticide in the solvent. This phenomenon of dissolution could overestimate concentration in the gas phase and underestimate concentration in the particulate phase since adhesion of the aerosols can occur in the bubbling tube of the impinger. 3.6. Next experiments In order to confirm previous observations, a sampling campaign was undertaken in the orchard on 14 September, 2000 Ž11.00᎐15.00 h. during phosalone treatment Ž125 grha, 250 lrha were applied.. During this campaign, four different collectors were placed at 1.50 m from ground at approximately 100 m in the wind of the orchard, which was low during sampling Ž- 1 m sy1 .. These collectors were: 1. An impinger with cyclohexane preceded by a
filter maintained at 15⬚C in a water bath Ž1.45 l miny1 .; 2. a sampling tube with Supelpak-2 preceded by a filter Ž1.45 l miny1 .; 3. a sampling tube with Supelpak-2 without filter Ž1.45 l miny1 .; and 4. Hi-Vol. sampler Zambelli equipped with a filter and resin Supelpak-2 Ž40 l miny1 ; 9.7 m3 .. The results obtained for the three first collectors are illustrated Fig. 4, and it appears that total pesticides concentrations observed Žparticulate and gaseous phases. are comparable for collectors Ž1. and Ž2.. Measurements with the Hi-Vol. sampler Ž4. are not totally available since a loss of resin occurred in the laboratory during extraction. However, the filter was analysed and the results obtained show concentrations of 19, 2 and 100 ng my3 for captan, cyprodynil and phosalone, respectively. Low concentrations obtained for phosalone vs. collectors Ž1. and Ž2. could be explained by the association of high flow rate and temperature, which could transfer pesticides from particles collected on filters to the gas-phase. The analysis of resin should demonstrate underestimation of the particulate phase. This phenomenon is well known when using high volume samplers ŽSanusi et al., 1999.. It appears that tubes are effective only for the
Fig. 4. Atmospheric concentrations of phosalone Žng my3 . in the different samplers used on 14 September, 2000.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
collection of the gas phase since concentrations detected in the tube without a filter are comparable to the concentrations obtained in the tube following a filter. This observation can be explained by the position of the entrance of the tube, which is vertically positioned, whereas the dust sampler is mounted horizontally with respect to the ground. Indeed, for this last device, particle sedimentation occurred on the filter and is favoured by flow rate, whereas in the first tube, sedimentation is against active pumping. Gas-particle partitioning, as it has been described previously, must be considered with many precautions. Indeed, with collectors Ž1. and Ž2., it appears that, even if total concentrations are comparable, strong partitioning differences between the two devices were underlined. However, the explanation of high concentration detected in cyclohexane could be due to the solubilisation of pesticides from the diffusion of solvent vapours through the filter. Concentrations measured, especially in the gas phase, are high in comparison to those obtained in previous campaigns. This difference could be explained by the relatively high temperature, which occurred during the campaign Ž41⬚C.. This campaign cannot confirm previous hypotheses in which spray-drift is negligible in the orchard. Nevertheless, relatively high concentrations obtained in the gas phase could be explained by sampling of fine aerosols ᎏ not retained on filters ᎏ coming from evaporation of droplets consecutively at high air temperature.
4. Conclusions The collection of airborne spray-drift pesticides released from a low-profile air-blast orchard sprayer and analysis using gas chromatography᎐mass spectrometry has been developed for the evaluation of different sampling techniques to characterise spray drift in a commercial apple orchard. Five types of sampler: Ž1. a Perkin-Elmer low volume automatic air sampler used with glass tube packed with Supelpak-2; Ž2. a high-volume air sampler; Ž3, 4. an impinger containing cy-
211
clohexane that could be preceded by a glass fibre filter; and Ž5, 6. glass cartridges packed with Supelpak-2 that could be preceded by a glass fibre filter were investigated for: 䢇
䢇
䢇
their capacity to be used for evaluation of spray-drift; their efficiency to collect pesticides in the atmosphere; and the influence of factors such as ambient humidity, temperature, amount of solid sorbents and sampling flow rate.
From experiments performed, spray-drift in the orchard does not support horizontal transfer of pesticides droplets, by the wind during treatment, but rather by post-application transfer of pesticides associated to wind-eroded dust. This phenomenon seems to be influenced by the spraying technique, as droplets produced in orchards are larger than those delivered by conventional sprayers Žgreat crops or mosquito control.. These droplets can rapidly sediment onto leaves or soils and act to reduce spray-drift. However, it has been observed that climatic conditions could invert this phenomenon, especially air temperature. Increase of temperature is in favour of the evaporation of droplets, and consequently, reduces their size. These smaller droplets can be more easily transported outside the orchard by the wind.
Acknowledgements This work was supported by the French Minis tr y o f E n v ir o n m e n t Ž c o n tr a c t n o . DGADrSRAEr97143.. The authors also wish to thank Mr. Olivier Ferron from SRPV-Rennes ŽFrance. for his help and Mr. Albert Navarre from la Claye orchard to have permitted us to perform air sampling on his domain. References Atkinson R, Guicherit R, Hites RA, Palm W-U, Seiber JN, de Voogt P. Transformation of pesticides in the atmosphere: a state of the art. Water Air Soil Poll 1999;115:219᎐243.
212
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213
Barber RT, Warlen SM. Organochlorine insecticide residues in sea fish from 2500 m in the Atlantic Ocean. Environ Sci Technol 1979;13:1146᎐1148. Ballschmiter K, Buchert H, Bihler S, Zell M. Baseline studies of global pollution: IV The pattern of pollution by organochlorine compounds in the North Atlantic as accumulated by fish. Fresenius Z Anal Chem 1981;306:323᎐339. Bidleman TF, Leonard R. Aerial transport of pesticides over the Northern Indian Ocean and adjacent seas. Atmos Environ 1982;16:1099᎐1107. Billings WN, Bidleman TF. High volume collection of chlorinated hydrocarbons in urban air using three solid adsorbents. Atmos Environ 1983;17:383᎐391. Briand, O, Clement, M, Seux, R, Millet, M, 2001. Field experi´ ment for assessing atmospheric transfer of pesticides during and after application. Part 1: Analysis of pesticides by Adsorptionrthermal desorption-GCrMS. Environ Sci Technol, submitted. Chevreuil M, Gargouma M, Teil MJ, Chesterikoff A. Occurrence of organochlorines ŽPCSs, pesticides. and herbicides Žtriazines, phenylureas. in the atmosphere and in the fallout from urban and rural stations of the Paris area. Sci Total Environ 1996;182Ž1᎐3.:25᎐37. Cherif S, Wortham H. A new laboratory protocol to monitor the volatilization of pesticides from soil. Inter J Environ Anal Chem 1997;68:199᎐212. Clement M, Arzel S, Le Bot B, Seux R, Millet M. Adsorp´ tionrthermal desorption-GCrMS for the analysis of pesticides in the atmosphere. Chemosphere 2000;40Ž1.:49᎐56. Eisenrich SJ, Looney BB, Thornton JD. Airborne organic contaminants in the Great Lakes Ecosystem. Environ Sci Technol 1981;15:30᎐38. Foster P, Ferrari C, Turloni S. Environmental behaviour of herbicides. Atrazine volatilisation study. Fresenius Environ Bull 1995;4:256᎐261. Giam CS, Richardson RL, Wong MK, Sackett WM. Polychlorinated biphenyls in Antartica Biota. Antartica J US 1974; 8:303᎐305. Glotfelty DE, Leech MM, Jersey J, Taylor AW. Volatilization and wind erosion of soil surface applied atrazine, simazine, alachlor, and toxaphene. J Agric Food Chem 1989; 37:546᎐551. Hargrave BT, Vas WP, Erickson PE, Fowler BR. Supply of atmospheric organochlorines to food webs in the Arctic ocean. Sixth Int. Symposium of the Comm. for Atmos. Chem. and Global Poll. on Global Atmospheric Chemistry, Peterborough, Ontario, August, 1987:23᎐29. Hoff RM, Muir DCG, Grift NP. Annual cycle of polychlorinated biphenyls and organohalogen pesticides in air in Southern Ontario. 1. Air concentration data. Environ Sci Technol 1992a;26:266᎐275. Hoff RM, Muir DCG, Grift NP. Annual cycle of polychlorinated biphenyls and organohalogen pesticides in air in Southern Ontario. 2. Atmospheric transport and sources. Environ Sci Technol 1992b;26:276᎐283. Index Phytosanitaire, 35th edition, 1999. ACTA Editions, Paris.
Jury WA, Spencer WF, Farmer WJ. Behaviour assessment model for trace organics in soil: IV. Review of experimental evidence. J Environ Qual 1984;13Ž4.:580᎐586. Kloppel H, Kordel W. Pesticide volatilization and exposure of ¨ ¨ terrestrial ecosystems. Chemosphere 1997;35Ž6.:1271᎐1289. Larsson P, Olka L. Atmospheric transport of chlorinated hydrocarbons to Sweden in 1985 compared to 1973. Atmos Environ 1989;23:1699᎐1711. Martinez-Vidal JL, Egea Gonzalez FJ, Glass CR, Martinez Galera M, Castro Cano MI. Analysis of lindane, alpha- and beta-endosulfan and endosulfan sulfate in greenhouse air by gas chromatography. J Chrom A 1997;765:99᎐108. McCaffrey CA, MacLachlan J, Brookes BI. Adsorbent tube evaluation for the preconcentration of volatile organic compounds in air for analysis by gas chromatography᎐mass spectroscopy. Analyst 1994;119:897᎐902. Millet M, Wortham H, Sanusi A, Mirabel Ph. A multiresidue method for determination of trace levels pesticides in air and water. Arch Environ Contam Toxicol 1996;31:543᎐556. Nash RG. Volatilization and dissipation of acidic herbicides from soil under controlled conditions. Chemosphere 1989;18:2363᎐2373. Oehme M. Further evidence for long-range air transport of polychlorinated aromates and pesticides: North America to the Arctic. Ambio 1991;20:293᎐297. Oehme M, Mano S. The long-range transport of organic pollutants to the Arctic. Fresenius Z Anal Chem 1984; 319:141᎐146. Pacyna JM, Oehme M. Long-range transport of some organic compounds to the Norwegian artic. Atmos Environ 1988;22:243᎐257. Patton GW, Hinckley DA, Walla MD, Bidleman TF, Hargrave BT. Airborne organochlorines in the Canadian High Arctic. Tellus 1989;41B:243᎐255. Payne NJ, Thompson DG. Off target glyphosate deposits from aerial silvicultural applications under various meteorological conditions. Pestic Sci 1992;53᎐59:1992. Reisinger LM, Robinson E. Long-range transport of 2,4-D. J Appl Meteorol 1976;15:836᎐845. Riley CM, Wiesner CJ, Ecobichon DJ. Measurement of aminocarb in the long-range distance drift following aerial application to forests. Bull Environ Contam Toxicol 1989;42:891᎐898. Risebrough RW, Carmignani GM. In: Parker BC, Lawrence KS, editors. Chlorinated hydrocarbons in Antarctica. Allen Press, 1972:63᎐78. Sanusi A, Millet M, Wortham H, Mirabel Ph. A multiresidue for determination of trace levels of pesticides in atmosphere. Analusis 1997;25:302᎐308. Sanusi A, Millet M, Mirabel Ph, Wortham H. Gas-particles partitioning of pesticides in atmospheric samples. Atmos Environ 1999;33:4941᎐4951. Sherma J, Shafik TM. A multiclass, multiresidue analytical method for determining pesticide residues in air. Arch Environ Contam Toxicol 1975;3:57᎐71.
O. Briand et al. r The Science of the Total En¨ ironment 288 (2002) 199᎐213 Seiber JN, Glotfelty DE, Lucas AD, McChesnay MM, Sagiebel JC, Wehner TA. A multiresidue method by high performance liquid chromatography-based fractionation and gas chromatographic determination of trace levels pesticides in the air and water. Arch Environ Contam Toxicol 1990; 19:583᎐592. Seiber JN, Woodrow JE. In: Ragsdale NN, Kearney PhC, Plimmer JR, editors. Origin and fate of pesticide in air. 8th International Congress of Pesticide Chemistry, Washington DC, 1995:157᎐172. Tomlin CDS, editor. The pesticide manual, 11th edition. Farnham, Surrey, UK: British Crop Protection Council, 1997. Van der Hoed N, Halmans MTH. Sampling and thermal desorption efficiency of tube type diffusive samplers: selection and performance of adsorbents. J Am Ind Hyg Assoc 1987;48:364᎐373.
213
Van der Werf HMG. Assessing the impact of pesticides in the environment. Agric Ecosys Environ 1996;60:81᎐96. Waite DT, Grover R, Westcott ND, Irvine DG, Kerr LA, Sommerstad H. Atmospheric deposition of pesticides in a small southern Saskatchewan watershed. Environ Toxicol Chem 1995;14Ž7.:1171᎐1175. Watanabe T. Determination of the concentration of pesticides in atmosphere at high altitudes after aerial application. Bull Environ Contam Toxicol 1998;60:669᎐676. Wehner TA, Woodrow JE, Kim Y-H, Seiber JE. Multiresidue analysis of trace organic pesticides in air. In: Keith LH, editor. Identification and analysis of organic compounds in air. Woburn, MA: Butterworths Publishers, 1984:273᎐290.