- Email: [email protected]
Insecticide drift from ground-based, hydraulic spraying of peas and brussels sprouts: bioassays for determining buffer zones
Agriculture, Ecosystems and Environment, 43 ( 1993) 93-108
93
Elsevier Science Publishers B.V., Amsterdam
Insecticide drift from ground-based, hydraulic spraying of peas and brussels sprouts: bioassays for determining buffer zones B.N.K. Davis, K.H. Lakhani, T.J. Yates, A.J. Frost and R.A. Plant NERC, Institute of Terrestrial Ecology, Monks Wood,Abbots Ripton, Huntingdon, PE17 2LS, UK (Accepted 15 June 1992)
ABSTRACT Davis, B.N.K., Lakhani, K.H., Yates, T.J., Frost, A.J. and Plant, R.A., 1993. Insecticide drift from ground-based, hydraulic spraying of peas and brussels sprouts: bioassays for determining buffer zones. Agric. EcosystemsEnviron., 43: 93-108. Studies were made on insecticide spray drift from conventional hydraulic sprayers under typical agricultural conditions. A bioassay technique with 2-day-old Pieris brassicaelarvae was used to assess drift of cypermethrin and triazophos from pea and brassica crops on six occasions. Earlier work had shown that this species was a suitable model for other butterflies. Replicated series of target larvae were placed downwind of the sprayed crops, and mortality data used to obtain profiles of spray drift related to physical features such as roadways, adjacent crops and hedges. Estimates of the distances at which 50%, 20% and 10% mortality occurred were obtained graphically or by fitting logistic models to the data. Buffer zones that limited mortality to 10% or less were between 16 and 24 m for cypermethrin and ! 2 m for triazophos. These estimates made no allowance for control mortality of about 5%, however, they were based conservatively on short-term effects. Hedges also probably limited the full extent of drift in two instances. The wider application of these results is discussed and comparisons are made with the use of unsprayed crop headlands.
INTRODUCTION
There have been many studies on spray drift using tracer dyes collected on fiat plates, tapes, nylon lines or other receptors; deposits are then rinsed off and measured by colorimetry or fluorimetry (e.g. Nordby and Skuterud, 1974; Byass and Lake, 1977; Grover et al., 1978; Maybank et al., 1978; Lloyd and Bell, 1982; Gilbert and Bell, 1988; Parkin and Merritt, 1988). Other techniques include atomic absorption spectrometry, which has been used to measure zinc and manganese in fungicide sprays (Kiimmel et al., 1991 ); colorimetric determination of acid-extracted copper in copper fungicides Correspondence to."B.N.K. Davis, NERC, Institute of Terrestrial Ecology, Monks Wood, Abbots Ripton, Huntingdon, PE 17 2LS, UK.
94
B.N.K. DAVIS ET AL.
(Herrington et al., 1981 ); measurements of conductivity for sprays containing calcium nitrate (Hall et al., 1991 ); and neutron activation of dysprosium used as a tracer (Dobson et al., 1983). It is also possible to measure drift deposits of some organic pesticides using gas liquid chromatography or high pressure liquid chromatography e.g. carbaryl, deltamethrin and glyphosate (Currier et al., 1982; Riley and Wiesner, 1989; Payne et al., 1990). All these methods show the relationship of drift deposits with distance, and the effects of particular variables such as nozzel type, boom height, wind speed, etc. However, they do not give direct information on the biological impact of drift. In the case of insecticides, such effects are not readily observed but two approaches are possible. Short range and generally short-term effects can be recorded in natural populations or communities of non-target organisms in habitats adjacent to sprayed crops (Dobson, 1986; Williams et al., 1987). This can be useful for distinguishing relatively sensitive and insensitive groups for different spray materials. However, bioassay experiments are more appropriate for measuring the width of buffer zones to protect beneficial insects and wildlife. Here, organisms are exposed in a controlled manner during pest control spraying. Davis et al. ( 1991 a) described a bioassay technique for evaluating the effects of wind speed on insecticide spray drift from a tractor-powered hydraulic sprayer. In order to examine the influence of this one variable other factors were maintained as uniformly as possible. Thus, site and operating conditions were controlled by using a field of short grass, by orienting the sprayed swathes at right angles to the prevailing wind on each occasion, by setting out targets (also in short grass ) in line with the wind, and by using the same compound at the same rate for each trial. This paper describes bioassay measurements of drift made in agricultural crops during spraying by farmers. The objectives were: ( 1 ) to see whether the techniques previously developed for controlled experiments could be used under typical agricultural conditions; (2) to relate drift deposition and its effects to physical features downwind of sprayed crops; (3) to compare results with those obtained under standardized experimental conditions at Monks Wood in 1988 and 1989; (4) to estimate the size of buffer zones that would protect species outside the sprayed crop. MATERIALS AND METHODS
Two insecticides were selected for these tests, cypermethrin and triazophos which are used against a wide spectrum of pests including Lepidoptera. The ground crops were peas and brussels sprouts which are sprayed for control of Pea-moth (Cydia nigricana (Fab.)) and Cabbage white (Pieris spp. ) caterpillars. Topical dosing experiments on Pieris brassicae L. were first carried
INSECTICIDE SPRAY DRIFT
95
out to obtain LDso values for cypermethrin and triazophos, and to compare these with the range of toxicity values already obtained for eight other compounds (Sinha et al., 1990). In the case of triazophos, 2-, 3- and 4-day-old larvae were compared simultaneously to determine the effect of age on sensitivity. An experimental drift trial with cypermethrin was also carried out at Monks Wood, using larvae of three different ages in case older larvae had to be used for field bioassays because of delayed spraying. Four farms were selected in Cambridgeshire and Bedfordshire. Six to seven days notice of spraying was needed to have 2-day-old P. brassicaelarvae ready on the day of spraying, but on the first occasion at Haverhill insufficient notice was given and the larvae were only 1-day-old. Bioassay targets were prepared essentially in the way described by Davis et al. ( 1991 a). To optimize the initial feeding of larvae, egg batches were placed directly on Horse-radish plants (Armoracia rusticana), and kept in shaded, humid conditions at about 20°C. The day after hatching, the larvae were transferred to freshly cut leaves selected and trimmed to be as similar as possible. These leaves were stuck in flower arrangers' foam ('Oasis') in plastic pots flooded with water. On the following day they were transported to the test site (30-80 km). The targets were set out in two to four replicate lines about 10 m apart, perpendicular to the downwind field edge at varying measured distances up to 25 m. Wind speed and direction were recorded on a portable anemograph (Vector R500, Vector Instruments, Rhyl). When the direction was clearly oblique to the target lines at an angle 0, the measured distance d was adjusted to the effective distance d sec 0. The pots were placed on the ground at field edges, on road verges and in low crops, but at Haverhill where there was a tall crop of field beans on the downwind side the pots were suspended from collapsible wooden tripods so that the insects were just above crop height. After spraying, the larvae were placed in Petri dishes on sections of sprayed leaf, transported back and kept in an incubator at 24°C. Mortality was assessed over 3 days. A water-sensitive paper was pinned out next to each target leaf. Although 'inefficient' as collectors of fine droplets, they provided a visual indication of drift deposition which was measured on an I2S image processing system as described by Sinha et al. (1990). An attempt was made at Haverhill to measure the cypermethrin drift directly by using 'efficient' aluminium mesh cylinders ('Isopon', 10 cm circumference×6 cm long) suspended next to the target leaves. These were washed in 10 ml hexane which was injected onto a Varian model 3400 gas liquid chromatograph using a capillary column and electron capture detector. However, the cypermethrin could not be detected even at 1 m downwind owing to the very low application rate (25 g a.i. ha -1 is equivalent to 0.25 #g cm-2).
96
B.N.K. DAVISET AL.
Statistical analyses The determinations of LDso from topical dosing with cypermethrin and triazophos were based on probit analyses of the dose response relationships using the Maximum Likelihood Programme (Ross, 1980 ). The estimates from the separate replicate experiments were pooled to obtain an overall weighted mean estimate as described in Davis et al. ( 199 lb). For field data, estimates of mortality, with standard errors, were required for each distance from the replicate transects. These were obtained by pooling the numbers of larvae exposed at each distance and noting the total number of larvae dying. Thus, using the binomial model, the mortality rate p was estimated by the observed proportion dying
#=ErdEn, where r; is number of deaths, n; is number exposed, at a given distance on the ith transect and S.E. (#) =,4~/~( 1-,0)/En,]. The distances at which larvae would be expected to suffer 50%, 20% and 10% mortality were estimated at Upper Caldecote 1 and Swavesey by fitting the generalized logistic model to the data as described by Davis et al. ( 1991a),
P=y+~/[ 1+exp(a-fld) ] where P is percentage dying at distance d. RESULTS
Laboratory toxicity tests Cypermethrin was found to be the most toxic of the ten compounds so far tested against P. brassicae, being nearly 100 times as toxic as endosulfan (used for standardized comparisons in Sinha et al., 1990), 6.6 times as toxic as triazophos and 3.8 times as toxic as diflubenzuron used in earlier drift trials (Tables 1 and 2). The dose response curve for cypermethrin was found to be very TABLE1 Toxicity of cypermethrin and triazophos in topical applications to P. brassicae larvae. LDso estimates from probit analysis_+ standard errors as a percentage of the estimates Insecticide
Larval age (days)
Cypermethrin Triazophos
2 2 3 4
Concentration_+ SE (%) 6.48X 10 -5 6.5 4.26X 10 -4 7.7
10.55× l0 -4 5.3 17.86× l0 -4 6.0
/lg per insect
/lg g - t
0.00016 0.0011 0.0026 0.0045
0.231 1.521 3.103 3.283
97
INSECTICIDE SPRAY DRIFT TABLE 2
Relative field hazard of four insecticides to 2-day-old P. brassicae larvae based on topical dose and application rate (active intredient) to field crops Insecticide
LDso (/tgg -1 )
Field rate ( g h a -I )
'Field hazard' = rate/LDso (million LDso units h a - ~)
Dimethoate Cypermethrin Diflubenzuron Triazophos
627 0.23 0.87 1.52
400 25 100 3361
0.6 108 115 221
IRate for peas, 1050 for carrots (full rate).
steep around the LDso value, i.e. a small change in dose concentration produced a large change in percentage mortality. The above comparisons of relative toxicity are therefore only valid for the LDso dose. In the case of triazophos, the LDso doses increased with 3- and 4-day-old larvae as expected. When the dose per unit weight is considered, the effective rate was doubled for 3-day-old larvae but showed little further increase for 4day-old larvae. Contact toxicity in the field, however, is determined as much by application rate (and formulation) as by intrinsic toxicity. Thus, Table 2 shows that the 'hazard index' for cypermethrin is comparable with that for diflubenzuron and about 40% of that for triazophos at the rate used on peas. All three compounds also act as stomach poisons so the actual field toxicity may often depend on feeding behaviour rather than on direct exposure to spray drift.
Drift trial at Monks Wood The pilot drift bioassay with cypermethrin at Monks Wood gave results that were comparable with those obtained for diflubenzuron, at a wind speed of 2.5 m s- 1 (Davis et at., 1991 a ), in that 20% mortality lay between 5 m and 10 m downwind for the youngest group of larvae (Table 3 ). However, it should be noted that those larvae were a day older than in the standard diflubenzuron trials, and that there was not a great difference in sensitivity among the three age groups. The lack of clear-cut differences in this drift trial may have been owing partly to the limited sample size (a single transect for each age group), but also to the greater consumption of pesticide residues, and therefore increased stomach poisoning mentioned above.
Crop spraying bioassays Site details Six spraying operations were studied, one in 1989 and five in 1990. Table 4 summarises site, spraying and meteorological variables. An aphicide, de-
98
B.N.IC DAVISET AL.
TABLE 3 Pilot drift bioassay trial with cypermethrin at Monks Wood. Percentage mortality in three age groups ofP. brassicae larvae Distance ( m )
0 2.5 5 10 20 30 Controls Mean larvae per station
Age (days) 3
5
7
100 31.3 41.7 16.0 6.3 3.4 5.6 28
100 14.3 17.1 18.2 8.9 2.0 4.8 43
100 40.0 17.4 8.3 5.4 4.2 0 25
Date/time, 6 July 1990/14: 45; Wind speed, 3 m s- ~.
meton-S-methyl or dimethoate, was included in the spray at Upper Caldecote and Swavesey. Previous bioassay experiments have shown that these compounds have a very low toxicity to P. brassicae compared with cypermethrin and triazophos (Williams et al., 1987; Sinha et al., 1990). The width of sprayed area upwind of the targets depended on the combination of boom width and number of passes. Drift from a single pass was assessed at Haverhill and Swavesey where the tractor started working from the far end of a large field after spraying the headland. At Upper Caldecote the tractor made three or four passes before the targets were taken up. At Chatteris the targets were exposed to drift from seven passes covering the whole field. However, there was some doubt as to whether the sprayer was giving the expected output at this last site because of calibration problems and the amount of spray left in the tank. The downwind terrain varied considerably from site to site, and depended on wind direction on the day. Most crops adjoin other crops, with or without an intervening grass bank, hedge or ditch, but it just happened that the downwind edge on five of the six occasions bordered a road or track. The time of spraying varied between 09: 40 and 18: 10, modal wind speed from 2.5 to 5.0 m s- 1 and temperature from 16 to 23 ° C (Table 4 ). There was a light drizzle during spraying on the first occasion at Upper Caldecote so no water-sensitive papers were used there.
Drift effects The percentage mortality data for P. brassicae larvae (Table 5) are arranged in seven distance classes to aid cross comparisons among sites, the actual spacing of targets being largely determined by physical and vegetation features. As mentioned previously, the results for the first Haverhill trial were atypical as the larvae were too young to survive exposure well, confirmed by
Haverhill 2
peas peas 28.6.90 10.7.90 15:15 14:30 cypermethrin cypermethrin 25 25 24 24 55 55 4.0 3.5 23 21 Albuz 110V 2.75
2plus dimethoate at 350 g a.i. h a - 1.
~Plus demeton-S-methyl at 325 g a.i. h a - ~.
Crop Date Time Insecticide Rate (g a.i. ha - I ) Sprayed width ( m ) Crop height ( c m ) Wind speed (m s - ~) Temperature ( ° C) Nozzles (BCPC code) Spray pressure ( b a r )
Haverhill 1 peas 16.7.90 15:45 cypermethrin 25 100 50 5.0 20 Chafer Red 2.5
Chatteris
Six field bioassays of insecticide spray drift: site, spraying and meteorological details
TABLE 4
U. Caldecote 2
brussels sprouts brussels sprouts 29.8.89 27.7.90 10:00 09:40 cypermethrin ~ cypermethrin ~ 25 25 25.5 34 50 40 2.5 2.5 16 22 Albuz A P G 110B 3.6
U. Caldecote 1
peas 11.6.90 18:10 triazophos 2 336 18 50 2.5 16 Lurmark 04F 110 3.5
Swavesey
15 25
1 7.5 .
21 3 (22.5) 2
15 25
3Underestimate, at least 16 m as per data, see t e x t .
2Larvae only 1-day-old.
69.6 43.3 41.9
79.1 .
100
1 7.5 .
39 3 14.0 21.6 24.0
47.9 5.6 1.3
65.3 .
Mort. 80.6
Dist. %
Dist. %
Mort.
Haverhill 2
Haverhill 1
q m m e d i a t e l y b e h i n d hedge line.
>12-16 > 16-25 Control Mean larvae/station No. o f transects LDso LD2o LDIo
0-1 > 1-2 >2-4 >4-8 >8-12
Distance class
15 25
1 5
15.8 7.4 9.0
25.5
34.6
Mort.
31 3 < 1 10.4 21.6
Dist. %
Chatteris
16 181
1 2 4 8
14.3 1.1 0
95.5 97.6 66.6 14.0
Mort.
41 2 5.1 + 0 . 7 7.3 + 1.2 8.6 4- 1.73
Dist. %
U. C a l d e c o t e 1
21.9
0.8 6.1 11.4
33 3 3.5 13.9 18.6
Dist. %
4.0 6.2
33.7 26.8
67.8
Mort.
U. C a l d e c o t e 2
5.4
68.9 23.5 1.7 10.3
100 100
Mort.
29 4 8.2+0.3 10.6 + 0.4 12.0+0.6
0 2 7 10 11.51 13.4 -
Dist. %
Swavesey
Six field bioassays o f insecticide s p r a y drift: m e a n p e r c e n t a g e m o r t a l i t y o f P . brassicae at v a r i o u s d i s t a n c e s d o w n w i n d , a n d d i s t a n c e s at w h i c h 50%, 20% a n d 10% mortality w o u l d be expected to o c c u r ( w i t h s t a n d a r d errors f r o m logistic m o d e l s )
TABLE 5
t"
INSECTICIDE SPRAY DRIFT
101
the heavy mortality in the controls. The controls at Chatteris also showed somewhat higher mortality than usual, which was probably the result of strong winds and unusually long exposure when the whole field was sprayed. These six cases are depicted in semi-diagrammatic form (Fig. 1 ) to show the mean mortality curves superimposed on the ground features. (As mentioned above, the targets were placed on the ground in all cases except in the field of beans. ) At Haverhill, Chatteris and Upper Caldecote 2 the mean mortality profiles were approximately straight lines over the range of distances considered (Fig. 1a,b,d). The mortality levels at Chatteris were lower at 1 m and 5 m than might be expected from the width of sprayed area and rather high wind speed; they were more comparable with those at 6 m and 11 m at Upper Caldecote 2. This might be attributed to wind currents over the raised road between the two targets but deposition measurements on water-sensitive papers do not support this theory (see below). Estimates of distances at which 50%, 20% and 10% of larvae would be expected to die were read directly from the graphs for these three occasions (Table 5 ). The curve for Upper Caldecote 1 was sigmoid (Fig. 1c), as in the earlier experiments with diflubenzuron at intermediate wind speeds (Davis et al., 1991 a). The tail of the mortality curve was considered to be truncated by the shelter provided by a hedge as larvae placed immediately behind it suffered only 1% mortality compared with those just in front with 14%. In fitting a logistic model to the data, the values obtained behind the hedge were therefore excluded. Estimates of LDso, LD2o and LDlo distances varied by 1-3 m between the two transects but the lowest standard errors were obtained by combining both sets of data. (The use of an extra pair of 'data values', by assuming zero mortality at 64 m, made virtually no difference to the estimated parameters.) The fitted relationship between percentage mortality (P) and distance (d) was P = 107.73/[ 1 + e x p ( 0 . 6 1 8 d - 3 . 0 3 4 ) ] Swavesey was the only site where triazophos was used (with dimethoate). The mortality curve here was even more convex over the first 10 m and a hedge again appeared to have some effect (Fig. l e). Two of the four lines of targets were placed opposite small gaps. Tall grass in these gaps sheltered targets placed just behind the hedge line but those placed 2 m further away on bare ground in front of a wheat crop had a higher average mortality (14%) than those that were sheltered by the hedge (7%), although the replication was insufficient to show significant difference. For fitting the logistic model here, the values immediately behind the hedge line were again excluded but the furthest values were retained. In Table 5 the results are based on the combined data from all four transects and are obtained from the fitted relationship P = 101.95/[ 1 + e x p ( 0 . 5 8 2 d - 4 . 7 5 9 ) ]
102
B.N.K. DAVIS ET AL
100' 60 30'
10. 8, 3'
100
i
80
1. 0.6"
60,
0.3.
=o E
40,
0.10.06"
20
0.03"
O:
0.01 1
7,5
15
25
1
Control
7
15
25
Distance (m)
DIstance downwind (m)
100
30 50 10 6
40 ¸
>,
=o E
30, 1 20.
0.6 0.3 CoTntrol
10 ¸ ,,..d road
sugar beet
I
I
0.1 0.06'
1
5
15
25
1'0
Downwind distance (m)
100-
c
80-
00-
40-
20traok 0 0
1
2
4
8 DIManca downwind (m)
1"5
Oistance (m)
16
Control
2'0
2"5
103
INSECTICIDESPRAY DRIFT 100
lO. 6- \~ h:
3.\~
100"
1.
80"
0.6"
=
0.3"
60"
0.1'
\"
0.06.
\
40.
0.03
20,
• \
,brulmels i sprouts
0
I~.
,'-----~,~ 0 0.8
~
~ ~, ~
8.1
~
~ ~,~
11.4
0.01
,~ll
Control
21.9
5
10
15
20
Distance (m)
Distance d o w n w i n d (m)
100'
10 g.
100"
e
-"
-
g 0.6.
~
4o. 0.06
hedge ,o
°°
pea crop ~'~|'~ ,~
(
farm track )
0 S,(,"%,;. . . . . . . . . . . . . . 0
2
)'~k
. ...............,.,,3 I~,/ 7
Distance downwind (m)
10
11.5
0.03
,
J;ll)l//
EILLLL__.r
13.4
Control
0.01
, ~ 0 2
, 7
, 10
, 15
Distance (m]
Fig. 1. Mortality profiles for P. brassicae larvae exposed to insecticide drift (means + standard errors), and equivalent drift deposition measurements on water-sensitive papers (separate transects for b, d, e). Details as in Tables 4 and 5. (a) Cypermethrin at Haverhill, 28 June 1990 ( O ), I 0 July 1990 ( • ) ; (b) Cypermethrin at Chatteris, 16 July 1990; (c) Cypermethrin at Upper Caldecote 1, 29 August 1989; (d) Cypermethrin at Upper Caldecote 2, 27 July 1990; (e) Triazophos at Swavesey, 11 June 1990: open circles beyond the hedge were from targets opposite gaps.
Figure 1 also shows the drift deposition results obtained from the watersensitive papers placed adjacent to the larvae. These deposition curves, plotted on log-normal axes, show rapid and almost linear decline up to about 10
104
B.N.K. DAVIS ET AL.
m at Swavesey and Upper Caldecote 2, but a more exponential fall-off at Haverhill and Chatteris. The similarity between the two spraying events at Haverhill is noteworthy. The results for Chatteris show greater drift at 15 m and 25 m downwind, as would be expected from the higher wind speed, although this was not reflected in higher mortalities of larvae. DISCUSSION
There is little direct evidence of insecticide drift damage to terrestrial invertebrates because it rarely causes economic damage and has not been looked for. Most reports involve aerial spraying. In Canada, Plowright and Rodd (1980) used caged bumblebees (Bombus sp. ), solitary bees and vespid wasps to demonstrate mortality from drift during forest spraying with fenitrothion. Aerial spraying of ground crops has likewise been associated with honeybee poisoning incidents in Britain (Stevenson et al., 1978 ) and with killing beneficial predatory insects in orchards in Israel (Harpaz, 1979). These results were attributed to drift but were not based on actual bioassays. Several studies have shown the effects of insecticide sprays on non-target terrestrial invertebrates such as Coleoptera, Diptera, Hymenoptera and Araneae within crops after ground-based applications (Vickerman and Sunderland, 1977; Shires, 1980; Cole and Wilkinson, 1984; Hill, 1985; Basedow et al., 1985 ). It is therefore to be expected that such groups would also be affected by drift into adjacent habitats. The use of butterfly larvae for bioassay studies of insecticide drift, and of P. brassicae in particular, was argued by Davis et al. (199 l b), but similar studies are needed for representatives of other groups (of. Hassan et al., 1983 ). Studies of spray drift carried out under commercial farming conditions are subject to many uncontrollable factors. Uncertainty in timing can prove a serious impediment to bioassay studies when these depend on using animals or plants at a critical, sensitive stage. In this respect, the standardized procedure used here with 2-day-old P. brassicae larvae requires the spraying date to be known, and fixed, 5-7 days in advance, whereas farmers and contractors must normally respond flexibly to changes in weather, pest status and problems with equipment. The production of large numbers of larvae to order depends on a reliable external source of eggs or good quality control in an inhouse breeding programme. Given these constraints, and the possibility that a bioassay trial may have to be aborted at the last minute, the studies reported here nevertheless demonstrate that the method can be used under a range of field conditions. As Sinha et al. (1990) point out, the success of the method largely depends on the tenacious habit of the insect in remaining in a sedentary aggregation after an initial settling down period. Even if the larvae select the underside of the leaf they usually stay in place when the leaves are turned over just before the trial, unless they are exposed to hot sun.
INSECTICIDE SPRAY DR1VI"
105
In the studies described, the large variation between spray events (in site, operator and meteorological variables) prohibits precise conclusions about the effects of any particular variable. The height of vegetation downwind of the sprayed crop varied considerably: from zero, where there was a track or road, to 40-50 cm for the sugar-beet at Upper Caldecote 2 and Chatteris, and 120 cm for the field beans at Haverhill. The tall beans would have deflected the drift cloud upwards creating turbulence for a considerable distance (cf. Bradley, 1968). Such 'zero plane displacement' and associated change in 'roughness length' makes it unrealistic in many field situations to expect the drift deposition and mortality curves to fit a simple logistic or exponential model, as was assumed in the experiments done over a uniform grass sward (Davis et al., 1991a). Under such circumstances, a combination of modelling with graphical determinations (Fig. 1 ) allows reasonable estimates of buffer zones to be made under the prevailing spraying and topographical conditions. If these are set by distances at which larval mortality was 10% or less, then trials two, three and five give similar estimates of the order of 16-24 m. These results are comparable with the higher estimates of LDIo distance in Davis et al. ( 1991 a) for diflubenzuron which were in the range 16-28 m. For cypermethrin in trial four, the LDIo estimate of 8.6 m, using the logistic model, is clearly an underestimate, because the field observations showed mortality of 14% up to 16 m. For these data, an exponential model gave 14.9 _+6.0 m. For triazophos in trial six, a lower LDto value of 12 m was obtained. Present estimates of LD~o distances make no allowance for control mortality of about 5% and should therefore be treated with caution. However, they are based on short-term mortality and do not take account of delayed or sublethal effects. Longer-term effects need to be studied to obtain estimates of the dosage levels or distances at which there is no detectable effects on fecundity. To be realistic, however, studies should also take account of behavioural response to sprays. The hazard of pyrethroid sprays to honeybees is reduced because their repellency results in avoidance behaviour (Shires, 1983 ). Similarly, P. brassicae larvae can show irritancy responses to pyrethroids (Tan, 1982) which could cause them to drop off a plant. This might reduce exposure to toxic residues but increase exposure to ground predators which are the main cause of larval mortality in brassicas (Dempster, 1968 ). Water-sensitive papers are useful for assessing deposition of larger droplets (MacCollom et al., 1986 ), but should only be used in the absence of a more precise method such as a tracer dye or element (see Introduction). Only 30% of drops 50/~m in diameter will impact on glass microscope slides in a 2 m s- ~wind (Byass and Lake, 1977 ), and Davis et al. ( 1991 a ) showed that mortality of P. brassicae larvae occurred beyond the distance at which drift could be detected on water-sensitive papers. However, tracers cannot generally be added to commercial spray operations.
106
B.N.K.DAVISETAL.
In extrapolating these results more widely, the following factors need to be borne in mind. Cypermethrin and triazophos are among the most toxic compounds to the test insect, so less toxic ones would be safely accommodated within these buffer zones. Similarly, a hedge or shelterbelt within 20 m can provide significant protection to habitats beyond, a topic currently under investigation. However, cumulative drift from a whole field was only examined at Chatteris but could add significantly to drift in some circumstances (Nordby and Skuterud, 1974). Likewise, stronger winds and spraying of taller crops than those studied could increase the amount of drift. A 10% mortality level for P. brassicae larvae is, of course, quite arbitrary, and the estimated buffer zones are considerably greater than the 6 m 'conservation headland' recommended by the Game Conservancy Council for increasing the invertebrate food of partridge chicks. The Conservancy's figure is based pragmatically on the width of spray boom that can conveniently be shut off when spraying around the edge of a field, and will certainly reduce drift into field margins (Cuthbertson, 1988; Cuthbertson and Jepson, 1988 ). Bioassay measurements with P. brassicae indicate that mortality at 6 m would be between 24% and 39% for trials three to five and 70-75% for trials two and six. However, such comparisons are not strictly applicable because the first 6 m in the present studies were generally over a roadway or short vegetation. Less drift could be expected to occur over the same width of cereal crop. ACKNOWLEDGEMENTS
The authors are very grateful to M. Haylock, C. Childs, P. Maudlin and J. Burgess for access to their land and their cooperation. We would also like to thank ICI Agrochemicals and Hoechst UK Ltd. for supplying technical grade cypermethrin and triazophos, respectively. The work was funded by the Department of Environment (PECD 7/8/168) and Nature Conservancy Council (F73-17-01).
REFERENCES Basedow, T., Rzehak, H. and Vob, K., 1985. Studies on the effect of deltamethrin sprays on the number of epigeal predatory arthropods occurring in arable fields. Pestic. Sci., 16:325-331. Bradley, E.F., 1968. A micrometeorological study of velocity profiles and surface drag in the region modified by a change in surface roughness. Q. J. R. Meteorol. Soc., 94: 361-379. Byass, J.B. and Lake, J.R., 1977. Spray drift from a tractor powered field sprayer. Pest. Sci., 8: I 17-126. Cole, J.F.H. and Wilkinson, W., 1984. Selectivity of pirimicarb in cereal crops. In" Proceedings of the 1984 British Crop Protection Conference, Pests and Diseases, Croydon, l: 311-316. Currier, W.W., MacCollom, G.B. and Baumann, G.L., 1982. Drift residues of air-applied carbaryl in an orchard environment. J. Econ. Entomol., 75: 1062-1068.
INSECTICIDESPRAYDRIFT
107
Cuthbenson, P., 1988. The pattern and level of pesticide drift into conservation and fully sprayed arable crop headlands. Aspects Appl. Biol., 17: 273-275. Cuthbertson, P.S. and Jepson, P.C., 1988. Reducing pesticide drift into the hedgerow by the inclusion of an unsprayed field margin. In: Brighton Crop Protection Conference, Pests and Diseases, Thornton Heath, 1988, 2:747-751. Davis, B.N.K., Lakhani, K.H., Yates, T.J. and Frost, A.J., 1991a. Bioassays of insecticide spray drift: the effects of wind speed on the mortality of Pieris brassicae larvae (Lepidoptera) due to diflubenzuron. Agric. Ecosystems Environ., 36: 141-149. Davis, B.N.K., Lakhani, K.H. and Yates, T.J., 199 lb. The hazards of insecticides to butterflies of field margins. Agric. Ecosystems Environ., 36:151-161. Dempster, J.P., 1968. The control of Pieris rapae with DDT. I. The natural mortality of the young stages ofPieris. J. Appl. Ecol., 4: 485-500. Dobson, C.M., 1986. Insecticide drift from sprayers and the effect on beneficial arthropods in winter wheat. Ph.D. thesis, University of London. Dobson, C.M., Minski, M.J. and Matthews, G.A., 1983. Neutron activation analysis using dysprosium as a tracer to measure spray drift. Crop Prot., 2: 345-352. Gilbert, A.J. and Bell, G.J., 1988. Evaluation of the drift hazards arising from pesticide spray application. Aspects Appi. Biol., 17: 363-376. Grover, R., Kerr, L.A., Maybank, J. and Yoshida, K., 1978. Field measurements of droplet drift from ground sprayers: I. Sampling, analytical and data integration techniques. Can. J. Plant Sci., 58:611-622. Hall, F.R., Cooper, J.A. and Ferree, D.C., 1991. Orchard geometry and pesticide placement. In: A. Lavers, P.A. Herrington and E.S.E. Southcombe (Editors), Air-Assisted Spraying in Crop Protection. British Crop Protection Council Monograph No. 46, Farnham, pp. 171-176. Harpaz, I., 1979. Regulatory measures adopted in Israel regarding the screening and field application of pesticides within the context of integrated control. Int. Pest Control, 21: 28-30. Hassan, S.A., Bigler, F., Bogenschutz et al., 1983. Results of the second joint pesticide testing programme by the IOBC/WPRS working group, 'Pesticides and beneficial arthropods'. Z. Angew. Entomol., 95:151-158. Herrington, P.J., Mapother, H.R. and Stringer, A., 1981. Spray retention and distribution on apple trees. Pestic. Sci., 12:515-520. Hill, I.R., 1985. Effects on non-target organisms in terrestrial and aquatic environments. In: J.P. Leahey (Editor), The Pyrethroid Insecticides, Taylor & Francis, London, pp. 151-262. KiJmmel, K., G6hlick, H. and Westphal, O., 1991. Development of practice-oriented control test methods for orchard spray machines by means of a vertical test stand. In: A. Lavers, P.A. Herrington and E.S.E. Southcombe (Editors), Air-Assisted Spraying in Crop Protection. British Crop Protection Council Monograph No. 46, Farnham, pp. 27-33. Lloyd, G.A. and Bell, G.J., 1982. Relative contamination of humans and the environment by spray drift from use of a tractor mounted drift sprayer (the Ulvamast) and conventional hydraulic equipment. In: Proceedings of the 1982 British Crop Protection Conference, Weeds, Croydon, 3: 1045-1054. MacCollom, G.B., Currier, W.W. and Baumann, G.L., 1986. Drift comparisons between aerial and ground orchard application. J. Econ. Ent., 79: 459-464. Maybank, J., Yoshida, K. and Grover, R., 1978. Spray drift from agricultural pesticide applications. J. Air Pollut. Control. Assoc., 28: 1009-1014. Nordby, A. and Skuterud, R., 1974. The effects of boom height, working pressure and wind speed on spray drift. Weed Res., 14: 385-395. Parkin, C.S. and Merritt, C.R., 1988. The measurement and prediction of spray drift. Aspects Appl. Biol., 17: 351-361. Payne, N.J., Feng, J.C. and Reynolds, P.E., 1990. Off-target deposits and buffer zones required around water for aerial glyphosate applications. Pestic. Sci., 30:183-198.
108
B.N.K.DAVISETAL.
Plowright, R.C. and Rodd, F.H., 1980. The effect of aerial insecticde spraying on hymenopterous pollinators in New Brunswick. Can. Ent., 112, 259-269. Riley, C.M. and Wiesner, C.J., 1989. Off-target deposition and drift of aerially applied agricultural sprays. Pestic. Sci., 26:159-166. Ross, G.J.S., 1980. Maximum Likelihood Programme. Rothamsted Experimental Station, Harpenden. Shires, S.W., 1980. Soil surface predators in arable land - - the effect of farming practices. Span, 23: 62-64. Shires, S.W., 1983. Pesticides and honeybees: case studies with ripcord and fastac. Span, 26: 118-120. Sinha, S.N., Lakhani, K.H. and Davis, B.N.K., 1990. Studies on the toxicity of insecticidal drift to the first instar larvae of the Large White butterfly Pieris brassicae (Lepidoptera: Pieridae). Ann. Appl. Biol., 116: 27-41. Stevenson, J.H., Needham, P.H. and Walker, J., 1978. Poisoning of honeybees by pesticides: investigations of the changing pattern in Britain over 20 years. Rep. Rothamsted Exp. St. 1977, 2: 55-72. Tan, K-H., 1982. Irritancy response to temperature after sublethal action ofpyrethroids against cabbage white caterpillars Pieris brassicae. Ent. Exp. Appl., 32:15 l-154. Vickerman, G.P. and Sunderland, K.D., 1977. Some effects of dimethoate on arthropods in winter wheat. J. Appl. Ecol., 14: 767-777. Williams, C.T., Davis, B.N.K., MalTs, R.H. and Osborn, D., 1987. Impact of Pesticide Drift. NERC contract report to NCC, Institute of Terrestrial Ecology, Huntingdon.
93
Elsevier Science Publishers B.V., Amsterdam
Insecticide drift from ground-based, hydraulic spraying of peas and brussels sprouts: bioassays for determining buffer zones B.N.K. Davis, K.H. Lakhani, T.J. Yates, A.J. Frost and R.A. Plant NERC, Institute of Terrestrial Ecology, Monks Wood,Abbots Ripton, Huntingdon, PE17 2LS, UK (Accepted 15 June 1992)
ABSTRACT Davis, B.N.K., Lakhani, K.H., Yates, T.J., Frost, A.J. and Plant, R.A., 1993. Insecticide drift from ground-based, hydraulic spraying of peas and brussels sprouts: bioassays for determining buffer zones. Agric. EcosystemsEnviron., 43: 93-108. Studies were made on insecticide spray drift from conventional hydraulic sprayers under typical agricultural conditions. A bioassay technique with 2-day-old Pieris brassicaelarvae was used to assess drift of cypermethrin and triazophos from pea and brassica crops on six occasions. Earlier work had shown that this species was a suitable model for other butterflies. Replicated series of target larvae were placed downwind of the sprayed crops, and mortality data used to obtain profiles of spray drift related to physical features such as roadways, adjacent crops and hedges. Estimates of the distances at which 50%, 20% and 10% mortality occurred were obtained graphically or by fitting logistic models to the data. Buffer zones that limited mortality to 10% or less were between 16 and 24 m for cypermethrin and ! 2 m for triazophos. These estimates made no allowance for control mortality of about 5%, however, they were based conservatively on short-term effects. Hedges also probably limited the full extent of drift in two instances. The wider application of these results is discussed and comparisons are made with the use of unsprayed crop headlands.
INTRODUCTION
There have been many studies on spray drift using tracer dyes collected on fiat plates, tapes, nylon lines or other receptors; deposits are then rinsed off and measured by colorimetry or fluorimetry (e.g. Nordby and Skuterud, 1974; Byass and Lake, 1977; Grover et al., 1978; Maybank et al., 1978; Lloyd and Bell, 1982; Gilbert and Bell, 1988; Parkin and Merritt, 1988). Other techniques include atomic absorption spectrometry, which has been used to measure zinc and manganese in fungicide sprays (Kiimmel et al., 1991 ); colorimetric determination of acid-extracted copper in copper fungicides Correspondence to."B.N.K. Davis, NERC, Institute of Terrestrial Ecology, Monks Wood, Abbots Ripton, Huntingdon, PE 17 2LS, UK.
94
B.N.K. DAVIS ET AL.
(Herrington et al., 1981 ); measurements of conductivity for sprays containing calcium nitrate (Hall et al., 1991 ); and neutron activation of dysprosium used as a tracer (Dobson et al., 1983). It is also possible to measure drift deposits of some organic pesticides using gas liquid chromatography or high pressure liquid chromatography e.g. carbaryl, deltamethrin and glyphosate (Currier et al., 1982; Riley and Wiesner, 1989; Payne et al., 1990). All these methods show the relationship of drift deposits with distance, and the effects of particular variables such as nozzel type, boom height, wind speed, etc. However, they do not give direct information on the biological impact of drift. In the case of insecticides, such effects are not readily observed but two approaches are possible. Short range and generally short-term effects can be recorded in natural populations or communities of non-target organisms in habitats adjacent to sprayed crops (Dobson, 1986; Williams et al., 1987). This can be useful for distinguishing relatively sensitive and insensitive groups for different spray materials. However, bioassay experiments are more appropriate for measuring the width of buffer zones to protect beneficial insects and wildlife. Here, organisms are exposed in a controlled manner during pest control spraying. Davis et al. ( 1991 a) described a bioassay technique for evaluating the effects of wind speed on insecticide spray drift from a tractor-powered hydraulic sprayer. In order to examine the influence of this one variable other factors were maintained as uniformly as possible. Thus, site and operating conditions were controlled by using a field of short grass, by orienting the sprayed swathes at right angles to the prevailing wind on each occasion, by setting out targets (also in short grass ) in line with the wind, and by using the same compound at the same rate for each trial. This paper describes bioassay measurements of drift made in agricultural crops during spraying by farmers. The objectives were: ( 1 ) to see whether the techniques previously developed for controlled experiments could be used under typical agricultural conditions; (2) to relate drift deposition and its effects to physical features downwind of sprayed crops; (3) to compare results with those obtained under standardized experimental conditions at Monks Wood in 1988 and 1989; (4) to estimate the size of buffer zones that would protect species outside the sprayed crop. MATERIALS AND METHODS
Two insecticides were selected for these tests, cypermethrin and triazophos which are used against a wide spectrum of pests including Lepidoptera. The ground crops were peas and brussels sprouts which are sprayed for control of Pea-moth (Cydia nigricana (Fab.)) and Cabbage white (Pieris spp. ) caterpillars. Topical dosing experiments on Pieris brassicae L. were first carried
INSECTICIDE SPRAY DRIFT
95
out to obtain LDso values for cypermethrin and triazophos, and to compare these with the range of toxicity values already obtained for eight other compounds (Sinha et al., 1990). In the case of triazophos, 2-, 3- and 4-day-old larvae were compared simultaneously to determine the effect of age on sensitivity. An experimental drift trial with cypermethrin was also carried out at Monks Wood, using larvae of three different ages in case older larvae had to be used for field bioassays because of delayed spraying. Four farms were selected in Cambridgeshire and Bedfordshire. Six to seven days notice of spraying was needed to have 2-day-old P. brassicaelarvae ready on the day of spraying, but on the first occasion at Haverhill insufficient notice was given and the larvae were only 1-day-old. Bioassay targets were prepared essentially in the way described by Davis et al. ( 1991 a). To optimize the initial feeding of larvae, egg batches were placed directly on Horse-radish plants (Armoracia rusticana), and kept in shaded, humid conditions at about 20°C. The day after hatching, the larvae were transferred to freshly cut leaves selected and trimmed to be as similar as possible. These leaves were stuck in flower arrangers' foam ('Oasis') in plastic pots flooded with water. On the following day they were transported to the test site (30-80 km). The targets were set out in two to four replicate lines about 10 m apart, perpendicular to the downwind field edge at varying measured distances up to 25 m. Wind speed and direction were recorded on a portable anemograph (Vector R500, Vector Instruments, Rhyl). When the direction was clearly oblique to the target lines at an angle 0, the measured distance d was adjusted to the effective distance d sec 0. The pots were placed on the ground at field edges, on road verges and in low crops, but at Haverhill where there was a tall crop of field beans on the downwind side the pots were suspended from collapsible wooden tripods so that the insects were just above crop height. After spraying, the larvae were placed in Petri dishes on sections of sprayed leaf, transported back and kept in an incubator at 24°C. Mortality was assessed over 3 days. A water-sensitive paper was pinned out next to each target leaf. Although 'inefficient' as collectors of fine droplets, they provided a visual indication of drift deposition which was measured on an I2S image processing system as described by Sinha et al. (1990). An attempt was made at Haverhill to measure the cypermethrin drift directly by using 'efficient' aluminium mesh cylinders ('Isopon', 10 cm circumference×6 cm long) suspended next to the target leaves. These were washed in 10 ml hexane which was injected onto a Varian model 3400 gas liquid chromatograph using a capillary column and electron capture detector. However, the cypermethrin could not be detected even at 1 m downwind owing to the very low application rate (25 g a.i. ha -1 is equivalent to 0.25 #g cm-2).
96
B.N.K. DAVISET AL.
Statistical analyses The determinations of LDso from topical dosing with cypermethrin and triazophos were based on probit analyses of the dose response relationships using the Maximum Likelihood Programme (Ross, 1980 ). The estimates from the separate replicate experiments were pooled to obtain an overall weighted mean estimate as described in Davis et al. ( 199 lb). For field data, estimates of mortality, with standard errors, were required for each distance from the replicate transects. These were obtained by pooling the numbers of larvae exposed at each distance and noting the total number of larvae dying. Thus, using the binomial model, the mortality rate p was estimated by the observed proportion dying
#=ErdEn, where r; is number of deaths, n; is number exposed, at a given distance on the ith transect and S.E. (#) =,4~/~( 1-,0)/En,]. The distances at which larvae would be expected to suffer 50%, 20% and 10% mortality were estimated at Upper Caldecote 1 and Swavesey by fitting the generalized logistic model to the data as described by Davis et al. ( 1991a),
P=y+~/[ 1+exp(a-fld) ] where P is percentage dying at distance d. RESULTS
Laboratory toxicity tests Cypermethrin was found to be the most toxic of the ten compounds so far tested against P. brassicae, being nearly 100 times as toxic as endosulfan (used for standardized comparisons in Sinha et al., 1990), 6.6 times as toxic as triazophos and 3.8 times as toxic as diflubenzuron used in earlier drift trials (Tables 1 and 2). The dose response curve for cypermethrin was found to be very TABLE1 Toxicity of cypermethrin and triazophos in topical applications to P. brassicae larvae. LDso estimates from probit analysis_+ standard errors as a percentage of the estimates Insecticide
Larval age (days)
Cypermethrin Triazophos
2 2 3 4
Concentration_+ SE (%) 6.48X 10 -5 6.5 4.26X 10 -4 7.7
10.55× l0 -4 5.3 17.86× l0 -4 6.0
/lg per insect
/lg g - t
0.00016 0.0011 0.0026 0.0045
0.231 1.521 3.103 3.283
97
INSECTICIDE SPRAY DRIFT TABLE 2
Relative field hazard of four insecticides to 2-day-old P. brassicae larvae based on topical dose and application rate (active intredient) to field crops Insecticide
LDso (/tgg -1 )
Field rate ( g h a -I )
'Field hazard' = rate/LDso (million LDso units h a - ~)
Dimethoate Cypermethrin Diflubenzuron Triazophos
627 0.23 0.87 1.52
400 25 100 3361
0.6 108 115 221
IRate for peas, 1050 for carrots (full rate).
steep around the LDso value, i.e. a small change in dose concentration produced a large change in percentage mortality. The above comparisons of relative toxicity are therefore only valid for the LDso dose. In the case of triazophos, the LDso doses increased with 3- and 4-day-old larvae as expected. When the dose per unit weight is considered, the effective rate was doubled for 3-day-old larvae but showed little further increase for 4day-old larvae. Contact toxicity in the field, however, is determined as much by application rate (and formulation) as by intrinsic toxicity. Thus, Table 2 shows that the 'hazard index' for cypermethrin is comparable with that for diflubenzuron and about 40% of that for triazophos at the rate used on peas. All three compounds also act as stomach poisons so the actual field toxicity may often depend on feeding behaviour rather than on direct exposure to spray drift.
Drift trial at Monks Wood The pilot drift bioassay with cypermethrin at Monks Wood gave results that were comparable with those obtained for diflubenzuron, at a wind speed of 2.5 m s- 1 (Davis et at., 1991 a ), in that 20% mortality lay between 5 m and 10 m downwind for the youngest group of larvae (Table 3 ). However, it should be noted that those larvae were a day older than in the standard diflubenzuron trials, and that there was not a great difference in sensitivity among the three age groups. The lack of clear-cut differences in this drift trial may have been owing partly to the limited sample size (a single transect for each age group), but also to the greater consumption of pesticide residues, and therefore increased stomach poisoning mentioned above.
Crop spraying bioassays Site details Six spraying operations were studied, one in 1989 and five in 1990. Table 4 summarises site, spraying and meteorological variables. An aphicide, de-
98
B.N.IC DAVISET AL.
TABLE 3 Pilot drift bioassay trial with cypermethrin at Monks Wood. Percentage mortality in three age groups ofP. brassicae larvae Distance ( m )
0 2.5 5 10 20 30 Controls Mean larvae per station
Age (days) 3
5
7
100 31.3 41.7 16.0 6.3 3.4 5.6 28
100 14.3 17.1 18.2 8.9 2.0 4.8 43
100 40.0 17.4 8.3 5.4 4.2 0 25
Date/time, 6 July 1990/14: 45; Wind speed, 3 m s- ~.
meton-S-methyl or dimethoate, was included in the spray at Upper Caldecote and Swavesey. Previous bioassay experiments have shown that these compounds have a very low toxicity to P. brassicae compared with cypermethrin and triazophos (Williams et al., 1987; Sinha et al., 1990). The width of sprayed area upwind of the targets depended on the combination of boom width and number of passes. Drift from a single pass was assessed at Haverhill and Swavesey where the tractor started working from the far end of a large field after spraying the headland. At Upper Caldecote the tractor made three or four passes before the targets were taken up. At Chatteris the targets were exposed to drift from seven passes covering the whole field. However, there was some doubt as to whether the sprayer was giving the expected output at this last site because of calibration problems and the amount of spray left in the tank. The downwind terrain varied considerably from site to site, and depended on wind direction on the day. Most crops adjoin other crops, with or without an intervening grass bank, hedge or ditch, but it just happened that the downwind edge on five of the six occasions bordered a road or track. The time of spraying varied between 09: 40 and 18: 10, modal wind speed from 2.5 to 5.0 m s- 1 and temperature from 16 to 23 ° C (Table 4 ). There was a light drizzle during spraying on the first occasion at Upper Caldecote so no water-sensitive papers were used there.
Drift effects The percentage mortality data for P. brassicae larvae (Table 5) are arranged in seven distance classes to aid cross comparisons among sites, the actual spacing of targets being largely determined by physical and vegetation features. As mentioned previously, the results for the first Haverhill trial were atypical as the larvae were too young to survive exposure well, confirmed by
Haverhill 2
peas peas 28.6.90 10.7.90 15:15 14:30 cypermethrin cypermethrin 25 25 24 24 55 55 4.0 3.5 23 21 Albuz 110V 2.75
2plus dimethoate at 350 g a.i. h a - 1.
~Plus demeton-S-methyl at 325 g a.i. h a - ~.
Crop Date Time Insecticide Rate (g a.i. ha - I ) Sprayed width ( m ) Crop height ( c m ) Wind speed (m s - ~) Temperature ( ° C) Nozzles (BCPC code) Spray pressure ( b a r )
Haverhill 1 peas 16.7.90 15:45 cypermethrin 25 100 50 5.0 20 Chafer Red 2.5
Chatteris
Six field bioassays of insecticide spray drift: site, spraying and meteorological details
TABLE 4
U. Caldecote 2
brussels sprouts brussels sprouts 29.8.89 27.7.90 10:00 09:40 cypermethrin ~ cypermethrin ~ 25 25 25.5 34 50 40 2.5 2.5 16 22 Albuz A P G 110B 3.6
U. Caldecote 1
peas 11.6.90 18:10 triazophos 2 336 18 50 2.5 16 Lurmark 04F 110 3.5
Swavesey
15 25
1 7.5 .
21 3 (22.5) 2
15 25
3Underestimate, at least 16 m as per data, see t e x t .
2Larvae only 1-day-old.
69.6 43.3 41.9
79.1 .
100
1 7.5 .
39 3 14.0 21.6 24.0
47.9 5.6 1.3
65.3 .
Mort. 80.6
Dist. %
Dist. %
Mort.
Haverhill 2
Haverhill 1
q m m e d i a t e l y b e h i n d hedge line.
>12-16 > 16-25 Control Mean larvae/station No. o f transects LDso LD2o LDIo
0-1 > 1-2 >2-4 >4-8 >8-12
Distance class
15 25
1 5
15.8 7.4 9.0
25.5
34.6
Mort.
31 3 < 1 10.4 21.6
Dist. %
Chatteris
16 181
1 2 4 8
14.3 1.1 0
95.5 97.6 66.6 14.0
Mort.
41 2 5.1 + 0 . 7 7.3 + 1.2 8.6 4- 1.73
Dist. %
U. C a l d e c o t e 1
21.9
0.8 6.1 11.4
33 3 3.5 13.9 18.6
Dist. %
4.0 6.2
33.7 26.8
67.8
Mort.
U. C a l d e c o t e 2
5.4
68.9 23.5 1.7 10.3
100 100
Mort.
29 4 8.2+0.3 10.6 + 0.4 12.0+0.6
0 2 7 10 11.51 13.4 -
Dist. %
Swavesey
Six field bioassays o f insecticide s p r a y drift: m e a n p e r c e n t a g e m o r t a l i t y o f P . brassicae at v a r i o u s d i s t a n c e s d o w n w i n d , a n d d i s t a n c e s at w h i c h 50%, 20% a n d 10% mortality w o u l d be expected to o c c u r ( w i t h s t a n d a r d errors f r o m logistic m o d e l s )
TABLE 5
t"
INSECTICIDE SPRAY DRIFT
101
the heavy mortality in the controls. The controls at Chatteris also showed somewhat higher mortality than usual, which was probably the result of strong winds and unusually long exposure when the whole field was sprayed. These six cases are depicted in semi-diagrammatic form (Fig. 1 ) to show the mean mortality curves superimposed on the ground features. (As mentioned above, the targets were placed on the ground in all cases except in the field of beans. ) At Haverhill, Chatteris and Upper Caldecote 2 the mean mortality profiles were approximately straight lines over the range of distances considered (Fig. 1a,b,d). The mortality levels at Chatteris were lower at 1 m and 5 m than might be expected from the width of sprayed area and rather high wind speed; they were more comparable with those at 6 m and 11 m at Upper Caldecote 2. This might be attributed to wind currents over the raised road between the two targets but deposition measurements on water-sensitive papers do not support this theory (see below). Estimates of distances at which 50%, 20% and 10% of larvae would be expected to die were read directly from the graphs for these three occasions (Table 5 ). The curve for Upper Caldecote 1 was sigmoid (Fig. 1c), as in the earlier experiments with diflubenzuron at intermediate wind speeds (Davis et al., 1991 a). The tail of the mortality curve was considered to be truncated by the shelter provided by a hedge as larvae placed immediately behind it suffered only 1% mortality compared with those just in front with 14%. In fitting a logistic model to the data, the values obtained behind the hedge were therefore excluded. Estimates of LDso, LD2o and LDlo distances varied by 1-3 m between the two transects but the lowest standard errors were obtained by combining both sets of data. (The use of an extra pair of 'data values', by assuming zero mortality at 64 m, made virtually no difference to the estimated parameters.) The fitted relationship between percentage mortality (P) and distance (d) was P = 107.73/[ 1 + e x p ( 0 . 6 1 8 d - 3 . 0 3 4 ) ] Swavesey was the only site where triazophos was used (with dimethoate). The mortality curve here was even more convex over the first 10 m and a hedge again appeared to have some effect (Fig. l e). Two of the four lines of targets were placed opposite small gaps. Tall grass in these gaps sheltered targets placed just behind the hedge line but those placed 2 m further away on bare ground in front of a wheat crop had a higher average mortality (14%) than those that were sheltered by the hedge (7%), although the replication was insufficient to show significant difference. For fitting the logistic model here, the values immediately behind the hedge line were again excluded but the furthest values were retained. In Table 5 the results are based on the combined data from all four transects and are obtained from the fitted relationship P = 101.95/[ 1 + e x p ( 0 . 5 8 2 d - 4 . 7 5 9 ) ]
102
B.N.K. DAVIS ET AL
100' 60 30'
10. 8, 3'
100
i
80
1. 0.6"
60,
0.3.
=o E
40,
0.10.06"
20
0.03"
O:
0.01 1
7,5
15
25
1
Control
7
15
25
Distance (m)
DIstance downwind (m)
100
30 50 10 6
40 ¸
>,
=o E
30, 1 20.
0.6 0.3 CoTntrol
10 ¸ ,,..d road
sugar beet
I
I
0.1 0.06'
1
5
15
25
1'0
Downwind distance (m)
100-
c
80-
00-
40-
20traok 0 0
1
2
4
8 DIManca downwind (m)
1"5
Oistance (m)
16
Control
2'0
2"5
103
INSECTICIDESPRAY DRIFT 100
lO. 6- \~ h:
3.\~
100"
1.
80"
0.6"
=
0.3"
60"
0.1'
\"
0.06.
\
40.
0.03
20,
• \
,brulmels i sprouts
0
I~.
,'-----~,~ 0 0.8
~
~ ~, ~
8.1
~
~ ~,~
11.4
0.01
,~ll
Control
21.9
5
10
15
20
Distance (m)
Distance d o w n w i n d (m)
100'
10 g.
100"
e
-"
-
g 0.6.
~
4o. 0.06
hedge ,o
°°
pea crop ~'~|'~ ,~
(
farm track )
0 S,(,"%,;. . . . . . . . . . . . . . 0
2
)'~k
. ...............,.,,3 I~,/ 7
Distance downwind (m)
10
11.5
0.03
,
J;ll)l//
EILLLL__.r
13.4
Control
0.01
, ~ 0 2
, 7
, 10
, 15
Distance (m]
Fig. 1. Mortality profiles for P. brassicae larvae exposed to insecticide drift (means + standard errors), and equivalent drift deposition measurements on water-sensitive papers (separate transects for b, d, e). Details as in Tables 4 and 5. (a) Cypermethrin at Haverhill, 28 June 1990 ( O ), I 0 July 1990 ( • ) ; (b) Cypermethrin at Chatteris, 16 July 1990; (c) Cypermethrin at Upper Caldecote 1, 29 August 1989; (d) Cypermethrin at Upper Caldecote 2, 27 July 1990; (e) Triazophos at Swavesey, 11 June 1990: open circles beyond the hedge were from targets opposite gaps.
Figure 1 also shows the drift deposition results obtained from the watersensitive papers placed adjacent to the larvae. These deposition curves, plotted on log-normal axes, show rapid and almost linear decline up to about 10
104
B.N.K. DAVIS ET AL.
m at Swavesey and Upper Caldecote 2, but a more exponential fall-off at Haverhill and Chatteris. The similarity between the two spraying events at Haverhill is noteworthy. The results for Chatteris show greater drift at 15 m and 25 m downwind, as would be expected from the higher wind speed, although this was not reflected in higher mortalities of larvae. DISCUSSION
There is little direct evidence of insecticide drift damage to terrestrial invertebrates because it rarely causes economic damage and has not been looked for. Most reports involve aerial spraying. In Canada, Plowright and Rodd (1980) used caged bumblebees (Bombus sp. ), solitary bees and vespid wasps to demonstrate mortality from drift during forest spraying with fenitrothion. Aerial spraying of ground crops has likewise been associated with honeybee poisoning incidents in Britain (Stevenson et al., 1978 ) and with killing beneficial predatory insects in orchards in Israel (Harpaz, 1979). These results were attributed to drift but were not based on actual bioassays. Several studies have shown the effects of insecticide sprays on non-target terrestrial invertebrates such as Coleoptera, Diptera, Hymenoptera and Araneae within crops after ground-based applications (Vickerman and Sunderland, 1977; Shires, 1980; Cole and Wilkinson, 1984; Hill, 1985; Basedow et al., 1985 ). It is therefore to be expected that such groups would also be affected by drift into adjacent habitats. The use of butterfly larvae for bioassay studies of insecticide drift, and of P. brassicae in particular, was argued by Davis et al. (199 l b), but similar studies are needed for representatives of other groups (of. Hassan et al., 1983 ). Studies of spray drift carried out under commercial farming conditions are subject to many uncontrollable factors. Uncertainty in timing can prove a serious impediment to bioassay studies when these depend on using animals or plants at a critical, sensitive stage. In this respect, the standardized procedure used here with 2-day-old P. brassicae larvae requires the spraying date to be known, and fixed, 5-7 days in advance, whereas farmers and contractors must normally respond flexibly to changes in weather, pest status and problems with equipment. The production of large numbers of larvae to order depends on a reliable external source of eggs or good quality control in an inhouse breeding programme. Given these constraints, and the possibility that a bioassay trial may have to be aborted at the last minute, the studies reported here nevertheless demonstrate that the method can be used under a range of field conditions. As Sinha et al. (1990) point out, the success of the method largely depends on the tenacious habit of the insect in remaining in a sedentary aggregation after an initial settling down period. Even if the larvae select the underside of the leaf they usually stay in place when the leaves are turned over just before the trial, unless they are exposed to hot sun.
INSECTICIDE SPRAY DR1VI"
105
In the studies described, the large variation between spray events (in site, operator and meteorological variables) prohibits precise conclusions about the effects of any particular variable. The height of vegetation downwind of the sprayed crop varied considerably: from zero, where there was a track or road, to 40-50 cm for the sugar-beet at Upper Caldecote 2 and Chatteris, and 120 cm for the field beans at Haverhill. The tall beans would have deflected the drift cloud upwards creating turbulence for a considerable distance (cf. Bradley, 1968). Such 'zero plane displacement' and associated change in 'roughness length' makes it unrealistic in many field situations to expect the drift deposition and mortality curves to fit a simple logistic or exponential model, as was assumed in the experiments done over a uniform grass sward (Davis et al., 1991a). Under such circumstances, a combination of modelling with graphical determinations (Fig. 1 ) allows reasonable estimates of buffer zones to be made under the prevailing spraying and topographical conditions. If these are set by distances at which larval mortality was 10% or less, then trials two, three and five give similar estimates of the order of 16-24 m. These results are comparable with the higher estimates of LDIo distance in Davis et al. ( 1991 a) for diflubenzuron which were in the range 16-28 m. For cypermethrin in trial four, the LDIo estimate of 8.6 m, using the logistic model, is clearly an underestimate, because the field observations showed mortality of 14% up to 16 m. For these data, an exponential model gave 14.9 _+6.0 m. For triazophos in trial six, a lower LDto value of 12 m was obtained. Present estimates of LD~o distances make no allowance for control mortality of about 5% and should therefore be treated with caution. However, they are based on short-term mortality and do not take account of delayed or sublethal effects. Longer-term effects need to be studied to obtain estimates of the dosage levels or distances at which there is no detectable effects on fecundity. To be realistic, however, studies should also take account of behavioural response to sprays. The hazard of pyrethroid sprays to honeybees is reduced because their repellency results in avoidance behaviour (Shires, 1983 ). Similarly, P. brassicae larvae can show irritancy responses to pyrethroids (Tan, 1982) which could cause them to drop off a plant. This might reduce exposure to toxic residues but increase exposure to ground predators which are the main cause of larval mortality in brassicas (Dempster, 1968 ). Water-sensitive papers are useful for assessing deposition of larger droplets (MacCollom et al., 1986 ), but should only be used in the absence of a more precise method such as a tracer dye or element (see Introduction). Only 30% of drops 50/~m in diameter will impact on glass microscope slides in a 2 m s- ~wind (Byass and Lake, 1977 ), and Davis et al. ( 1991 a ) showed that mortality of P. brassicae larvae occurred beyond the distance at which drift could be detected on water-sensitive papers. However, tracers cannot generally be added to commercial spray operations.
106
B.N.K.DAVISETAL.
In extrapolating these results more widely, the following factors need to be borne in mind. Cypermethrin and triazophos are among the most toxic compounds to the test insect, so less toxic ones would be safely accommodated within these buffer zones. Similarly, a hedge or shelterbelt within 20 m can provide significant protection to habitats beyond, a topic currently under investigation. However, cumulative drift from a whole field was only examined at Chatteris but could add significantly to drift in some circumstances (Nordby and Skuterud, 1974). Likewise, stronger winds and spraying of taller crops than those studied could increase the amount of drift. A 10% mortality level for P. brassicae larvae is, of course, quite arbitrary, and the estimated buffer zones are considerably greater than the 6 m 'conservation headland' recommended by the Game Conservancy Council for increasing the invertebrate food of partridge chicks. The Conservancy's figure is based pragmatically on the width of spray boom that can conveniently be shut off when spraying around the edge of a field, and will certainly reduce drift into field margins (Cuthbertson, 1988; Cuthbertson and Jepson, 1988 ). Bioassay measurements with P. brassicae indicate that mortality at 6 m would be between 24% and 39% for trials three to five and 70-75% for trials two and six. However, such comparisons are not strictly applicable because the first 6 m in the present studies were generally over a roadway or short vegetation. Less drift could be expected to occur over the same width of cereal crop. ACKNOWLEDGEMENTS
The authors are very grateful to M. Haylock, C. Childs, P. Maudlin and J. Burgess for access to their land and their cooperation. We would also like to thank ICI Agrochemicals and Hoechst UK Ltd. for supplying technical grade cypermethrin and triazophos, respectively. The work was funded by the Department of Environment (PECD 7/8/168) and Nature Conservancy Council (F73-17-01).
REFERENCES Basedow, T., Rzehak, H. and Vob, K., 1985. Studies on the effect of deltamethrin sprays on the number of epigeal predatory arthropods occurring in arable fields. Pestic. Sci., 16:325-331. Bradley, E.F., 1968. A micrometeorological study of velocity profiles and surface drag in the region modified by a change in surface roughness. Q. J. R. Meteorol. Soc., 94: 361-379. Byass, J.B. and Lake, J.R., 1977. Spray drift from a tractor powered field sprayer. Pest. Sci., 8: I 17-126. Cole, J.F.H. and Wilkinson, W., 1984. Selectivity of pirimicarb in cereal crops. In" Proceedings of the 1984 British Crop Protection Conference, Pests and Diseases, Croydon, l: 311-316. Currier, W.W., MacCollom, G.B. and Baumann, G.L., 1982. Drift residues of air-applied carbaryl in an orchard environment. J. Econ. Entomol., 75: 1062-1068.
INSECTICIDESPRAYDRIFT
107
Cuthbenson, P., 1988. The pattern and level of pesticide drift into conservation and fully sprayed arable crop headlands. Aspects Appl. Biol., 17: 273-275. Cuthbertson, P.S. and Jepson, P.C., 1988. Reducing pesticide drift into the hedgerow by the inclusion of an unsprayed field margin. In: Brighton Crop Protection Conference, Pests and Diseases, Thornton Heath, 1988, 2:747-751. Davis, B.N.K., Lakhani, K.H., Yates, T.J. and Frost, A.J., 1991a. Bioassays of insecticide spray drift: the effects of wind speed on the mortality of Pieris brassicae larvae (Lepidoptera) due to diflubenzuron. Agric. Ecosystems Environ., 36: 141-149. Davis, B.N.K., Lakhani, K.H. and Yates, T.J., 199 lb. The hazards of insecticides to butterflies of field margins. Agric. Ecosystems Environ., 36:151-161. Dempster, J.P., 1968. The control of Pieris rapae with DDT. I. The natural mortality of the young stages ofPieris. J. Appl. Ecol., 4: 485-500. Dobson, C.M., 1986. Insecticide drift from sprayers and the effect on beneficial arthropods in winter wheat. Ph.D. thesis, University of London. Dobson, C.M., Minski, M.J. and Matthews, G.A., 1983. Neutron activation analysis using dysprosium as a tracer to measure spray drift. Crop Prot., 2: 345-352. Gilbert, A.J. and Bell, G.J., 1988. Evaluation of the drift hazards arising from pesticide spray application. Aspects Appi. Biol., 17: 363-376. Grover, R., Kerr, L.A., Maybank, J. and Yoshida, K., 1978. Field measurements of droplet drift from ground sprayers: I. Sampling, analytical and data integration techniques. Can. J. Plant Sci., 58:611-622. Hall, F.R., Cooper, J.A. and Ferree, D.C., 1991. Orchard geometry and pesticide placement. In: A. Lavers, P.A. Herrington and E.S.E. Southcombe (Editors), Air-Assisted Spraying in Crop Protection. British Crop Protection Council Monograph No. 46, Farnham, pp. 171-176. Harpaz, I., 1979. Regulatory measures adopted in Israel regarding the screening and field application of pesticides within the context of integrated control. Int. Pest Control, 21: 28-30. Hassan, S.A., Bigler, F., Bogenschutz et al., 1983. Results of the second joint pesticide testing programme by the IOBC/WPRS working group, 'Pesticides and beneficial arthropods'. Z. Angew. Entomol., 95:151-158. Herrington, P.J., Mapother, H.R. and Stringer, A., 1981. Spray retention and distribution on apple trees. Pestic. Sci., 12:515-520. Hill, I.R., 1985. Effects on non-target organisms in terrestrial and aquatic environments. In: J.P. Leahey (Editor), The Pyrethroid Insecticides, Taylor & Francis, London, pp. 151-262. KiJmmel, K., G6hlick, H. and Westphal, O., 1991. Development of practice-oriented control test methods for orchard spray machines by means of a vertical test stand. In: A. Lavers, P.A. Herrington and E.S.E. Southcombe (Editors), Air-Assisted Spraying in Crop Protection. British Crop Protection Council Monograph No. 46, Farnham, pp. 27-33. Lloyd, G.A. and Bell, G.J., 1982. Relative contamination of humans and the environment by spray drift from use of a tractor mounted drift sprayer (the Ulvamast) and conventional hydraulic equipment. In: Proceedings of the 1982 British Crop Protection Conference, Weeds, Croydon, 3: 1045-1054. MacCollom, G.B., Currier, W.W. and Baumann, G.L., 1986. Drift comparisons between aerial and ground orchard application. J. Econ. Ent., 79: 459-464. Maybank, J., Yoshida, K. and Grover, R., 1978. Spray drift from agricultural pesticide applications. J. Air Pollut. Control. Assoc., 28: 1009-1014. Nordby, A. and Skuterud, R., 1974. The effects of boom height, working pressure and wind speed on spray drift. Weed Res., 14: 385-395. Parkin, C.S. and Merritt, C.R., 1988. The measurement and prediction of spray drift. Aspects Appl. Biol., 17: 351-361. Payne, N.J., Feng, J.C. and Reynolds, P.E., 1990. Off-target deposits and buffer zones required around water for aerial glyphosate applications. Pestic. Sci., 30:183-198.
108
B.N.K.DAVISETAL.
Plowright, R.C. and Rodd, F.H., 1980. The effect of aerial insecticde spraying on hymenopterous pollinators in New Brunswick. Can. Ent., 112, 259-269. Riley, C.M. and Wiesner, C.J., 1989. Off-target deposition and drift of aerially applied agricultural sprays. Pestic. Sci., 26:159-166. Ross, G.J.S., 1980. Maximum Likelihood Programme. Rothamsted Experimental Station, Harpenden. Shires, S.W., 1980. Soil surface predators in arable land - - the effect of farming practices. Span, 23: 62-64. Shires, S.W., 1983. Pesticides and honeybees: case studies with ripcord and fastac. Span, 26: 118-120. Sinha, S.N., Lakhani, K.H. and Davis, B.N.K., 1990. Studies on the toxicity of insecticidal drift to the first instar larvae of the Large White butterfly Pieris brassicae (Lepidoptera: Pieridae). Ann. Appl. Biol., 116: 27-41. Stevenson, J.H., Needham, P.H. and Walker, J., 1978. Poisoning of honeybees by pesticides: investigations of the changing pattern in Britain over 20 years. Rep. Rothamsted Exp. St. 1977, 2: 55-72. Tan, K-H., 1982. Irritancy response to temperature after sublethal action ofpyrethroids against cabbage white caterpillars Pieris brassicae. Ent. Exp. Appl., 32:15 l-154. Vickerman, G.P. and Sunderland, K.D., 1977. Some effects of dimethoate on arthropods in winter wheat. J. Appl. Ecol., 14: 767-777. Williams, C.T., Davis, B.N.K., MalTs, R.H. and Osborn, D., 1987. Impact of Pesticide Drift. NERC contract report to NCC, Institute of Terrestrial Ecology, Huntingdon.