Control of spotted wing drosophila (Drosophila suzukii) in sweet cherry and raspberry using bait sprays

Spotted wing drosophila (Drosophila suzukii Matsumura) is a major pest of fruit crops with global significance. Effective control is reliant on uniformly spraying insecticides on all crop foliage. To encourage pest attraction and ingestion of insecticides, phagostimulant baits can be employed in ‘attract and kill’ strategies. In semi-field trials, we compared (1) full-field foliar sprays of two insecticides spinosad and cyantraniliprole, with (2) reduced [40%] and (3) low [4%] rates of the insecticides to control D. suzukii and reduce insecticide residues in fruit in sweet cherry and raspberry. The low rates of the insecticides were also combined with baits, (4) Combi-protec, a proprietary mixture of plant extracts, proteins and sugars and (5) molasses; treatment (6) was an untreated control. In both crops, when combined with baits, low rates of insecticides gave comparable control of D. suzukii in fruits compared with the full-rate sprays in most cases, and D. suzukii was significantly lower in these treatments compared with the untreated controls. Crop spray coverage was eight and thirty times higher in the full-rate applications compared with the bait and low-rate sprays, in raspberry and cherry, respectively. The reduction in applied insecticide was achieved by lower concentrations and volumes and a narrower band spray applied across the middle of the crop canopy. Spinosad and cyantraniliprole residues were detected in cherries taken from trees sprayed with full-rate applications but not in fruit from trees given the low rates of insecticides with bait. This study demonstrates that bait sprays can be effectively employed on crops with complex canopies for D. suzukii control.


Introduction
Spotted wing drosophila (Drosophila suzukii Matsumura) is a major pest of stone and soft fruit crops with global significance (Asplen et al. 2015;Tait et al. 2021), and in Europe, sweet cherry (Prunus avium L.) is reported as the most susceptible crop (Shawer et al. 2019). Repeated applications of a limited number of effective pesticides can give control (Van Timmeren and Isaacs 2013;Rosensteel and Sial 2017), and fruit growers have become reliant on their use (Lidley et al. 2020). However, there are risks associated with reliance on pesticides. These include resistance to active ingredients such as spinosad (Gress and Zalom 2019), pesticide residues in fruit and regulatory authorities reducing the maximum residue levels (MRLs) or permitted application Communicated by Kent Marshall. rates and even withdrawing approval of permitted products (Arena et al. 2018;Lainsbury 2020). In 2022, the permitted application rate of spinosad on fruit crops in the UK is being reduced from 120 to 48 g ha −1 (Anonymous 2020).
Targeting D. suzukii with sprays is difficult because juvenile stages are inside fruits, and the adults spend most time on the underside of leaves and in the middle of the crop canopy (Eaton 2014). The complex canopy structure of most fruit crops increases the difficulty in reaching these areas with sprays (Lewis and Hamby 2020). However, Mermer et al. (2021) found that certain insecticides, including spinosad and cyantraniliprole, were effective in increasing mortality of immature life stages of D. suzukii, and Wise et al. (2015) found that certain insecticides including spinetoram had curative effects on fruit post-infestation with D. suzukii larvae.
Baits have been added to sprays in 'attract and kill' strategies to improve the efficacy of existing and alternative insecticides for D. suzukii control by encouraging pest attraction to and ingestion of active ingredients (Cowles et al. 2015;Knight et al. 2016). Inclusion of sugar, corn steep liquor and/or brewer's yeast (Saccharomyces cerevisiae Meyen ex E.C. Hansen, 1883) into sprays had only a limited effect on the efficacy of insecticides against D. suzukii (Diepenbrock and Burrack 2015;Fanning et al. 2021). Several proprietary products containing natural phagostimulants have been shown to increase the efficacy of insecticides and reduce the effective dose of insecticide needed. These include Combiprotec, a mixture of plant extracts, proteins and sugars (Dederichs, 2015), a protein bait based on spent brewery waste (Cai et al. 2018), and the unspecified formulations HOOK SWD (Klick et al. 2019) and ACT TRA SWD (Babu et al. 2021). However, the cost saving in pesticide may be negated by the price of the bait (Babu et al. 2021;Noble et al, 2021). The by-product, molasses, was shown to be equally effective to Combi-protec for D. suzukii control in a semi-field-scale raspberry (Rubus idaeus L.) trial (Noble et al. 2021).
Drosophila suzukii is associated with and attracted to the yeast Hanseniaspora uvarum (Niehaus), which is also effective as a bait spray (Rehermann et al. 2021). However, a semi-field-scale strawberry (Fragaria × ananassa Duch.) trial showed that Combi-protec was more effective than H. uvarum, required less preparation and had a longer shelf life (Noble et al. 2021).
The aims of these studies were to determine the D. suzukii control efficacy, fruit insecticide residues and crop spray coverage of using low rates of insecticides with and without baits compared with full field and reduced permitted rates of the insecticides, in comparison with an untreated control. This was attempted by reducing the application rate of insecticides to sweet cherry by using a combination of lower insecticide concentration and water volume, modified spray equipment and targeted and faster application. Two phagostimulant baits were compared: Combi-protec, which is approved for use as a spray adjuvant in the UK (Lainsbury, 2020) and molasses. The D. suzukii control efficacy and crop spray coverage of treatments applied to a semi-field-scale cherry crop was determined and compared with similar treatments applied to a raspberry crop using unpublished larvae flotation data and published spray coverage data from a previous experiment (Noble et al. 2021).

Experimental sites and set-up
The cherry experiment was conducted at NIAB EMR in a 2008 planted orchard, with cv. Penny trees planted at 2-m spacing within the row and 3 m between rows orientated north-south. Trees were ~ 3.4-m high and 2.0-m wide, with no canopy below 0.5 m. In April 2021, after flowering had finished, two central tree rows, each of 76 trees, including guards at each end, were covered under a polythene tunnel. The tunnel had a central height of 3.7 m, side walls of 1.2-m height and width of 7.9 m. To exclude birds, the side walls and adjacent rows of trees were covered with 1-cm square netting. The tunnel was divided into 35 plots by polythene walls spaced at ~ 5.4-m intervals so that each plot contained four trees and was ~ 42.7 m 2 . The dividing walls and roof prevented migration of D. suzukii adults directly between adjacent plots but the netted side walls allowed movement of D. suzukii to and from the surrounding untreated orchard. Trees were irrigated through a drip irrigation system. Air temperature and relative humidity (RH) among the trees were recorded using sensors (type HMP 31 UT, Vaisala, Helsinki, Finland) and data loggers (Grant Instruments, Cambridge, UK). Average daily maximum and minimum temperatures in the cherry tree canopies were 21.6ºC and 12.9 ºC; average daily maximum and minimum RH was 97.5% and 74.5%.
Details of the 2020 raspberry experimental set-up are presented in Noble et al. (2021). Briefly, potted long cane raspberry plants were grown in 12 small cropping tunnels covered with fine mesh to prevent entry or exit of flies. Each tunnel was divided in two using a wall of fine mesh, and each of the 24 half-tunnel compartment plots contained ten potted raspberry plants, each with two canes.

Insecticide treatments
The entire cherry orchard was sprayed with acetamiprid at 75 g ha − 1 (product Gazelle, Certis Europe, Utrecht, The Netherlands) for control of black cherry aphid (Myzus cerasi Fabricius), on 26 May 2021, three weeks before the experimental treatments were applied. No further pesticide sprays were applied to the experiment or surrounding trees, other than the experimental treatments. Methods used for applying the treatments, including the spray nozzles used, are summarised in Table 1. Sprays were applied with the nozzle held 0.5 m from the crop foliage, from the alleyway between the two rows of trees. The tree canopy was sprayed from one side and designated as 'near', 'mid' and 'far' from the spray nozzle. These canopy sections corresponded to depths across the canopy from one side of the tree to the other of 0-0.6, 0.7-1.3 and 1.4-2.0 m. Trees were first sprayed on 17 June 2021 at white fruit stage (Shaw et al. 2019a), and then three further times, with alternating sprays of spinosad (product Tracer, Dow AgroSciences, Indianapolis, IN) and cyantraniliprole (product Exirel, Du Pont, Wilmington, DE) ( Table 2). Spinosad was applied at the full field rate permitted in 2021, a reduced permitted 2022 rate or at a low rate. Cyantraniliprole was applied at the full field rate, which was the same for 2021 and 2022, or at a low rate (Table 2). Insecticide applications at the full field or reduced rates (treatments 1 and 2) were air assisted with a high water volume, equivalent to 500 L ha − 1 . Low-rate applications were made with a water volume equivalent to 40 L ha − 1 , either without (treatment 3) or with one of two baits (Tables 1 and 2). The bait sprays were prepared by pre-mixing 5% v/v Combiprotec (treatment 4) or molasses made from raw cane sugar (Holland & Barrett, Nuneaton, UK) (treatment 5) in water at 30 °C. Control plots remained unsprayed (treatment 6). There were five replicate compartments of each treatment arranged in a randomised block design. Data of a seventh insecticide treatment, which was part of the same experiment blocking structure, were included in the statistical analysis, thereby improving estimation of the effect of blocks, but the results are not presented.
Target volumes of insecticide spray per tree to achieve the full-field-, reduced-and low-rate spray application rates were calculated from a nominal cropping density of 1,500 trees ha − 1 and were applied by calibrating the spray nozzle outputs and adjusting the speed of application to the four trees in each compartment. The spraying time in each individual plot was measured at 97 s for full-rate sprays and 10 s for low-rate sprays with and without baits. Volumes of spray applied were determined by measuring the initial and final volumes in the spray tank, before and after application to five replicate plots. All insecticide sprays were applied with an electric motorised knapsack sprayer (Birchmeier 14 REC ABC; Birchmeier AG, Stetten, Switzerland) at 3 bar. For full-and reduced-rate insecticide sprays (treatments 1 and 2), a hollow cone nozzle (Orange Albuz, ATR 80; Solcera, Evreux, France) with an additional motorised mist blower (Birchmeier AS1200; Birchmeier AG) was used. The fulland reduced-rate sprays were applied over the entire sprayed side of the trees, and the droplet spectra size was very fine to fine (British Crop Protection Council [BCPC], Alton, Hants., UK) ( Table 1). The target spray application volume was 333 ml tree − 1 . Low-rate insecticide sprays, with (treatments 4 and 5) or without baits (treatment 3), were applied with a hand lance supplied with the above Birchmeier 14 REC ABC sprayer, through an air-injector flat spray nozzle with a coarse droplet spectra (IDK 120-015 green nozzle, Lechler GmbH, Metzingen, Germany). This was used to spray a 1-m-width swath with the centre of spray aimed at Table 1 Summary of details used for applying and measuring the effects of full-field-, reduced-and low-rate insecticide and bait spray treatments in each of two applications a Cyantraniliprole was used at the same rate in both the 'full' and 'reduced' application rate treatments b Details of baits are shown in Table 2 Spray  Table 2 Weekly insecticide and bait spray treatments. Full recommended field rate at the time of writing, reduced (rdd) and low rates of spinosad (SP) and cyantraniliprole (CY) are shown in Table 1 a Only spinosad is at the reduced rate ( the middle of the 'near' side of trees. The target spray application volume was 26.6 ml tree − 1 , and the droplet spectra size was very fine to fine (Table 1). Details of the five treatments applied to four or five replicate plots in the raspberry experiment are presented in Noble et al. (2021). Briefly, two alternating sprays of spinosad and cyantraniliprole at the full field rates approved for raspberry (treatment 1) or low, 4% rates, without and with Combi-protec or molasses baits (treatments 2, 3 and 4) were compared with an untreated control (treatment 5).

Assessments of D. suzukii in fruit
In June 2021, one day after the first treatment spray, each cherry compartment was artificially infested with 20 female and ten male adult summer morph D. suzukii. Approximately, equal numbers of flies were deployed on either side of the tunnel compartments.
The ripest but not overripe cherries were used for each assessment. A sample of 24 cherries from the entire height and depth of all four trees in each plot was picked six days after the second sprays, and again six days after the two subsequent sprays. Cherries were incubated in ventilated clear plastic boxes as described in Noble et al. (2021), and the numbers of D. suzukii and other Drosophila species adults were recorded separately (Noble et al. 2021). Boxes were discarded after 17 days so there was no possibility of a second generation of flies. For a pre-treatment assessment, samples of 20 cherries from the entire height and depth of all four trees in each plot was picked three days before the first spray. Further, samples of 20 cherries were picked six days after each of the four sprays from the near, mid and far canopy sections of the entire height of all four trees in each compartment. The samples of 20 cherries were incubated for two days at 20 °C; larvae were then extracted using the sugar flotation method in Shaw et al. (2019b).
In June 2020 in raspberry, each compartment plot was artificially infested with ten female and ten male adult D. suzukii one day after the first treatment sprays, and with 20 females and ten males one day after the second and third sprays. Samples of 20 ripe fruits were taken from the top (above 1 m), middle (0.6-1 m) and bottom (below 0.6 m) of plants in each plot, six days after each of the four treatment sprays and one, three and five weeks thereafter, and flotation tested for larvae (Shaw et al. 2019b).

Insecticide residue testing of fruit
Samples of cherries for pesticide residue analysis were taken immediately after the second application of spinosad (with one intervening cyantraniliprole spray) and after the second cyantraniliprole application (with two previous spinosad sprays), i.e. after sprays 3 and 4 in Table 2. Pooled, 1-kg samples of cherries from all replicates were taken from the near, mid and far canopy sections of the trees from treatments 1-4 (Tables 2 and 3). Samples were analysed by QTS, Sittingbourne, Kent, UK, using liquid chromatography-mass spectrometry (LC-MS) (ISO 9001 certification). The detection limit for pesticide residues was 0.01 mg kg −1 fruit. Most of the fruit in the experiment ripened to a harvestable stage by weeks 2 and 3 of the experiment.

Spray deposition on foliage
The spray deposition of the full field rate (application volume 500 L ha − 1 ) and low rate with Combi-protec (application volumes 40 and 80 L ha − 1 applied at 10 and 20 s per compartment), without the added insecticides, was assessed separately from the main spray experiment but on the same cherry trees using a handheld imaging fluorometer (Chelsea Technologies Ltd, Molesey, UK) and fluorescence tracer dye (NIAB EMR) (Whitfield et al. 2019;Noble et al. 2021). Dye solutions (2% v/v in water) were prepared with and without Combi-protec at 5% v/v. Trees in the guard areas of the tunnel were sprayed using the appropriate spray equipment and settings for each treatment, seven days after the main spray experiment had been completed. For measurements, cherry tree canopies were divided into nine sections: near, mid and far distances from the sprayer nozzle (as previously described) × top (above 2.5 m), middle (1.5-2.5 m) Table 3 Residues of spinosad and cyantraniliprole in cherries taken from different spray treatments and tree canopy sections (near, mid and far from the sprayer nozzle); average of values in samples taken after the second spinosad and cyantraniliprole sprays (mg kg − 1 fruit). Mean values of the same insecticide with the same letters are not significantly different (p = 0.05), n = 2 a Only spinosad is at the reduced rate (Table 1)

Statistical analysis
Statistical analysis was carried out separately for each experiment in R 4.1.1 (Anonymous 2021). Analyses used generalised linear mixed models (GLMM), generalised linear models (GLM) or linear models (LM) as appropriate to the experimental design and data. GLMMs were fitted preferentially with the glmmTMB package (Brooks et al. 2017) or LME4 (Bates et al. 2015). For mixed models, likelihood ratio tests were performed to test for statistical differences between the fixed effects. Post hoc tests were carried out with the emmeans package (Lenth 2021), with p-values corrected for false discovery rate using the Tukey method. Adult emergence data were analysed using a GLMM with negative binomial family; to control for complete separation in the data, interaction effects between assessment and treatment were fitted as separate assessment points using a Poisson GLM implementing Firth's correction (Kosmidis and Firth 2020).
Cherry larvae flotation data were analysed with a negative binomial GLMM, with treatment, canopy section, assessment week and their interactions as the fixed effects. Raspberry larvae flotation data were fitted as a negative binomial GLMM in LME4 with the negative binomial family from the MASS package (Venables and Ripley 2002).
Spray deposition data were analysed with a LM using a logit transformation of the proportion of plant coverage.

Assessments of D. suzukii in fruit
There were linear relationships between larvae flotation (f) and adult D. suzukii emergence (e) measurements determined on equivalent numbers of cherries (f = 0.298e + 0.163, r = 0.666, p < 0.001) and raspberries (f = 0.192e + 0.549, r = 0.686, p < 0.001) taken from plots in assessment weeks 2, 3 and 4 in each of the experiments. This indicated that the larvae flotation method detected 29.8% and 19.2% of the potential adult D. suzukii emergence from the cherry and raspberry fruit samples.

D. suzukii control in cherry, 2021
In D. suzukii emergence tests on fruit samples, the first D. suzukii adults emerged 5-8 days after incubation, with the majority emerging by day 13. All the emerged adults were identified as D. suzukii, and similar proportions of females and males emerged in all boxes; overall, the ratio of D. suzukii females to males was 54:46.
There were significant effects of spray treatment (Χ 2 6 = 56.26; p < 0.001) and assessment week (Χ 2 2 = 66.68; p < 0.001) on the numbers of D. suzukii adults that emerged from cherry fruit. Approximately, three times more D. suzukii adults emerged from the untreated control fruit in week 4 than in weeks 2 and 3 ( Fig. 1) (z > 4.47; p < 0.001). However, by week 4, most fruit was overripe, and commercially would have been picked in week 3.
The low-rate insecticide sprays without baits reduced D. suzukii adult emergence from cherry fruit, although the effect was only significant in week 4 (z = 3.62; p < 0.001) (Fig. 1). Overall and in individual assessment weeks 2 and 3, the full field rate, reduced rate and low rate with molasses bait spray applications resulted in further reductions in D. suzukii (z ≥ 3.20; p ≤ 0.011). However, only the full-and reduced-rate treatments significantly lowered adult emergence compared with the low rate without bait treatment in week 4 (z ≥ 2.68; p ≤ 0.007). There were no significant differences in adult emergence between the full and reduced insecticide rate treatments, or between the low-rate insecticide treatments with molasses or Combi-protec baits.
Fruit sampled for cherry flotation tests three days before the treatments were applied contained a mean 1.5 (± SE 1.0) larvae per fruit before the introduction of laboratory reared D. suzukii. There were no significant differences in background numbers of larvae between plots that were allocated for different subsequent treatments. There were significant effects of spray treatment (Χ 2 6 = 45.34; p < 0.001), assessment week (Χ 2 3 = 42.28; p < 0.001), canopy section (Χ 2 2 = 37.78; p < 0.001) and of treatment × assessment week (Χ 2 18 = 47.30; p < 0.001) and treatment × canopy section (Χ 2 12 = 40.24; p < 0.001) interactions on the numbers of larvae in flotation tests.
There was a large increase in the number of larvae detected in week 4 compared with earlier assessment weeks in the untreated control plots (Fig. 2) (t 108 = 3.70; p = 0.001). In assessment week 1, the reduced rate of spinosad resulted in a significant reduction in the number of larvae detected in flotation tests (z = 2.62; p = 0.046); differences between other treatments were not significant (Fig. 2). Overall, the low-rate insecticide without baits reduced the number of larvae in fruit compared with the untreated control (t 108 = 3.26; p = 0.007), although the effect was only significant in week 4 (t 108 = 3.88; p < 0.001) (Fig. 2). In weeks 2, 3 and 4, the full-field-rate and reduced-rate sprays resulted in significantly fewer larvae than the low-rate sprays without baits (t 108 > 2.67; p ≤ 0.041) (Fig. 2). Averaged across all weeks, the full rate, reduced rate and low rate with bait sprays were not significantly different, and all resulted in fewer larvae than the low rate without bait sprays (t 108 > 3.41; p ≤ 0.004) (Fig. 2). This difference between the low-rate sprays without bait and other insecticide treatments was only significant for samples taken from the mid and far distances from the spray nozzle on the tree canopy (t 108 > 2.98; p ≤ 0.017) (Fig. 2).
Although the numbers of larvae in untreated control plots were the highest in the mid-tree canopy section (t 108 = 2.35; p = 0.05), equally good control of D. suzukii was obtained with full and reduced insecticide rate treatments and lowrate insecticide with bait spray treatments in all tree canopy sections. However, for the low-rate insecticide without bait spray, larvae numbers were significantly higher in the mid and far canopy sections than in fruit collected from the sides of the trees closest to the sprayer nozzle (t 108 ≥ 2.68; p ≤ 0.021) (Fig. 2).

D. suzukii control in raspberry, 2020
Fruit sampled three days before the treatments were applied, and before the introduction of laboratory reared D. suzukii, contained a mean 0.2 (± SE 0.1) larvae per fruit. There were no significant differences in background numbers of larvae between plots that were allocated for different future treatments. There were significant effects of spray treatment (Χ 2 4 = 37.38; p < 0.001), assessment week (Χ 2 6 = 133.90; p < 0.001), canopy section (Χ 2 2 = 9.90; p = 0.007) and of treatment × assessment week (Χ 2 19 = 42.04; p < 0.001) interaction on the numbers of larvae in flotation tests.
During weeks 1, 2 and 3, D. suzukii larvae numbers in samples from untreated control plots were low but increased from week 3 onwards (Fig. 3). Overall, the low insecticide rate without bait reduced larvae numbers (z = 2.74; p = 0.022), although the effect was not significant in individual assessment weeks. Insecticides at full field rate or low rate with bait almost completely controlled D. suzukii larvae numbers and remained significantly lower than untreated controls until week 5, two weeks after the final insecticide sprays were applied (z > 3.15; p < 0.003) (Fig. 3). Thereafter, D. suzukii larvae numbers in low-rate insecticide with bait spray treatments increased at a similar rate to those in untreated plots or plots treated with low-rate insecticides alone. By week 9, six weeks after the final insecticide sprays were applied, larvae numbers in the full-rate treatment remained significantly lower (z > 3.28; p < 0.008) than in the untreated control and low rate without bait treatments.
During weeks 1 to 5, there was no significant effect of the height at which fruit samples were taken from plants on the number of larvae detected in fruit samples in flotation tests. In weeks 7 and 9, samples taken from the top of plants (above 1.0 m) had significantly more larvae than samples taken from the bottom (below 0.6 m) of plants (z ≥ 2.99; p ≤ 0.008). However, compared with treatment effects, there were only small differences in the number of larvae detected in fruit samples taken from different plant heights, and the

Insecticide residues in cherry fruits
Insecticide residue concentrations detected in cherries picked immediately after spraying were within the EU MRLs for spinosad (0.2 mg kg − 1 fruit), cyantraniliprole (6 mg kg − 1 fruit) and acetamiprid (1.5 mg kg − 1 fruit). Residue levels of acetamiprid were < 0.02 mg kg − 1 fruit in all cherry fruit samples.
There was no significant difference between residue concentrations of fruit samples taken from the same treatments after the second spinosad (spray 3) or cyantraniliprole (spray 4) applications; means of these values are therefore shown in Table 3. There were significant effects of treatment (F 4,15 > 8.22; p < 0.001), tree canopy section (F 2,15 > 27.84; p < 0.001) and treatment × canopy section interaction (F 2,15 > 6.07; p ≤ 0.001) on the residues of spinosad and cyantraniliprole detected in cherries. Where insecticide residues were detected, these were significantly higher in fruit samples taken from the sprayed sides than from the mid or far sides of the trees (t 15 > 13.56; p < 0.001). No spinosad or cyantraniliprole residues were detected in fruit taken from the far sides of trees or from any tree canopy sections sprayed with the low-rate insecticide with bait. The higher residues of spinosad in samples taken from full-rate plots than from reduced-rate plots (t 15 > 5.39; p < 0.001) corresponded with the higher spinosad concentration in the full-rate applications. There was no significant difference in cyantraniliprole residues between treatments 1 and 2 which received the same cyantraniliprole rate.

Spray application and deposition analysis
Measured spray volumes per tree were 339 (± 13) ml for the full and reduced rates and 26 (± 3) ml for the low-rate sprays, corresponding with 102% and 101% of the target values in Table 1. Total amounts of spinosad applied in two full-, reduced-or low-rate applications were 165, 64 and 6 mg a.i. tree −1 , respectively. Total amounts of cyantraniliprole applied in two full-or low-rate applications were 123 and 5 mg a.i. tree − 1 , respectively.
The difference in spray deposition coverage between the low rate with Combi-protec applications at 40 and 80 L ha − 1 was not significantly different; only the former results are therefore presented here. There were significant effects of spray treatment (F 2,1008 = 430.89; p < 0.001), canopy section (F 2,1008 = 316.82; p < 0.001) and height 1 3 (F 2,1008 = 5.79; p = 0.003), leaf surface (F 1,1008 = 9.47; p < 0.001) and interactions between these factors (F 8,1008 ≥ 3.07; p < 0.01), except canopy section × leaf side, on spray deposition coverage. For the full-rate spray (500 L ha − 1 ), leaf surfaces nearest the spray nozzle received the highest coverage (> 20%) and, with the exception of upper leaves at the top of trees, leaves furthest from the nozzle received the lowest spray deposition (≤ 0.5%) (t 1008 > 81.60; p < 0.001) (Fig. 4). For the bait spray (40 L ha − 1 ), leaf surfaces nearest the spray nozzle received a higher coverage than leaves in the middle of the canopy or furthest from the spray nozzle (t 1008 > 32.22; p < 0.001). Spray deposition coverage was significantly (t 1008 > 2.73; p < 0.018) higher (between 5 and 134 times higher in corresponding positions) for the full-rate application than for the low rate with Combi-protec application, except on leaves furthest from the spray nozzle at the middle tree height (both leaf surfaces) and top of trees (upper leaf surface) (Fig. 4).
Spray deposition data and analysis of full field rate and low rate with bait spray applications for raspberry are presented in Noble et al. (2021).
No phytotoxicity symptoms were observed on the foliage of any of the cherry trees or raspberry plants.

Discussion and conclusions
In agreement with Shawer et al. (2019) and Shaw et al. (2019a), full-field-rate applications of spinosad and cyantraniliprole provided good control of D. suzukii in a sweet cherry crop under polythene cladding. Semi-field-scale cherry and raspberry experiments demonstrated that using phagostimulant baits, the full field rate of these insecticides Fig. 4 Spray deposition as mean percentage (n = 25, ± SE) coverage on the leaves of cherry trees across canopy sections near, mid and far from the sprayer nozzle, and heights and leaf surfaces (Logit transformation). Trees were sprayed at (a) full field rate or at (b) low rate with Combi-protec. Asterisks denote leaf surfaces where spray deposit coverage was significantly lower (p = 0.05*, 0.01** or 0.001***) for the low rate with Combi-protec spray than for the full-rate spray could be reduced by 96% without loss in D. suzukii control. The only exception was in the final assessment on cherries when they were picked from trees with mainly overripe fruit and under severe pest pressure. The reduction in insecticide rate was attained by dilution, applying the spray as a band across the middle height of the trees rather than to the entire canopy, and by increasing the speed of application across the crop. The lower rates of application resulted in 11% lower detectable insecticide residues than from raspberries sprayed with the full rates (Noble et al. 2021), while in cherries, no insecticide residues could be detected. Insecticide residues detected in cherries sampled from the sides of trees sprayed with one or two full-field-rate applications (Table 3) were within the ranges of cherry fruit residues of cyantraniliprole (0.04-3 mg kg − 1 ) and spinosad (0.03-0.12 mg kg − 1 ) detected by VanWoerkam et al. (2022) following one to three applications of these insecticides. They used the same application rate for spinosad but an application rate for cyantraniliprole that was 67% higher than that used here. There was no loss in efficacy when the applied rate of spinosad was reduced by 60% without bait, when combined with alternating sprays of cyantraniliprole. However, the ability to apply insecticides at much lower doses with baits, without compromising efficacy, minimises environmental impact and the risk of pesticide resistance, since the ingested dose with baits may be increased (Noble et al. 2019). The results for raspberry showed that the efficacy decline of the full-rate sprays was slower than that of the low-rate insecticides applied with bait sprays. This potentially increases the risk of pest exposure to a sub-lethal dose, compared with a low-rate application, where there was faster loss in efficacy.
In cherry trees, the spray coverage was on average 30 × higher with the full-field-rate application than with the bait and low-rate spray. In raspberry, crop spray coverage was on average eightfold higher with the full field rate than with the bait and low-rate insecticide spray (Noble et al. 2021). In raspberry, larvae numbers were lower following full-rate or low-rate insecticide with bait sprays than low rate without bait sprays, irrespective of plant position. For low rate without bait sprays on cherry, the decreasing spray coverage on leaves with increasing distance from the spray nozzle corresponded with decreasing control efficacy of D. suzukii and contrasted with the good D. suzukii control efficacy in all canopy sections when bait was added to the spray. This suggests that the pest was attracted from other parts of the tree canopy to spray droplets, and that uniform spray coverage is less important when spraying in combination with baits. This offers an advantage over conventional sprays where good coverage of fruit crop canopies is needed to give uniform control of D. suzukii (Lewis and Hamby 2020).
Our study demonstrates that more rapid and less uniform application of bait sprays with insecticide can reduce application time by more than 90% compared with high volume, full-rate sprays. However, with very high pest pressure, it is possible that higher levels of D. suzukii control would be achieved by bait spraying both sides of cherry trees, rather than a single side as in this experiment.
Combi-protec and molasses were equally effective as bait sprays although molasses was significantly cheaper (Noble et al. 2021). However, molasses was not an approved for this use at the time of conducting this experiment but could reduce costs compared with currently approved full rates of insecticides by at least 50% (Noble et al. 2021). Field walnut trials showed that when used with an insecticide, cattle feed molasses were at least as effective as commercial baits based on corn steep liquor or gluten meal in control of walnut husk fly (Rhagoletis completa Cresson), although laboratory studies showed that feed molasses were less effective than store bought molasses as a phagostimulant bait (Van Steenwyk 2018). The difference may be due to a higher sugar content of the store bought molasses, which were used here, but the influence of different molasses sources and analysis on their effectiveness as phagostimulant baits for control of D. suzukii, R. completa and other fruit fly pests requires further investigation.
The attractiveness of molasses (~ 55% sugars) applied as a low-rate bait spray to non-target species has not been investigated. Whereas low-rate bait sprays are applied as a band to the crop foliage, full-field-rate insecticide sprays result in 8 × to 30 × greater coverage on foliage and application to flowers and fruit of soft fruit crops, potentially increasing the risk of contact with pollinators and beneficial insects. No effect on the mortality of bees was found when a spray containing spinosad and the sugar-containing bait, Combi-protec, were applied to field grapes (Görlich et al., 2016). Toledo-Hernandez et al. (2021) found that ACT TRA SWD bait, containing 36% sugars, was not attractive to honeybees. Cabrera-Marín et al. (2016) found that a highdensity-rate application of spinosad caused high mortality in bees, whereas mortality following a low-density-rate application, with or without a corn steep liquor (2.5% sugars) bait, GF-120, was low and not significantly different to a water control.
All emerged adults in the cherry emergence tests were identified as D. suzukii. Similarly, in an experiment on strawberry, > 96% Drosophila species adults in emergence tests were D. suzukii (Noble et al. 2021). This contrasts with the raspberry experiment where 16% to 45% of adults in emergence tests on ripe fruit were other dipteran species, predominantly D. melanogaster. Studies in grape by Rombaut et al. (2017) showed that entry wounds made in the fruit by D. suzukii can be exploited by D. melanogaster and the same may be true for raspberry.
The lower D. suzukii larvae numbers compared with adults emerging from fruit samples agrees with Shaw et al. (2019b) and may be due to the non-detection of eggs and the younger first instar larvae which can be difficult to detect using the flotation method (Van Timmeren et al. 2017). However, in both the raspberry and cherry experiments, there was a close correlation between larvae flotation and adult emergence numbers, except in the final sample of cherries where the pest pressure was very high. Unlike the raspberry experiment, ripe unused cherries were not removed from compartments, and experimental trees were surrounded by an untreated cherry orchard, resulting in high pest pressure by the fourth week of the experiment.
In the UK, control of D. suzukii accounts for 30%, 48% and 50% of insecticide applied to soft fruit, grapes and cherries, respectively (Ridley et al. 2020a,b). This study has demonstrated that phagostimulant baits can be used to substantially reduce the amount of insecticide applied to fruit crops and hence surrounding environment. There is also potential for bait sprays to broaden the range of available effective insecticides. The D. suzukii control efficacies of lambda-cyhalothrin and deltamethrin were increased by Combi-protec in laboratory bioassays (Helsen and van der Sluis 2018;Noble et al. 2019). Future studies should also test the potential of combining baits with biocontrol agents (Yousef et al. 2018;Lee et al. 2019) so that fruit growers are no longer reliant on applying synthetic agrochemicals for control of D. suzukii.

Author contributions
RN, MTF, BS and ECW conceived the study. AH, AW and RN calculated and calibrated the spray applications. AW, ECW, AD-P and RN collected the data. GD performed the statistical analysis. RN drafted the first version of the manuscript, and all authors contributed to the final text.