853

Environmental Toxicology and Chemistry, Vol. 28, No. 4, pp. 853–863, 2009� 2009 SETAC

Printed in the USA0730-7268/09 $12.00 � .00

ESTIMATING ECOTOXICOLOGICAL EFFECTS OF PESTICIDE DRIFT ONNONTARGET ARTHROPODS IN FIELD HEDGEROWS

STEFAN OTTO,*† LUCA LAZZARO,‡ ANTONIO FINIZIO,§ and GIUSEPPE ZANIN‡†Institute of Agro-environmental and Forest Biology—CNR, ‡Department of Environmental Agronomy and Crop Sciences,

University of Padova, Agripolis-Viale dell’Universita 16, 35020 Legnaro, Italy§Department of Environmental Sciences, University of Milano Bicocca, Piazza della Scienza 1, 20126 Milano, Italy

(Received 5 June 2008; Accepted 17 October 2008)

Abstract—When hedgerows grow in orchards where pesticides are applied, they can play a double role: Providing a barrier forchemical spray drift and as a refuge for beneficial arthropods such as pollinators and predators. Effectiveness of hedgerows asbarriers to drift depends mainly on canopy density (that can be estimated through optical porosity) and wind speed. When opticalporosity is low, the hedgerow can intercept a significant amount of spray drift and act as an effective barrier, but the interceptedpesticide can negatively affect the beneficial arthropods living there. A drift model was used to simulate drift in a hedgerow–vineyard system, and a deposition distribution model was used to calculate the pesticide spatial pattern distribution on a hedgerowwith different optical porosity and wind speed conditions. The possible ecotoxicological effects were estimated for 28 activeingredients with different median lethal rates for two nontarget arthropods, Aphidius rhopalosiphi and Typhlodromus pyri. Aspatialized risk assessment for a hedgerow is suggested to improve procedures based on application rate, standard drift, and vegetationdistribution values, as in the hazard quotient approach. An alternative method for calculation of the exposure is also proposed, witha step-by-step example of a toxicity/exposure ratio calculation. The results highlighted the importance of the spatial pattern of driftand proved that a hedgerow can be an effective barrier against spray drift. Analysis of the toxicity/exposure ratio values showedthat a hedgerow can continue its shelter and feeding function for nontarget arthropods when low-toxicity pesticides are used, thereis no significant wind interference, or both.

Keywords—Hedgerow Pesticide drift Risk assessment Aphidius rhopalosiphi Typhlodromus pyri

INTRODUCTION

Field hedgerows are landscape infrastructure that usuallyconsists of different layers of herbaceous, shrub, and tree spe-cies, sometimes with a lateral grass strip. As such, they con-stitute an important source of biodiversity [1]. A hedgerow canenrich the local arthropod population for a distance of approx-imately three to ten times its height to the leeward side andequal to or twice its height to the windward [2], and this is ofparticular importance in monocultures. Moreover, during fieldcrop and orchard spraying, hedgerows can intercept spray driftand protect the leeward areas from contamination ([3,4];http://www.agdrift.com/PDF�FILES/drift%20filtration.PDF).Insecticide spray drifting into natural and seminatural habitatsadjacent to treated fields endangers not only biodiversity butalso biological pest control, because these habitats offer alter-native prey or hosts for generalist predators, provide food sourc-es such as pollen and nectar, and are regarded as importantrefuges and overwintering sites for epigeal predators [5].

Hedgerow capacity in reducing spray drift may be veryhigh: In comparison to no hedgerow situations, according tosome authors [6,7], reduction percentages range from 50 to90%. Nevertheless, interception of drift negatively affects thehedgerow itself, because beneficial arthropods living in it maybe exposed to pesticides [8,9].

Directive 91/414 of the European Commission for placingnew plant protection products on the market [10] includes theneed for specific risk assessment for nontarget arthropods. This

* To whom correspondence may be addressed(stefan.otto@ibaf.cnr.it).

Published on the Web 12/19/2008.

is calculated using the hazard quotient (HQ) approach, fol-lowing the guidance document on terrestrial ecotoxicology[11] and the results of the ESCORT 2 workshop ([12];http://www.iobc-wprs.org/). According to these documents, atTier I, the HQ is calculated by dividing the crop-specific ap-plication rates (in-field exposure scenario) or drift rates (off-field exposure scenario) by the median lethal rate (LR50); thelast is the application rate causing 50% mortality of the twotest organisms Typhlodromus pyri Scheuten (Acari, Phyto-seiidae) and Aphidius rhopalosiphi DeStefani-Perez (Hyme-noptera, Braconidae) chosen as the most sensitive species [13–15]. The off-field HQ is calculated with the model

drift factorAR � MAF � � �vegetation distribution factor

off-field HQ �LR50

� correction factor (1)

where AR and MAF are the application rate and multiple ap-plication factor [16], respectively; the drift factor (% drift/100)derives from BBA studies [17]. The obtained exposure (i.e.,predicted environmental concentration [PEC]) is then cor-rected by a vegetation distribution factor (default value � 10)to account for foliar surface area effects on exposure. The HQalso takes into consideration a correction factor of 10 (safetyfactor) to account for uncertainty with the extrapolation fromT. pyri and A. rhopalosiphi to other nontarget arthropods.Further details are reported in ([12,18]; http://www.iobc-wprs.org/pub/book2000.pdf). At Tier 1, an HQ triggervalue of 2 for both species is considered. Such an approachhas the advantage of being simple, but it is somewhat limiting

854 Environ. Toxicol. Chem. 28, 2009 S. Otto et al.

Fig. 1. Layout of sprayer, poles, collectors on poles, and hedgerowsin the simplified scenario (hedgerow parallel to crop rows and pes-ticide sprayed directly on the hedgerow).

because it does not take into account the spatial distributionof exposure, even estimated. Better models may be thoseadapted to specific agronomic practices and environmentalconditions for different countries, so research is needed in thearea of off-field drift estimation and measurement. Further-more, recent research proved the relationship between leafdeposits and mortalities of test organisms, and the connectionbetween drift deposit measurement and exposure bioassay pro-vides a promising approach to determining effects of insec-ticide drift on nontarget arthropods [19].

A complex study on drift generated by an air-assisted spray-er was conducted in 2004–2005 [20] to evaluate hedgerowefficacy in reducing spray drift and develop a model to predictspray drift level in the areas surrounding treated fields in atypical northeast Italian vineyard–hedgerow system.

In the present study, the previous work was used to calculateexposure in a field hedgerow after a pesticide application.Another model was constructed to calculate the spatial patterndistribution of drift in the hedgerow canopy, for its wholewidth, to highlight the importance of spatial variability of thePECs, not accounted for in the HQ approach. Given that suchvariability can be of importance in some cases, a direct com-parison of the spatialized PECs with the LR50s of 28 pesticidesapproved for use in Italian vineyards was then performed, andsome exemplar situations are discussed. This can improve theunderstanding of the effects of drift on nontarget arthropodsand provide a rational basis for an off-field exposure assess-ment that can integrate that proposed by the ESCORT 2 guid-ance document [12].

MATERIALS AND METHODS

Modeling the spatial pattern of drift

With the aim of evaluating the effect of a hedgerow on thespatial pattern of spray drift, a new procedure was proposedby the Department of Environmental Agronomy and Crop Sci-ences of Padova University [20]. The procedure is based onthe application of a model for the estimate of spray drift spatialpattern from parameters such as wind speed, distance down-wind from the sprayer, and optical porosity of an antidrifthedgerow barrier [21,22], because optical porosity has beenproven to be an alternative to aerodynamic porosity [20].

The first step of the procedure calculates the quantity ofactive ingredient sprayed per meter run by the sprayer (Q,g/m). The second step is the application of a negative expo-nential drift model to estimate the spray drift still in the airat distance x from the spray origin [20]:

xy � Q exp � (2)[ ]3 OP exp(WS)

where y is the drift (g/m) at distance x, Q is the quantity ofsprayed active ingredient (g/m), 3 is a model parameter(shape), OP is the optical porosity (fraction), and WS is thewind speed (m/s). The shape parameter has been introducedin the model to fit the experimental drift data [20].

This model can be applied every time that spray drift im-pacts on a porous barrier, for example, one of the crop rowsbetween sprayer and hedgerow. In the present study, the termimpact refers only to the physical interaction of droplets (thedrift) with a porous barrier. Contiguous barriers have to beconsidered as one, whereas for separated ones a sequentialcalculation is required for each, using the model output afterone barrier as input for the decay calculation for the subsequentbarrier. In the absence of vertical obstacles, drift decreases

undisturbed downwind by ground deposition. By applicationof Equation 2 to an obstacle with a given optical porosity, theestimation of the decay curve of drift beyond it is possible.

Measuring the deposition distribution in the hedgerow

According to what is commonly observed in field condi-tions, the drift that impacts the hedgerow is assumed to remainthere, so drift distribution in the hedgerow canopy can bestudied for its whole width. Leaf area is taken as uniformlydistributed along the vertical profile of the hedgerow and alsothe leaf area index of hedgerow (i.e., the one-sided green leafarea per unit ground area, LAI, dimensionless), whereas thedrift deposition amount changes: More drift would depositcloser to the sprayer, and less would deposit on the oppositeside of the hedgerow. Some research has shown the patchinessof drift deposition [19]; however, in this scenario, the driftdeposition at a defined distance from the sprayer is assumedto be uniform.

A vertical profile distribution of drift was obtained fromdata derived from experiments conducted in northern Italy onthree dates (September 2004, December 2004, and October2005) to evaluate the capacity of hedgerows to reduce driftcaused by broadcast air-assisted sprayers. On each date, threehedgerow scenarios were considered (no hedgerow, singlehedgerow, and double hedgerows) combined with two sprayer–hedge interactions: sprayer starting at 0.5 m from the hedgeand moving perpendicularly to it (best-normal case) or sprayertraveling along the hedge at a distance of 0.5 m and sprayingit directly (worst case). In reference to the double hedgerow,the September experiment was done at very low optical po-rosity values (10.8%, i.e., dense hedge canopy), that of De-cember with very high values (74.2%, i.e., sparse hedge can-opy), and that of October at intermediate (44.9%).

The sprayer tank was filled with a 0.25% E102 (tartrazyneyellow) solution (Dynemic, Ahmedabad, Gujarat, India). Col-lectors for the interception of aerial drift were 10 � 25 cmair-conditioner synthetic buffer rectangles (type CM 360 man-ufactured by Camfil, Cinisello Balsamo, Italy) [23], whichproved again to be efficient in terms of both interception andrecovery rate [20]. The collectors were positioned 1 m aparton 5-m-tall poles, from 0 to 5 m. Four poles were set out ina straight line perpendicular to the hedgerow, the first one infront of the hedge and the others at 4, 8, and 12 m from thefirst. This provided a total of 6 � 4 � 24 sampling positions.Where two hedgerows were included in the study, they werebetween the first and second and second and third poles. Forthe interception of ground deposit, collectors were placed hor-izontally at soil level (Fig. 1). Other details (such as type ofsprayer, runs, tracer, and analysis) are as in Lazzaro et al. [20].

Ecotoxicological effects of pesticide drift in hedgerows Environ. Toxicol. Chem. 28, 2009 855

Table 1. Active ingredients used in the simulationsa

Active ingredient TypeApplicationrate (g/ha)

Aphidius rhopalosiphi

LR50 (g/ha) Trial length

Typhlodromus pyri

LR50 (g/ha) Trial length

Alpha-cypermethrin Insecticide 35 0.256 48 h 0.02b 7 dCarbaryl Insecticide 430 0.0247 48 h 457b 7 dChlormequat chloride Growth regulator 230.5 2,200 2,250Cyanamide Growth regulator 7,800 432.1 14 d 445.6 14 dCyfluthrin Insecticide 10 1.63 48 h 0.42b 7 dCymoxanil Fungicide 110 480 480Cyproconazole Fungicide 10 80 48 h 35.6 7 dCyprodinil Fungicide 150 14,700 48 h 2,250 7 dDimethoate Insecticide 141.75 2.24 7 dDimetomorph Fungicide 125 1,800 48 h 1,800 7 dDithianon Fungicide 330 4,200 48 h 4,200 7 dDodine Fungicide 325 1,800 48 h 4,000 7 dEtofenprox Insecticide 105 0.42 48 h 0.7 7 dFenbutatin oxide Acaricide 275 85.9 48 h 468.6 7 dFormetanate hydroxide Insecticide/acaricide 625 0.93 48 h 0.78 7 dFosetyl-Al Fungicide 650 80,000 48 hHexythiazox Acaricide 25 300 48 h 65.9 7 dLambda-cyhalothrin Insecticide 18.75 0.2 7 dLufenuron Insecticide 25 100 7 d 21c 7 dMalathion Insecticide 500 0.061 48 h 85.4Methiocarb Insecticide/acaricide 500 0.47 48 h 33.7 7 dMethomyl Insecticide 200 0.25 48 h 12.8 7 dPirimicarb Insecticide 175 620 48 h 835 7 dPropargite Acaricide 142.5 7.2 4.5Pyridaben Acaricide 60 47 47Tau-fluvalinate Insecticide 60 0.049 48 h 0.48 7 dTebuconazole Fungicide 33.75 62.5 58.5Zeta-cypermethrin Insecticide 12 5.03 48 h 23.18 7 d

a Application rate from [25], median lethal rate (LR50) from the European Union Footprint project database (http://sitem.herts.ac.uk/aeru/footprint/).b Protonymph.c Coccinella septempunctata L.

In the present study, the data for drift interception obtainedwith the sprayer traveling along the hedge and spraying itdirectly, on the three dates and for the double hedgerow, wereused. Such hedgerows, up to hundreds of meters in length andup to some meters in width, are typical of the Po Valley land-scape when a field is surrounded by a ditch or narrow channel,and its importance increases with the intensification of crop-ping systems. For this scenario, all depositions on all samplersfor every run were considered. Given that the absolute amountof deposit varies across dates by a factor of 2, as these actedas a random factor, within each date, the deposit values oneach pole were rescaled on their 0 to 5 mean, allowing theproportion of active ingredient deposited at each height to beestimated independently of the initial sprayed quantity. As aresult, 18 dimensionless values were obtained and then inter-polated directly.

Modeling of the deposition distribution in the hedgerow

The 18 averaged and rescaled experimental values obtainedwere arranged in a skewed bell shape, then interpolated withan exponential equation

ah � d h � d 1z � exp � (3)� � � �[ ] [ ]b b bc

where z is the deposition amount, dimensionless, expressed asa fraction of the mean value of deposition, h is the height fromground, and a, b, c, and d are equation parameters.

The fitting was performed using the nonlinear procedure ofStatistica 7.1 [24]. The exponential equation was chosen be-cause it is very flexible and can fit even very skewed bell-

shaped values. Other models could have been used, becausethe input parameters do not require a physical meaning butjust have to satisfactorily fit the experimental values.

Risk assessment for hedgerow

After application of the drift model (Eqn. 2), the amountof pesticide drift (g/ha) retained by the hedgerow foliage iscalculated (i.e., the average PEC for the entire hedgerow,which can also be expressed in terms of mass of depositionon the canopy surface [�g/cm2]). After application of the de-position distribution model (Eqn. 3), the vertical pattern dis-tribution of this average PEC is calculated. The iterative ap-plication of Equations 2 and 3 for a 1-m-wide vertical sectionof hedgerow finally allows calculation of the PEC pattern inthe entire hedgerow width. This approach to the calculationof the PEC differs from that of the ESCORT 2 report [12],because it is based on experimental values of the spatial patternof drift and allows an assessment of more realistic conditions.The PEC was calculated for 28 active ingredients used in vine-yards in Italy and normally applied with an air-assisted sprayerand considering the application rate suggested on the productlabel [25,26] (Table 1).

In accordance with the guidance document on terrestrialecotoxicology [11] and the results of the ESCORT 2 workshop[12], T. pyri and A. rhopalosiphi were chosen as representativeof nontarget arthropods species for the present study. The tox-icological endpoint considered was the LR50 (g/ha) [15], ob-tained from the European Union Footprint project database(http://sitem.herts.ac.uk/aeru/footprint/) (Table 1).

In accordance with a well-accepted simple approach

856 Environ. Toxicol. Chem. 28, 2009 S. Otto et al.

Fig. 2. Deposit fraction, rescaled to the overall mean, as a functionof height (reversed axis). Mean of experimental data for each fieldtrial date (20 September 2004, □; 13 December 2004, �; 13 October2005, �) and fitting model (Eqn. 3, solid curve). Overall mean ofthe dates (#) and standard error bars are indicated. Standard errorsfor each of the 18 means vary from 0.0042 to 0.0335 and are lowerthan the marker. Model parameters (standard error): a � 4.04 (3.22);b � 0.62 (0.25); c � 4.19 (2.05); d � 0.62 (0.25); general fitting, r� 0.96, r2 � 0.92. Original values are in mg/m. The vertical solidline is the overall mean deposition.

[10,27], in the present study the risk to nontarget arthropodsin the hedgerow was based on toxicity to exposure ratio (TER)values:

�1lethal rate 50 (g ha )TER � (4)

�1PEC (g ha )

When TER � 1, the risk is low; when TER � 1, the risk ishigh. Aware of the probabilistic meaning of this ratio, in thepresent study we considered that, when an arthropod is inconditions where TER � 1, then it is safe and can survive.

Case study scenarios

Equation 1 can be applied iteratively each time that thespray drift impacts a crop row or hedgerow. To estimate theeffect of spraying on the crop row farthest from the hedgerow,initial spray value (i.e., initial exposure) can be considered asthe quantity of pesticide emitted by the sprayer per meter ofrun, and the beginning of the vegetation structure of the nextrow can be considered as the distance from the origin of thedrift. The model output can then be considered as the inputfor the next row and so on for all crop rows between sprayerand hedgerow. This processing may be repeated until the driftinto a hedgerow from pesticide applications to a whole adja-cent field is considered. This may be achieved by adding to-gether all of the drift inputs, taking into account attenuationof drift by all crop rows. Equation 2 can be used for themodeling of virtually all types of crop–hedgerow systems, butto keep the calculation simple, two particular case study sce-narios were used in the present study.

In a simplified scenario, a simple sprayer–hedgerow systemwas considered with the hedgerow parallel to crop rows andpesticide sprayed directly on the hedgerow as if it was a croprow. Two hedgerow optical porosity values (0.7 and 0.3) andtwo wind speeds (0.5 and 2.0 m/s) were used in the simulation,and Equation 2 was applied once for the estimation of driftintercepted by the hedgerow. For each active ingredient andeach of the four factor combination scenarios (two opticalporosities for two wind speeds), an average PEC (g/ha) forthe whole hedgerow was calculated with Equation 2, and itsvertical pattern distribution was calculated with Equation 3.A 6-m-wide by 5-m-high hedgerow was hypothesized, whichwas divided into six vertical sections (slices) of 1 m in widthby 5 m in height: The 0 to 1 m section is at the side nearestto the sprayer, the subsequent 2 to 3 m, 3 to 4 m, and 4 to 5m are at progressive distances (i.e., deep in the hedgerow),and the last 5 to 6 m is that at the far side, opposite the sprayer.In each section, with Equations 2 and 3 iteratively, the two-dimensional (2-d) pattern distributions of the PEC are calcu-lated (i.e., how drift varies vertically [0–5 m high] and hori-zontally within each vertical section of 1 m in width of hedge-row). The average 2-d pattern distribution is also calculatedfor the entire hedgerow width (0–6 m).

In a second more realistic field condition scenario, whengood agricultural practices are followed, the hedgerow is notdirectly treated but intercepts drift generated from the appli-cations to the nearest crop rows. According to the results ofEquation 2 and the standard drift values [17], the drift invineyards is negligible 10 to 15 m from the sprayer, so astandard scenario can be considered, given by one treated1-m-wide row and a hedgerow 3 m distant and interceptingthe generated drift. In this scenario, Equation 2 was used it-eratively to calculate, when spraying the crop row, the driftintercepted by the crop row and then retained by the hedgerow.

This calculation provides a correction factor to be applied tothe application rate for the calculation of the PEC in the hedge-row. The calculation then proceeds as in the simplified scenariowith the determination of vertical and horizontal pattern dis-tribution of the PEC in the hedgerow. The main differencebetween the two scenarios is therefore the amount of pesticideintercepted by the hedgerow, which is a simple fraction of theapplication rate in the case of the simplified scenario, or afraction adjusted by a scale factor in the case of the standardscenario.

RESULTS AND DISCUSSION

Modeling the deposition distribution in the hedgerow

The fitting of the rescaled values with the vertical distri-bution deposition model (Eqn. 3) showed a clear relationship(r2 � 0.92): At a height of between 1.0 and 2.0 m above theground, the deposition is nearly double that of the mean, andat a height of 5 m deposition is approximately one-fifth thatof the mean (Fig. 2). Given that equipment and operatingparameters used in the experiments were standard for northernItaly [28], the bell-shaped pattern of deposition obtained canbe considered representative of the field conditions in thatregion. Obviously, the deposition pattern may change signif-icantly for any combination of sprayer type, tractor speed,nozzles, or sprayer fan capacity.

Hedgerow interception depends on optical porosity con-ditions and wind speed: Although the hedge effect is significanteven with a very sparse canopy (i.e., OP � 0.7), with a densecanopy and low wind speed, the fraction of air drift is 2 ordersof magnitude lower on the leeward side than that on the wind-ward side (i.e., with OP � 0.3 and WS � 0.5 m/s in thesimplified scenario, 81.7% of the sprayed amount impacts, butonly 1.8% can bypass the hedgerow) (Table 2).

Ecotoxicological effects of pesticide drift in hedgerows Environ. Toxicol. Chem. 28, 2009 857

Table 2. Estimated fraction of application rate impacting and bypassing a 6-m-wide hedgerow for each wind speed and optical porosity consideredin the simplified scenario (direct treatment on hedgerow) and scale factor to be introduced for the standard scenario (hedgerow intercepting the

drift generated by the treatment of one adjacent crop row)a

Optical porosity (dimensionless) 1.0 1.0 0.3 0.3 0.7 0.7Wind speed (m/s) 0.5 2.0 0.5 2.0 0.5 2.0Distance from source of drift (m) 1.0 1.0 6.0 6.0 6.0 6.0LAI of the hedgerow (dimensionless) 6.0 6.0 3.0 3.0LAI of the herbaceous vegetation beneath the hedgerow (dimensionless) 1.0 1.0 1.0 1.0Fraction impacting the hedge in the simplified scenario 0.817 0.956Fraction bypassing the hedge in the simplified scenario 0.018 0.406 0.177 0.679Scale factor for the standard scenario (1 row) 0.063 0.540 0.136 0.641

a Leaf area index (LAI) of hedgerow is based on Davis and William [29] and authors’ expertise; optical porosity � 1.0 means no hedgerow orcrop row.

Calculating the TER and graphical representation of thePEC

For the simplified scenario (pesticide sprayed directly onthe hedgerow), the PEC and TER values were calculated forthe 28 active ingredients for the two nontarget arthropods inthe hedgerow in the four optical porosity and wind speed com-binations (Table 3).

As an example of PEC and then TER calculation, the caseof the active ingredient propargite for T. pyri with OP � 0.7and WS � 0.5, which has TER � 1.131, may be considered.Hypothesizing a treatment with a volume of 500 L/ha on ahedgerow 10,000 m in length, each linear meter of hedgerowreceives 500/10,000 � 0.05 L of the theoretical applied so-lution. It should be noted that the spraying volume is inde-pendent of the length of the hypothetical hedgerow, becauseeach meter receives exactly the same amount.

According to Equation 2, a fraction of 0.817 impacts on ahedgerow 1 m distant from the sprayer (Table 2), correspond-ing to 0.05 � 0.817 � 0.041 L. Given a concentration of 0.5g/L of a commercial formulate containing 0.57 L/L of pro-pargite, each linear meter of hedgerow receives (0.041 � 0.5� 0.57) � 0.0116 g of active ingredient. When OP � 0.7 andWS � 0.5, Equation 2 calculates that the first 1-m section ofhedgerow (section 0–1 m) is bypassed by a fraction of 0.749,so a fraction of (1 � 0.749) � 0.251 is retained in this section,corresponding to (0.251 � 0.0116) � 0.0029 g of active in-gredient. A hedgerow with OP � 0.7 has an estimated LAI �3 [29], so by hypothesis of a uniform repartition of that amountof active ingredient in the canopy, the average deposition inthe first 1-m section of hedgerow and on the soil beneath is[0.0029/(3 � 1 � 1)] � 1,000,000/10,000 � 0.073 �g/cm2.The vertical bell-shaped repartition of this average amount isthen calculated with Equation 3. The same calculation appliesiteratively for each 1-m section (e.g., the 1-to-2-m section hasan average deposition of 0.055 �g/cm2, the 2-to-3-m section0.041 �g/cm2, the 3-to-4-m section 0.031 �g/cm2, the 4-to-5-m section 0.023 �g/cm2, and the 5-to-6-m section 0.017 �g/cm2). The whole hedgerow therefore has an average depositionof (0.073 � 0.055 � 0.041 � 0.031 � 0.023 � 0.017)/6 �0.0398 �g/cm2. The calculation of both the average and thehighest (worst-case) deposition allows a risk assessment evenwhen it is not known exactly where arthropods reside.

Besides terms of the mass of the active ingredient on thecanopy surface, the same calculations can be done in terms ofapplication rate, such as a sprayed solution concentration (i.e.,ml/L) or mass applied on a defined surface (i.e., g/ha). Forexample, the application rate of propargite is 142.5 g/ha (Table1). When this rate is applied directly to the hedgerow with anair-assisted sprayer, a fraction of 0.817 impacts on the hedge-

row, and a fraction of 0.177 can pass beyond it (Table 2). Sothe fraction of 0.817 � (1 � 0.177) � 0.673 remains in thehedgerow, which corresponds to (142.5 � 0.673) � 95.84g/ha of propargite, that is, the average PEC in the whole hedge-row and the integral of the depositions previously expressedin �g/cm2. Given that the hedgerow has LAI � 3 and soil hasLAI � 1 (Table 2) and that the whole 6 m width must beconsidered, this same integral can be obtained using the av-erage deposition: (0.0398 �g/cm2) � [(3 � 6 � 1 � 6) �100,000,000 cm2/ha] � [(1/1,000,000) g/�g] � 95.84 g/ha.

Deposition of active ingredient can then be expressed bothin mass of active ingredient on canopy surface and in massof active ingredient on soil (field) surface simply with a changeof the scale (Fig. 3). The LR50 for T. pyri is 4.5 g/ha (Table1) or 0.045 �g/cm2, so TER � 0.045/0.0398 � 1.131 (Table3). This means that, on average, the concentration in the hedge-row is slightly lower than the toxicological endpoint and thatT. pyri is safe (see vertical blue solid line in Fig. 3). The samecan be concluded for A. rhopalosiphi, which has an LR50 of0.072 �g/cm2 and TER � 2.356 (see vertical red solid line).But this does not mean that all individuals are safe. The de-position of 0.0398 �g/cm2 (or 95.84 g/ha) remaining in thehedgerow (see black solid line) is the average concentrationin the entire hedgerow, but results from experimental trials[20] show that the distribution of drift in a hedgerow varieswith height and horizontal distance from the sprayer (i.e., thehedgerow sections closer to the sprayer intercept more drift).The consequence is that even if the average PEC is higherthan the LR50, refuges can be found in some parts of thehedgerow (i.e., in the higher or deeper parts). Likewise, whenthe average PEC is lower than the LR50, it is possible thatthe pesticide concentration in some parts of the hedgerow canreach unacceptable values for arthropod survival. In a 6-m-wide hedgerow, this 2-d pattern distribution of the PEC canbe calculated using Equation 3 for each 1-m-wide section.There are therefore six 1-m-wide sections, from 0 to 1 m(closer to the sprayer) to 5 to 6 m (at the opposite side), plottedfrom right (dotted blue line) to left (dotted gray line). Thevertical black solid line is the average PEC on the canopy(0.0398 �g/cm2), the solid black curve is its vertical patterndistribution, the vertical blue solid line is the LR50 for T. pyri(0.045 �g/cm2), the vertical red solid line is the LR50 for A.rhopalosiphi (0.072 �g/cm2), and the vertical green solid lineis the theoretical field deposition rate (0.059 �g/cm2). Becausethe theoretical field application rate is 142.50 g/ha, when thewhole amount remains in the hedgerow this rate correspondsto 142.50/[(6 � 3 � 6 � 1) � 100] � 0.059 �g/cm2 of the-oretical deposition in the hedgerow canopy.

As described above, the solid black curve crossing the ver-

858 Environ. Toxicol. Chem. 28, 2009 S. Otto et al.

Tab

le3.

Ave

rage

pred

icte

den

viro

nmen

tal

conc

entr

atio

n(P

EC

)an

dto

xici

tyto

expo

sure

rati

o(T

ER

)va

lues

(rou

nded

),ca

lcul

ated

wit

hE

quat

ion

4,fo

rA

phid

ius

rhop

alos

iphi

and

Typ

hlod

rom

uspy

riin

the

hedg

erow

for

28ac

tive

ingr

edie

nts

inth

efo

urco

mbi

nati

ons

ofop

tica

lpo

rosi

ty(O

P,di

men

sion

less

)an

dw

ind

spee

d(W

S,

m/s

)in

the

sim

plifi

edsc

enar

io(h

edge

row

para

llel

tocr

opro

ws

and

pest

icid

esp

raye

ddi

rect

lyon

the

hedg

erow

)

Act

ive

ingr

edie

nt

PE

C(�

g/cm

2 )

OP

0.7

WS

0.5

OP

0.7

WS

2.0

OP

0.3

WS

0.5

OP

0.3

WS

2.0

TE

R

A.

rhop

alos

iphi

OP

0.7

WS

0.5

OP

0.7

WS

2.0

OP

0.3

WS

0.5

OP

0.3

WS

2.0

T.

pyri

OP

0.7

WS

0.5

OP

0.7

WS

2.0

OP

0.3

WS

0.5

OP

0.3

WS

2.0

Alp

ha-c

yper

met

hrin

0.01

00.

004

0.00

70.

005

0.26

20.

575

0.38

40.

543

0.02

00.

045

0.03

00.

042

Car

bary

l0.

120

0.05

50.

082

0.05

80.

002

0.00

50.

003

0.00

438

.07

83.5

355

.83

78.9

Chl

orm

equa

tch

lori

de0.

064

0.02

90.

044

0.03

134

1.9

750

501.

470

834

9.7

767

512.

872

4C

yana

mid

e2.

177

0.99

21.

485

1.05

11.

985

4.35

42.

910

4.11

12.

047

4.49

03.

001

4.24

0C

yflut

hrin

0.00

30.

001

0.00

20.

001

5.83

912

.81

8.56

12.1

01.

505

3.30

12.

206

3.11

7C

ymox

anil

0.03

10.

014

0.02

10.

015

156.

334

3.0

229.

232

3.8

156.

334

3.0

229.

232

3.8

Cyp

roco

nazo

le0.

003

0.00

10.

002

0.00

128

6.6

628.

842

0.2

593.

712

7.5

279.

818

7.0

264.

2C

ypro

dini

l0.

042

0.01

90.

029

0.02

03,

511

7,70

25,

148

7,27

353

7.4

1,17

978

8.0

1,11

3D

imet

hoat

e0.

040

0.01

80.

027

0.01

90.

566

1.24

20.

830

1.17

3D

imet

omor

ph0.

035

0.01

60.

024

0.01

751

5.9

1,13

275

6.4

1,06

951

5.9

1,13

275

6.4

1,06

9D

ithi

anon

0.09

20.

042

0.06

30.

044

455.

91,

000

668.

694

545

5.9

1,00

066

8.6

945

Dod

ine

0.09

10.

041

0.06

20.

044

198.

443

5.3

290.

941

1.0

440.

996

764

6.5

913

Eto

fenp

rox

0.02

90.

013

0.02

00.

014

0.14

30.

314

0.21

00.

297

0.23

90.

524

0.35

00.

495

Fen

buta

tin

oxid

e0.

077

0.03

50.

052

0.03

711

.19

24.5

516

.41

23.1

861

.04

133.

989

.51

126.

5F

orm

etan

ate

hydr

oxid

e0.

174

0.08

00.

119

0.08

40.

053

0.11

70.

078

0.11

00.

045

0.09

80.

066

0.09

3F

oset

yl-A

l0.

181

0.08

30.

124

0.08

84,

409

9,67

36,

465

9,13

4H

exyt

hiaz

ox0.

007

0.00

30.

005

0.00

342

9.9

943

630.

489

194

.43

207.

213

8.5

195.

6L

ambd

a-cy

halo

thri

n0.

005

0.00

20.

004

0.00

30.

382

0.83

80.

560

0.79

2L

ufen

uron

0.00

70.

003

0.00

50.

003

143.

331

4.4

210.

129

6.9

30.0

966

.02

44.1

362

.34

Mal

athi

on0.

140

0.06

40.

095

0.06

70.

004

0.01

00.

006

0.00

96.

119

13.4

28.

9712

.68

Met

hioc

arb

0.14

00.

064

0.09

50.

067

0.03

40.

074

0.04

90.

070

2.41

45.

297

3.54

15.

002

Met

hom

yl0.

056

0.02

50.

038

0.02

70.

045

0.09

80.

066

0.09

32.

293

5.03

03.

362

4.75

0P

irim

icar

b0.

049

0.02

20.

033

0.02

412

6.9

278.

518

6.1

262.

917

0.9

375.

025

0.6

354.

1P

ropa

rgit

e0.

040

0.01

80.

027

0.01

91.

810

3.97

12.

654

3.75

01.

131

2.48

21.

659

2.34

4P

yrid

aben

0.01

70.

008

0.01

10.

008

28.0

661

.57

41.1

558

.13

28.0

661

.57

41.1

558

.13

Tau

-fluv

alin

ate

0.01

70.

008

0.01

10.

008

0.02

90.

064

0.04

30.

061

0.28

70.

629

0.42

00.

594

Teb

ucon

azol

e0.

009

0.00

40.

006

0.00

566

.34

145.

597

.28

137.

462

.09

136.

291

.05

128.

6Z

eta-

cype

rmet

hrin

0.00

30.

002

0.00

20.

002

15.0

232

.94

22.0

231

.11

69.2

015

1.8

101.

514

3.4

Ecotoxicological effects of pesticide drift in hedgerows Environ. Toxicol. Chem. 28, 2009 859

Fig. 3. Example of the predicted environmental concentration (PEC), median lethal rate (LR50), and application (appl.) rate values for propargitefor two optical porosity (OP) and wind speed (WS) combinations in the simplified scenario (hedgerow parallel to crop rows and pesticide sprayeddirectly on the hedgerow), with double PEC and LR50 scale. In each graph the vertical red and blue solid lines indicate LR50 for Aphidiusrhopalosiphi and Typhlodromus pyri, respectively, the vertical green solid line the application rate, the vertical black solid line the average PECfor the whole hedgerow, and the solid black curve its vertical pattern distribution calculated with Equation 3. The six dotted curves are the PECvalues for each 1-m-wide hedgerow section: blue � 0–1 m; red � 1–2 m; green � 2–3 m; fuchsia � 3–4 m; black � 4–5 m; gray � 5–6 m;in each graph these curves are from right, blue dotted (the section nearest the sprayer), to left, gray dotted (the section farthest from the sprayer).

tical one in Figure 3 is the variation of the average PEC withthe height: The PEC is higher (0.065 �g/cm2) at a height ofapproximately 1.5 m and lower at a height above 3 m or below0.5 m. On average and for any height, T. pyri is completelysafe only when it is in the deeper sections of hedgerow, startingfrom the 4 to 5 m section; instead, A. rhopalosiphi is alwayssafe except when it is in the two sections nearest to the sprayer,even if the risk is quite low in the higher or lower part. Sofor A. rhopalosiphi, the TER � 1 (i.e., on the basis of theaverage PEC value), it is safe, reliable index of the low-riskconditions in the hedgerow, whereas the real risk conditionsfor T. pyri are better described by the 2-d pattern of the PEC.

The conclusions would be different when, still with pro-pargite and with OP � 0.7, a higher wind speed (WS � 2.0)is considered. In these conditions a fraction of 0.679 bypassesthe hedgerow, which retains a fraction of 0.956 � (1 � 0.679)� 0.307 that corresponds to (142.50 � 0.307) � 43.68 g/haof propargite, or 0.018 �g/cm2, which is the average PEC inthe hedgerow. This is less than half of the average PEC cal-culated in the previous scenario because the higher wind speedinduces different partitions of the PEC throughout the hedge-row width, and both nontarget arthropods are safe in any partof the hedgerow. It is worth noting that when those PECs areexpressed in terms of mass fraction (mg/kg) the results are inagreement with the values reported by other researchers ([30],p 2205, Table 7]) and used in support of worst-case, early-tierassessments of terrestrial nontarget organisms.

This is what the model calculates for the simplified sce-nario, where a sprayer sprays the hedgerow directly, but inreal field conditions the hedgerow would receive the drift gen-erated by the treatment to the crop row, which is the standardscenario. With the same model for crop interception of driftas that used for the hedgerow, when OP � 0.7 and WS � 2.0m/s, only a fraction of 0.641 of the application rate impactson the hedgerow (Table 2), and in the case of propargite, thiscorresponds to (142.5 � 0.641) � 91.34 g/ha. This scale factorof 0.641 can be included in the TER calculation to accountfor specific spraying conditions. In the case of low wind speed,

the fraction reaching the hedgerow is low, and the scale factoris 0.136.

On average, when OP � 0.7, a proportion of [(0.641 �0.136)/2] � 0.389 of the application rate impacts on the hedge-row. A very similar proportion (0.302) is obtained in the caseof low optical porosities. Therefore, in the standard scenarioaround one-third of the application rate bypasses the crop andimpacts on the hedgerow (compared with four-fifths to five-fifths in the simplified scenario). For the four OP–WS scenario,the fraction bypassing the crop ranges from 0.063 to 0.641.These fractions are in full agreement with the interceptionvalues reported in specific studies (i.e., in Linders et al. [30]),the interception ranges from 50% (leaf development) to 85%(ripening) (p 2203, Table 4) or from 30% (leaf development)to 80% (flowering/development of fruit/ripening) (p 2214, Ta-ble 10).

Equation 2 can be applied iteratively each time that thespray drift impacts a crop row or hedgerow and can be usedto model virtually all types of crop–hedgerow systems. In ascenario where many crop rows are sprayed and each rungenerates a drift directed at the hedgerow, all drifts are ac-cumulated in the hedgerow. In the case of 10 rows, the scalefactor ranged from 1.5 (OP � 0.3) to 2.0 (OP � 0.7) timesthe rate applied to a single crop row in one run. It should beremembered that the calculation of the scale factor was per-formed with a model built for a hedgerow, so it is likely notfully adequate for a crop, but nonetheless the results suggestthat when the crop has sparse vegetation the adjacent hedgerowcontamination can be very high. Models are applied under theassumption that the spray drift cannot pass over crop rows,but it must be taken into account that this assumption can beviolated. For example, Davis et al. [31] have shown that sub-stantial amounts of pesticide travel above a 2-m hedge. Onthe one hand, the model is reliable in as much as the hypothesesare respected; on the other hand, its application can take ad-vantage of specific environmental conditions or experimentalresults that can be fully implemented (e.g., through modifi-

860 Environ. Toxicol. Chem. 28, 2009 S. Otto et al.

cation of the applied rate or the fraction drifting directly to aspecific distance, bypassing the crop rows).

This approach to the calculation of the PEC differs fromthat in HQ. It shows that a simple relationship exists betweendrift factor and wind speed and optical porosity and that thelatter is also linked to the vegetation distribution factor. Buteven with the same drift factor and vegetation distributionfactor, this approach can provide a more precise estimate ofthe PEC on leaves. First, it shows that even with only oneapplication the hedgerow can accumulate the drift originatedfrom the treatment of each crop row. Second, with the HQapproach, the PECs have the same dimension as the applicationrate, usually g/ha, whereas in this the PECs are mass depositionon leaf surfaces (�g/cm2). Third, it also shows how the (av-erage) PEC varies inside the hedgerow canopy. Finally, thepresent study focuses on direct comparison between the PECand the LR50, and, as in HQ, a proper correction (safety) factor(e.g., 10) can be introduced, being aware that this would actonly as a scale factor reducing the portion of the hedgerowwhere beneficial arthropods are supposed to be safe.

TER values and main sources of variability

Besides the two case study scenarios, Equations 2 and 3can be applied to any scenario of interest. In any case, thedrift deposition values on the hedgerow (the PECs) are notequally distributed on the vertical profile, the higher valuesbeing between the heights of 1.0 and 1.5 m, which is the heightof the sprayer’s fan. As a consequence, the risk level is gen-erally higher in this range than that in the remaining profile,so it would be considerable even with low-toxicity pesticides.Risk levels in the various parts of the hedgerow vary in a2-d pattern. The deeper the considered section is, the lowerthe risk, which is minimum on the side of the hedgerow op-posite the sprayer. For a given distance from the sprayer, thelevel diminishes down toward the soil and more quickly higherup, according to the bell-shaped pattern of the PEC. Thus,some areas remain protected (e.g., those higher, deep in thecanopy, or both).

The main source of variability among the different sce-narios is pesticide toxicity (Table 3 and Fig. 4). Applicationrate is also important. Furthermore, low wind speed increasesthe PEC especially when the optical porosity is low (i.e., withOP � 0.3 and WS � 0.5 the hedgerow intercepts 98.2% ofthe application rate) (Table 2). For the 28 considered activeingredients, there may be many combinations of toxicity, ap-plication rate, and environmental conditions, but the simula-tions show that a grouping in four case studies is indeed pos-sible, and Figure 4 reports six examples.

When a low-toxicity pesticide for both nontarget arthropodsis used, the risk in the hedgerow is always low in any windcondition and porosity considered, even on sections nearest tothe crop and sprayer. This is the case for most of the activeingredients used in the simulations (i.e., cymoxanil, cypro-conazole, dithianon, lufenuron [Fig. 4, first row], pirimicarb,and others for a total of 15 active ingredients), for which inall four scenarios 1 � TER � 3 (Eqn. 4), and the risk is low.

When toxicity is very high for both nontarget arthropods,the risk is always high, and on average, TER � 0.75, whenthe application rate is both low (i.e., etofenprox) and high(formetanate hydroxide) (Fig. 4, second row).

Between these two extreme and concordant situations, an-other is possible. When for the same active ingredient thetoxicity for the two chosen nontarget arthropods is very dif-

ferent, the risk conditions would result as extreme but opposite(i.e., high for one nontarget arthropod and low for the other).For example, for carbaryl, malathion (Fig. 4, third row), meth-iocarb, and methomyl, there is always a high risk for A. rho-palosiphi and a low risk for T. pyri.

But the most interesting situations are those where the PECsare close to the LR50, when intermediate risk conditions areobtained for at least one nontarget arthropod and some partsor sections of the hedgerow are not or partially affected, high-lighting the importance of the 2-d pattern distribution of drift,depending on wind speed and porosity conditions. For ex-ample, for cyanamide, the risk is low for both nontarget ar-thropods when wind speed is high, but when wind speed islow, there is a certain risk, but only in the first section of thehedgerow and only at sprayer height. When the toxicity in-creases, the portion of the hedgerow unaffected or less affectednarrows; just the highest part of the canopy on the oppositeside from the sprayer remains unaffected and only if windspeed is low. This is the case of alpha-cypermethrin (Fig. 4,fourth row), dimethoate, and lambda-cyhalothrin for A. rho-palosiphi, which is safe only in the deepest sections of thehedgerow, propargite (Fig. 4, fifth row) and cyfluthrin for bothnontarget arthropods, and tau-fluvalinate for T. pyri (Fig. 4,sixth row). In other conditions or parts of the hedgerow, thePEC curves remain on the right of the LR50 lines, so TER �1, and risk is high.

As a general rule, wind blows the spray drift deeper throughthe hedgerow and dilutes the pesticide over a greater foliagesurface, so as wind speed increases the deposition reduces onthe first canopy sections and increases on the subsequent ones,making deposition more homogeneous throughout the hedge-row. This implies that, with a wind speed higher than 2.0m/s and low-toxicity pesticide, the risk for the arthropods inthe hedgerow is low even on the sprayed side, whereas withmore toxic pesticides the risk is high in practically all of thehedgerow. Optical porosity, which is negatively correlated toleaf area, influences deposition dynamics: Low optical porositylevels preserve the side of hedgerow opposite that of the spray-er, although they may play an important role especially withlow wind speeds, because with high wind speed all of thehedgerow will be affected, even when optical porosity is low.In general, the use of high-toxicity pesticides affects largeportions of the hedgerow, whereas with low-toxicity pesticidesdamage can be null or negligible, except, at worst, in the partof the canopy near the sprayer.

Risk assessment for soil and herbaceous vegetation

The TER on soil and vegetation beneath and beyond thehedgerow was only low when high-toxicity pesticides wereused; nonetheless, the hedgerow seemed able to protect partof the soil surface on the side opposite of the spray source,even in severe simulation conditions.

In general, high-risk conditions are possible beneath thehedgerow in the case of high-toxicity products, but the risklevel decreases significantly moving from the side near thesprayer toward the far side, unless with higher wind speed:High-risk conditions can be reached even in the vegetationbeyond the hedgerow at a distance of 6 to 15 m from thesprayer. With low wind speeds, the ecotoxicological risk be-yond the hedgerow is very low, confirming the barrier effectof the hedgerow.

Ecotoxicological effects of pesticide drift in hedgerows Environ. Toxicol. Chem. 28, 2009 861

Fig. 4. Predicted environmental concentration (PEC), median lethal rate (LR50), and application rate values (all in �g/cm2) for six active ingredientsin the four simulation conditions in the simplified scenario (hedgerow parallel to crop rows and pesticide sprayed directly on the hedgerow). Forgraph legend, see Figure 3.

862 Environ. Toxicol. Chem. 28, 2009 S. Otto et al.

CONCLUSIONS

With drift and deposition distribution models in definedsimulation scenarios, where some assumptions hold [20], it ispossible to estimate both the hedgerow effect on drift and theeffect of drift on the hedgerow when a pesticide is sprayed.The results have shown that hedgerows are a very effectivebarrier against spray drift and that only a small fraction of thedrift can pass through a hedgerow with low optical porosity[20].

Hedgerows may thus be considered a valid aid to spraydrift interception and can play an important role in environ-mental protection. However, hedgerows also provide shelterand feeding grounds for beneficial arthropods [32]. To maintainthe biota function, close attention must be paid to choice ofpesticide and environmental conditions when spraying, be-cause the hedgerow can receive a spray contribution fromvirtually all of the field runs, and even if not directly sprayed,the hedgerow could intercept an amount of pesticide higherthan that sprayed on a single crop row. This conclusion needsan independent quantitative validation.

For a given pesticide, a close relationship exists betweenthe PECs and the environmental conditions during application.When low-toxicity pesticides are used in low wind speed con-ditions (i.e., less than 1.0 m/s), no danger levels occur fornontarget arthropods in the hedgerow. On the contrary, withmore toxic pesticides and high wind speeds, the hedgerowfauna would be variously affected. An approach that takes intoaccount the spatial pattern of drift deposition could be usefulfor studying and understanding the realistic behavior of non-target arthropods when risk conditions are intermediate, so asurvival pattern of these arthropods is likely and can improvethe tier approach based on HQ trigger values. Furthermore,the 2-d spatial distribution of the PECs can constitute a basefor a probabilistic approach to a refined risk assessment inwhich the 95th percentiles of the PECs may be used insteadof the mean deterministic PEC of the entire hedgerow. Thiscould allow elimination of the HQ trigger value of 2.

When a hedgerow intercepts a significant amount of pes-ticide drift, the field beyond it is protected, so it would be ofinterest to know the preferential habitat of the nontarget ar-thropods and consider it in the risk assessment, because theirfate would be different if this is the tree canopy or soil veg-etation. T. pyri and A. rhopalosiphi were chosen as represen-tative nontarget arthropods species because they are very sen-sitive to pesticides, but they are also representative of differenthabitat and feeding habits. T. pyri is a generalist phytoseiidthat feeds on spider mites, insects, and pollen [33,34], and itsabundance is likely to be higher in the hedgerow. A. rhopa-losiphi is instead a specialist parasitoid, and its abundanceshould be linked to that of its prey [35]. The population re-sponse during or soon after the treatments would also be ofinterest, but specific studies are needed [36]. In general, manynontarget arthropods live in crops, and their distribution in avineyard–hedgerow system is of great importance because therisk can also be low for a very sensitive species when itspreferential habitat allows reduced exposure.

Hedgerows are useful as a barrier for spray drift but mustalso be considered sites for protection, because they are ageneral source of biodiversity [37]. Recent surveys carried outon farms in northern Italy showed the importance of ecologicalcompensation areas for maintaining arthropod populations[38], and it is extremely important that policymakers and man-

agers understand the enormous value of such landscape com-ponents, which are often mistakenly considered as marginal.

Acknowledgement—This research was partly financed by the ItalianMinistry of University and Research (Project: Geographical Infor-mation Systems (GIS)-based assessment and spatial distribution ofthe ecotoxicological risk deriving from pesticide use, Progetti di Ril-evante Interesse Nazionale, PRIN 2002).

REFERENCES

1. Boller EF, Hani F, Poehling HM. 2004. Ecological Infrastruc-tures. Idealbook on Functional Biodiversity at the Farm Level.Temperate Zones of Europe. Mattenbach AG, Winterthur, Swit-zerland.

2. Altieri MA, Nicholls CI. 2004. Biodiversity and Pest Manage-ment in Agroecosystems, 2nd ed. The Haworth Press, New York,NY, USA.

3. Ucar T, Hall FRH. 2001. Windbreaks as a pesticide drift mitigationstrategy: A review. Pest Manag Sci 57:663–675.

4. Hewitt J. 2001. Drift Filtration by Natural and Artificial Col-lectors: A Literature Review. Stewart Agricultural Research Ser-vices, Macon, GA, USA.

5. Langhof M, Gathmann A, Poehling HM, Meyhofer R. 2003. Im-pact of insecticide drift on aphids and their parasitoids: Residualtoxicity, persistence and recolonisation. Agric Ecosyst Environ94:265–274.

6. Richardson GM, Walklate PJ, Baker DE. 2004. Spray drift fromapple orchards with deciduous windbreaks. Aspects of AppliedBiology 71:149–156.

7. Van de Zande JC, Michielsen JMGP, Stallinga H, De Jong A.2000. The effect of windbreak height and air assistance on ex-posure of surface water via spray drift. Proceedings, British CropProtection Conference, Pests and Diseases, Brighton, UK, No-vember, 13–16, pp 91–96.

8. Leistra M, Zweers AJ, Warinton JS, Crum SJH, Hand LH, Belt-man WHJ, Maund SJ. 2003. Fate of the insecticide lambda-cy-halothrin in ditch enclosures differing in vegetation density. PestManag Sci 60:75–84.

9. Crane M, Whitehouse P, Comber S, Watts C, Giddings J, MooreDRJ, Grist E. 2003. Evaluation of probabilistic risk assessmentof pesticides in the UK: Chlorpyrifos use on top fruit. Pest ManagSci 59:512–526.

10. European Commission. 1991. Council Directive of 15 July 1991concerning the placing of plant protection products on the market.91/414/EEC. Official Journal of the European Communities L230.Brussels, Belgium.

11. European Commission. 2002. Guidance document on terrestrialecotoxicology under council directive 91/414/EEC. Working Doc-ument SANCO/10329/2002. Brussels, Belgium.

12. Candolfi MP, Barrett KL, Campbell PJ, Forster R, Grandy N,Huet MC, Lewis G, Oomen PA, Schmuck R, Vogt H, eds. 2000.ESCORT 2. Guidance Document on Regulatory Testing and RiskAssessment Procedures For Plant Protection Products with Non-target Arthropods. A Joint BART, EPPO/CoE, OECD, and IOBCWorkshop, Wageningen, The Netherlands, March 21–23. IOBC/WPRS, Gent, Belgium, p 50.

13. Candolfi MP, Bakker F, Canez V, Miles M, Neumann C, PillingE, Primiani M, Romijn K, Schmuck R, Storck-Weyhermuller S,Ufer A, Waltersdorfer A. 1999. Sensitivity of non-target arthro-pods to plant protection products: Could Typhlodromus pyri andAphidius spp. be used as indicator species? Chemosphere 39:1357–1370.

14. Vogt H, Brown K, Candolfi M, Kuhner C, Moll M, Travis A,Ufer A, Waldburger M, Waltersdorfer A, Bigler F. 2000. Labo-ratory method to test effects of plant protection products on larvaeof Chrysoperla carnea (Neuroptera: Chrysopidae). In CandolfiMP, Blumel S, Forster R, Bakker FM, Grimm C, Hassan SA,Heimbach U, Mead-Briggs MA, Reber B, Schmuck R, Vogt H.eds, Guidelines to Evaluate Side-Effects of Plant ProtectionProducts to Non-target Arthropods. IOBC/WPRS, Gent, Bel-gium, pp 27–44.

15. Grimm C, Schmidli H, Bakker F, Brown K, Campbell P, CandolfiM, Chapman P, Harrison EG, Mead-Briggs M, Schmuck R, UferA. 2001. Use of standard toxicity tests with Typhlodromus pyriand Aphidius rhopalosiphi to establish a dose-response relation-ship. Anzeiger fur Schadlingskunde 74:72–84.

Ecotoxicological effects of pesticide drift in hedgerows Environ. Toxicol. Chem. 28, 2009 863

16. Gonzalez-Valero JF, Campbell PJ, Fritsch HJ, Grau R, RomijnK. 2000. Exposure assessment for terrestrial non-target arthro-pods. J Pest Sci 73:163–168.

17. Biologische Bundesanstalt fur Land- und Forstwirtschaft (FederalBiological Agency of Agriculture and Forestry, Germany). 2000.Bekanntmachung des Verzeichnisses Risikomidernder Anwen-dungsbedingungen fur Nichtzielorganism. Bundesanzeiger 100:9878–9880.

18. Candolfi MP, Blumel S, Forster R, Bakker F, Grimm C, HassanSA, Heimbach U, Mead-Briggs MA, Reber B, Schmuck R, VogtH. 2000. Guidelines to evaluate side-effects of plant protectionproducts on non-target arthropods. IOBC, BART and EPPO JointInitiative. IOBC/WPRS, Ghent, Belgium.

19. Langhof M, Gathmann A, Poehling HM. 2005. Insecticide driftdeposition on noncrop plant surfaces and its impact on two ben-eficial nontarget arthropods, Aphidius colemani Viereck (Hy-menoptera, Braconidae) and Coccinella septempunctata L. (Co-leoptera, Coccinellidae). Environ Toxicol Chem 24:2045–2054.

20. Lazzaro L, Otto S, Zanin G. 2008. Role of hedgerows in inter-cepting spray drift: Evaluation and modelling of the effects. AgricEcosyst Environ 123:317–327.

21. Kenney WA. 1987. A method for estimating windbreak porosityusing digitized photographic silhouettes. Agric For Meteorol 39:91–94.

22. Heisler GM, DeWalle DR. 1988. Effects of windbreak structureon wind flow. Agric Ecosyst Environ 22/23:41–69.

23. Balsari P, Marucco P, Tamagnone M. 2007. A test bench for theclassification of boom sprayers according to drift risk. Crop Prot26:1482–1489.

24. Statsoft. 2005. Statistica 7.1. Tulsa, OK, USA.25. Valmori I. 2000. Fitofarmaci per Coltura. Calderini Edagricole,

Bologna, Italy (in Italian).26. Muccinelli M. 2004. Prontuario dei Fitofarmaci, 10th ed. Eda-

gricole, Bologna, Italy.27. Hartnik T, Sverdrup LE, Jensen J. 2008. Toxicity of the pesticide

alpha-cypermethrin to four soil nontarget invertebrates and im-plications for risk assessment. Environ Toxicol Chem 27:1408–1415.

28. Savi D. 2001. Attrezzature per la Difesa delle Piante—Guida

alla Scelta e al Corretto Impiego. Edizioni L’Informatore Agra-rio, Verona, Italy (in Italian).

29. Davis BNK, Williams CT. 1990. Buffer zone widths for honey-bees from ground and aerial spraying of insecticides. EnvironPollut 63:247–259.

30. Linders J, Mensink H, Stephenson G, Wauchope D, Racke K.2000. Foliar interception and retention values after pesticide ap-plication. A proposal for standardized values for environmentalrisk assessment. Pure Appl Chem 11:2199–2218.

31. Davis BNK, Brown MJ, Frost AJ. 1993. Selection of receptorsfor measuring spray drift deposition and comparison with bio-assays with special reference to the shelter effect of hedges. Pro-ceedings, Brighton Crop Protection Conference—Weeds, Brigh-ton, UK, November 26–29, pp 139–144.

32. Holland J, Fahrig L. 2000. Effect of woody borders on insectdensity and diversity in crop fields: a landscape-scale analysis.Agric Ecosyst Environ 78:115–122.

33. Luh H-K, Croft BA. 2001. Quantitative classification of life-styletypes in predaceous phytoseiid mites. Exp Appl Acarol 25:403–424.

34. Duso C, Pozzebon A, Capuzzo C, Bisol PM, Otto S. 2003. Grapedowny mildew spread and mite seasonal abundance in vineyards:Evidence for the predatory mites Amblyseius andersoni and Ty-phlodromus pyri. Biological Control 27:229–241.

35. Micha SG, Kistenmacher S, Molk G, Wyss U. 2000. Tritrophicinteractions between cereals, aphids and parasitoids: Discrimi-nation of different plant-host complexes by Aphidius rhopalosiphi(Hymenoptera: Aphidiidae). Eur J Entomol 97:539–543.

36. Hazel RP, Andrew JE, Derek M. 2006. Aphid population responseto agricultural landscape change: a spatially explicit, individual-based model. Ecol Model 4:451–463.

37. Marshall EJP. 2001. Biodiversity, herbicides, and nontargetplants. Pesticide behaviour in soils and water. British Crop Pro-tection Council 78:419–426.

38. Burgio G, Ferrari R, Pozzati M, Boriani L. 2004. The role ofecological compensation areas on predator populations: An anal-ysis on biodiversity and phenology of Coccinellidae (Coleoptera)on non-crop plants within hedgerows in Northern Italy. Bulletinof Insectology 57:1–10.