The use of light spectrum blocking films to reduce populations of Drosophila suzukii Matsumura in fruit crops

Spotted wing drosophila, Drosophila suzukii, is a serious invasive pest impacting the production of multiple fruit crops, including soft and stone fruits such as strawberries, raspberries and cherries. Effective control is challenging and reliant on integrated pest management which includes the use of an ever decreasing number of approved insecticides. New means to reduce the impact of this pest that can be integrated into control strategies are urgently required. In many production regions, including the UK, soft fruit are typically grown inside tunnels clad with polyethylene based materials. These can be modified to filter specific wavebands of light. We investigated whether targeted spectral modifications to cladding materials that disrupt insect vision could reduce the incidence of D. suzukii. We present a novel approach that starts from a neuroscientific investigation of insect sensory systems and ends with infield testing of new cladding materials inspired by the biological data. We show D. suzukii are predominantly sensitive to wavelengths below 405 nm (ultraviolet) and above 565 nm (orange & red) and that targeted blocking of lower wavebands (up to 430 nm) using light restricting materials reduces pest populations up to 73% in field trials.


Results
Laboratory choice tests. Initial laboratory choice tests investigated the spectral sensitivity of D. suzukii before comparing their spectral preferences. D. suzukii were given a choice of two visual stimuli (light of different spectra or dark controls). The ratio of insects choosing to approach each stimulus after 30 min exposure was used to define the attraction index (AI). AIs of 1 or -1 would indicate that all individuals approached either source, whereas an AI of 0 indicates an equal preference.
To reveal spectral sensitivity flies chose between a dark stimulus and a set waveband of light (Table 3, 340 to 660 nm)with perceptible light triggering the animals innate phototactic approach behaviour 18 . D. suzukii females preferentially approached light in the near-UV range (340, 365, 405 nm) in common with D. melonagastor 30 that possess UV-sensitive photoreceptor R7p and R7y and true colour vision with photoreceptor rhodopsins Rh3 and Rh4 [31][32][33] . However, D. suzukii females did not preferentially approach light in the 430-565 nm range as would be predicted possessing similar blue-green sensitive RX and RY photoreceptors as D. melonagastor but rather show preferences for orange and red light (617 and 660 nm), raising the potential for divergence in spectral sensitivities between species (Fig. 1i).
Then differential phototaxis experiments 18 were conducted which tested wavelength preference between the discrete wavebands listed above and a 365 nm control (Table 3). No significant preference was observed between 365 and 340 nm stimuli but D. suzukii showed a preference for 365 nm when presented in combination with higher wavebands (405 nm-660 nm) (Fig. 1ii). The strength of the effect increased from AI's of − 0.25 and − 0.3 (p < 0.05) for lower wavebands representing violet (405 nm) and blue light (430 nm), to a consistent c.-0.5 (p < 0.01) for all higher wavebands. In summary, D. suzukii showed a differential and preferential phototactic response to ultra-violet light over visible light.
Semi-field "no-choice" fruit damage experiments. A hypothesis was then tested to investigate whether removing ultra-violet radiation in the terrestrial environment via spectral filters applied to greenhouse cladding materials could reduce pest impacts.
Spectral transmissions of the materials are shown in Fig. 2; experimental materials absorbed UV up to and between 350 to 430 nm.
In the first year (2016) semi-field experiments with strawberry crops, the effect of cladding absorbing up to 350 nm of UV against an "open" control cladding on D. suzukii emergence from strawberry fruit was tested. In addition, the impact of clear versus highly diffuse materials which scattered a significant proportion of inbound radiation was also tested. These experiments showed there were no significant differences between the UV light transmission clear or diffuse cladding in the umbers of D. suzukii which emerged per gram of fruit; UV light transmission (F [1,6] (Fig. 3)). Statistical analyses of the three-year combined data set showed significant differences between the UVattenuating claddings for D. suzukii emergence per gram of fruit (F[4,38.7] = 4.4, p = 0.005). Adult D. suzukii emergence was reduced up to 8%, 22%, 34%, and 73% for UV350, UV370, UV400, and Lumitec claddings respectively, compared to the UVopen cladding (Fig. 4).
In choice experiments between two cloche compartments, both containing blueberry fruit in a Petri dish (no plant foliage), but with different cladding materials, D. suzukii preferred to lay eggs under UVopen film compared to the light exclusion (dark) cloche (t = 6.34, df = 9, p < 0.001, Fig. 5ii). There was no preference when UVopen and Lumitec (absorbs up to c.430 nm) were compared (t = 0.58, df = 9, p = 0.575), nor Lumitec and light exclusion (t = 0.1, df = 9, p = 0.923, Fig. 5ii). The generally increased oviposition in UVopen vs. Lumitec, and Lumitec vs. the covered/dark cloche (Fig. 5ii) was noteworthy.  Table 1. Numbers of adult D. suzukii emerging per gram of strawberry fruit from tunnels clad with different UV-attenuating films in 2016 and 2017 (mean (± S.E)). There was no significant difference between cladding treatments in either year, although there was a suggestion of decreasing D. suzukii with increasing UV attenuation.    www.nature.com/scientificreports/ many insects 34 with clear benefits for sky detection 35 which, when combined with a positive phototactic reflex, will naturally guide individuals to the outdoors. This hypothesis is supported by the data from cloche experiments in which D. suzukii females preferentially entered tunnels illuminated with UV (and near visible) light over those with the same spectra attenuated. It is noted however that attraction to red/orange and not blue/green light diverges from the predictions of a highly conserved spectral sensitivity across fruit-fly species 26 with D. melanogaster sensitivity decreasing above > c.480 nm 21,22 . It is fascinating to consider whether D. suzukii have evolved away from the common fruit-fly not just in its ovipositor function but also in its visual sensing with likely adaptations to improve visual fruit detection. Neuroanatomical analysis of the characteristics of the photoreceptors present in D. suzukii's eye present an obvious research avenue.

Discussion
In differential phototaxis experiments, D. suzukii showed a preference towards UV light (365 nm) over higher wavelengths (405-430 nm: weak preference; 455-660 nm: strong preference). It is therefore clear that the need to remain outdoors (expressed by a preference for UV light) is dominant in D. suzukii, but a secondary attraction to wavelengths (orange and red) that provide a possible secondary mechanism allowing for robust visual fruit detection was also noticed. Colour-opponency is thought to underpin colour perception in fruit flies, and has recently been used to enhance the performance of artificial fruit detectors 36 . Computational modelling of the sensory and neural perception systems facilitating fruit detection in insects will allow evermore targeted and nuanced interventions and provides an excellent channel for future investigation.
Short term choice experiments in field cloches showed adult females were more likely to enter the UVOpen treatment compared to Lumitec (430 nm absorption), however, there was no difference in numbers of eggs laid between treatments. These effects are not surprising since Drosophila spp. have complex responses to UV, whilst orientation has a preferential response to UV, females have an aversion to laying eggs under UV 24 . In addition, cues other than phototaxis could play a role, for example, olfactory responses have a significant role in fruit localisation by Drosophila spp. 37,38 . The complexity and contradictory nature of these responses to UV challenges whether the use of UV absorbing materials can support efficacious control of the pest in greenhouse crops. Net effects of UV on insect populations can only be established through longer term projects over multiple lifecycles.
Here, net effects were tested in "no choice" experiments conducted over a full production season. Although the magnitude of the response of D. suzukii to lower wavelengths of UV-attenuating claddings was not significant in the low replicated (× 3) no-choice experiments in the first two years the trend was encouraging and following experiments showed a significant reduction in D. suzukii in fruit with increasing UV-attenuation in experiments with higher UV-attenuation and increased replication (× 8). These showed D. suzukii egg laying decreased significantly with increasing attenuation of light transmission up to 430 nm (maximum tested). Numbers of eggs laid (measured by adult D. suzukii emergence from fruit) were reduced up to 8%, 22%, 34%, and 73% for UV350, UV370, UV400, and Lumitec (absorbs up to 430 nm) respectively, compared to the UVopen claddings. These experiments confirm the impact of UV attenuation over multiple insect lifecycles and suggest that the approach can be used to suppress population and infestation pressure. D. suzukii were less likely to reproduce in areas under UV attenuation of 430 nm even in the presence of a fully cropping field strawberries which resulted in less fruit damage and losses of yield. These impacts are of commercial significance to fruit farmers but additional work is required to quantify the mechanisms driving the response. It was shown that short-term variance in UV/blue light can impact fly orientation including egg laying aversion but longer-term mechanisms have not been fully explored. This is the first time that UV-attenuating claddings have shown efficacy for contributing to the control of D. suzukii and the first study conducted using strawberry 14 . In Fennell et al. (2019) 14 the main mechanisms of cladding suppression of insect pests were considered to be (i) positive phototaxis to ultraviolet light sources, and (ii) reduced take-off and flight behaviour when UV was absent. Suppression of other common pests including thrips 13,39-41 and aphids 13,15,39,[42][43][44][45][46][47][48][49][50][51][52][53] are attributed to both a reduction in pest immigration into the crop 15 , and within crop movement 54 . In our field crop studies, D. suzukii was artificially introduced into the cropping area demonstrating a within-crop effect. In addition, the cloche experiments relied on D. suzukii immigrating into the cladded areas. The latter was not affected by the claddings when fruit was available indicating that the primary driver when given no choice was to orientate to fruit-most likely with olfactory cues associated with both fruit compounds and associated fermenting yeasts 37,55 .
The claddings appear to be interfering with host location and subsequent egg laying. In the blueberry cloche experiment where fruits were introduced on a Petri dish and not with background plant foliage, D. suzukii successfully laid eggs. It is possible that D. suzukii locates fruit visually by colour contrast (with plant foliage) as suggested by Little et al. (2018) 29 rather than colour appearance. From their experiments D. suzukii had a limited ability to comprehend red. Female D. suzukii were attracted to purple sticky discs (5 cm) on a white background 29 . In another similar recent study red, purple, and black disks were more attractive when presented against a white background. Male and female D. suzukii responded identically in these tests. Significantly more male and female D. suzukii were captured on the red and yellow disks than those presenting the corresponding grayscale for that colour 28 . Hence, background contrast to fruits is important in fruit recognition and it is possible that claddings used in this work interfered with this contrast and hence orientation to fruits within the strawberry crop, but not in blueberries placed on the ground on a Petri dish.
Potentially a more effective strategy would be to incorporate the claddings as part of an integrated pest management approach including repellents, to further inhibit D. suzukii entering crops 56,57 , and attractants where semiochemical based baits are placed outside the crop to intercept and further reduce immigration into the cropping area 58 . This could have significant consequences for reducing the need for chemical plant protection products in fruit crops and the reduction of future insecticide resistance 59,60 .
The impact of such claddings on natural enemies 49 and pollinators 61 , which are key to the production of many fruit crops, requires further investigation. The commercially produced biological control agents Orius laevigatus and Amblyseius swirskii exhibited reduced dispersal rate under and a preference for the lower UV  49 . In addition, impact of these claddings on other fruit types is recommended as they may yield different responses.   Laboratory choice tests. The attractiveness of 9 wavelengths isolated in dark environment were assayed in binary choice "Y-maze" 62 tests using 6-12 days old female D. suzukii. Light attraction experiments were conducted between 13 Dec 2017 and 17 Jan 2018 in a laboratory at NIAB EMR. A 3-D printed Y-maze with a central release chamber (length 116 mm, height 30 mm, depth 34 mm) and two arms with the light sources at the distal ends was constructed. The arms were black anodised aluminium beam tubes (THORLABS, SM1E60, outer diameter 30.5 mm) to minimise light reflection inside the arms. Access to the arms was blocked by use of UV-fused silica windows to prevent flies leaving Y-maze arms. The LED light wavelengths ranged between 340 and 660 nm. LEDs were fitted with a collimation adapter to ensure a uniform light beam was illuminated inside the Y-maze arms. The light beam was controlled by an aperture iris. Since the actual wavelength of LED light sources had a bandwidth of ± 10-30 nm (depending on the light source) bandpass filters were used to reduce the bandwidth and ensure the output wavelength was as close as possible to the nominal wavelength as shown in Table 2. Figure 6 shows the experimental optical Y-maze apparatus.
The apparatus was tested for bias within the Y-maze choice test and it was shown that where identical light treatments were applied to both arms, D. suzukii females had no significant preferences.
All of the optical setup apparatus was fixed to an aluminium breadboard (450 mm × 600 mm × 12.7 mm). Sixty to seventy female D. suzukii (6 -12 days old) were transferred into the release chamber. Only females were used as these are responsible for fruit damage and, therefore, the focus of behavioural disruption. Subsequently the gates were opened simultaneously, and the flies were exposed to the light/dark conditions. After 30 min the gates were closed and the number of flies in each arm and the release chamber was counted and proportions determined. All experiments were carried out at 21 ± 1 °C and in a dark room to prevent interference with artificial and/or natural light. The light intensity in each arm was measured by an Ocean-FX spectrometer and set to 10 µmolm −2 s −1 using LED controllers to ensure flies were exposed to equal light photon flux density in both arms for all tests.
Each test was repeated 6 times and approximately 60 flies were used in each test. The tests were classified into 3 groups (Table 3). The first group, control was to ensure there was no bias between the arms of the equipment. The second group tested light vs dark (phototaxis 18 ) to ensure that D. suzukii responded to the different wavelengths in the absence of light in the other arm. Finally, the wavelengths were tested against 365 nm for preference to orientate (differential phototaxis 18 ). 365 nm was chosen with prior knowledge of D. melanogaster sensitivities to blue and UV 21,22  Twelve tunnels (12 m × 2 m × 2.1 m; Knowle Nets Ltd, Bridport, Dorset, UK) were covered with insect mesh (1 mm × 1 mm, Knowle Nets Ltd, Bridport, Dorset, UK) with cladding materials (treatments) over the top. Test films were cut off at 30 cm above the ground to provide ventilation to the strawberry plants in the tunnels. The tunnels were orientated in north-south direction. In 2016, two UV-blocking films with different UV light transmission levels were tested. The films were coded UVopen and UV350 in a clear and diffuse version; there were three replicates per treatment. The following year four clear films were tested; UVopen, UV350, UV370, and UV400 with three replicates. Then in 2018 the 12 existing tunnels were divided by fine mesh (Dunelm Ltd., Syston, Leicestershire, UK) into two compartments and three films were compared; UVopen, UV370, and Lumitec, resulting in eight replicates of each (Table 4). Table 4. Overview of the number of replicates used in the field tunnel trials in each year. * = not tested in that year. See (Fig. 2) for wavelength and transmission measurement. www.nature.com/scientificreports/ The light transmission of the experimental polyethylene film was measured from 300 to 2,500 nm at 2 nm steps using an Agilent Cary 7,000 Universal Measurement Spectrophotometer equipped with a diffuse reflectance accessory; an integrating sphere to capture all scattered light post transmission. Transmissions were measured on a 2 × 2 cm of film samples with the beam incident angle at 90°. Spectral transmissions of each of the materials are shown in Fig. 2.
Bare root plants or plug plants (cv. Finesse and proprietary cultivar, respectively) were planted in 1 m peat bags. Bags were arranged end-to-end in the tunnels in one (2016 and 2017) or two rows (2018). In 2018, both cultivars were in all tunnels (10 bags of each, side-by-side in a row). Fertigation (fertilizer plus irrigation) was set to commercial standard and pesticides for disease control were applied only if necessary. At pink fruit stage (BBCH 81 -85) 63 , known numbers of D. suzukii males and females were released into the tunnels (Table 5). In 2016 and 2017, air temperature and relative humidity was recorded in one tunnel of each treatment using a data logger (EL-USB-2; Lascar Electronics Ltd, Whiteparish, Wiltshire, UK). In 2018, a data logger (Tinytag plus 2 TGP-4500; Gemini Data Loggers Ltd, Chichester, West Sussex, UK) in each tunnel recorded air temperature and relative humidity.
During each trial period, where possible and where available, samples of 20 ripening and ripe fruits from the tunnels were collected every two weeks. During the experiments two cloches were connected on opposite sides of the release chamber with the remaining two sockets sealed. The first experiment was carried out from 4 September to 3 October 2018. Twenty female D. suzukii were released into the central chamber. Five blueberries cv. Duke were placed in a Petri dish in each cloche and the flies were left to oviposit for 24 h. Blueberries were used instead of strawberries because it is easier, and therefore more accurate, to observe eggs in these fruits. Subsequently, eggs in each fruit were counted under a light microscope. All film treatments were successively tested against each other. Each combination was repeated ten times. In a second experiment, eight releases of 50 female D. suzukii were done between 22 October and 3 November 2018. All treatments were simultaneously compared to one-another. Instead of blueberries, yellow sticky traps were used to trap D. suzukii adult females orientating into each cloche arm. This omitted any effect of fruit volatile compounds attracting flies and relied upon visual orientation.

Statistical analysis
Data from the laboratory experiment was used to calculate Attraction Index (AI) which is the number of times a light treatment on one arm of the Y-maze was more attractive to flies compared to the light treatments 65 . AI = (N1-N2)/ (N1 + N2) where N1 and N2 are the number of the flies which have chosen arms 1 or 2, respectively. Then binominal distribution with Null hypothesis was used (equal distribution (= 0.5) between Y-maze arms). For this the number of flies (left and right) from all test replicates (6 reps per test) were summed and the values tested whether the sample was significantly (P < 0.05) different from the control.
The data from semi-field trials (numbers per gram of strawberry in each compartment at each sampling date), from each year, were analysed using repeated measures analysis of variance (ANOVA) in Genstat 13 (VSN International Ltd., 2010). When the assumptions of ANOVA were not met, data were square root transformed before analysis. However, reported means are from untransformed data for presentation purposes. The differences between means were compared using Fisher's least significant difference (LSD) test at the 5% confidence level. The combined data set of the three consecutive years was analysed using restricted maximum likelihood (REML) in Genstat 13 (VSN International Ltd., 2010). The diffuse versions of the UVopen and UV350 films www.nature.com/scientificreports/ were excluded from the analysis as they were only tested in 2016. The differences between predicted means were compared using Tukey's honest significant difference (HSD) test at the 5% confidence level. The data from both cloche trials (numbers of eggs laid and numbers of D. suzukii on sticky traps) were analysed using a two-sample paired t-test in Genstat 13 (VSN International Ltd., 2010). When the assumptions of normality were not met, data was square root transformed before analysis. However, reported means are from untransformed data for presentation purposes. Means were compared using Tukey's honest significant difference (HSD) test at the 5% confidence level. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.