Sweet-potato whitefly, Bemisia tabaci (Gennadius), is the first insect pest found in Korea in 1998, which is known to affect more than 900 host species, including many vegetables and ornamental crops (Lee et al., 2000;Helmi, 2011). The most problematic biotypes among the more than 24 globally known biotypes are the B and Q biotype (Lee et al., 2005;Yang et al., 2009;Lee et al., 2012). Biotype Q is known to mediate more than 100 viruses including tomato yellow leaf curl virus (TYLCV) (Kim et al., 2008) and shows high resistance to neonicotinoid-based insecticides (Nauen et al., 2002;Lee et al., 2012). Therefore, it is necessary to identify diverse new strategies to control B. tabaci.

The push-pull strategy was first used in 1987 in Australia to study repellent and attractive stimuli for the control of Helicoverpa sp. This approach has been developed as an integrated pest management tool that changes the distribution and density through behavioral control of pests or natural enemies (Cook et al., 2007;Khan and Pickett, 2008;Bhattacharyya, 2017). In the early days, companion crops (attractant or repellent plants) were mixed with crop cultivation, but in recent years, attempts have been made to utilize volatile compounds capable of attracting or avoiding insect pests (Cook et al., 2007;Bhattacharyya, 2017).

Apart from this, technologies for pest forecasting have been developed by using colors and light sources. If we use these forecasting techniques as a control strategy, we think that the push-pull strategy will be more successful. Colored traps are mainly used to monitor the density of insects. Kim and Lim (2011) devised a sticky trap with a yellow circular pattern on a black background as a way to capture more B. tabaci by using this visual cue. In recent years, along with colored traps, artificial light sources such as LED have also been widely used as a means of integrated pest management (Shimoda and Honda, 2013;Zheng et al., 2014). Most plants have functional compounds such as repellents, feeding deterrents, toxins, and growth regulators to defend against the attack of herbivorous insects (Maia and Moore, 2011). Plants that are damaged by insect pests are known to produce a variety of volatile organic compounds (VOCs) to protect them. Among these volatile substances, herbivore-induced plant volatiles (HIPVs) are closely related to each other, attracting other insects or natural enemies (Dicke and Baldwin, 2010;Xiao et al., 2012). Darshanee et al. (2017) have reported that Trialeurodes vaporariorum were strongly attracted to volatile compounds released from tomato leaves. Yang et al. (2010) have reported strong contact toxicity with essential oil extracted from Thymus vulgaris (insecticidal effect) and strong avoidance effect of the oil extracted from Pogostemon cablin (repellant effect) for B. tabaci. In order to enhance biological control of insect pest species, banker plants are widely used as a means to maintain insects that are natural enemies of these pests for a long period in the crop fields (Frank, 2010).

The purpose of this study was to select a companion plant, color, volatile matter, LED (light source), and banker plant that can regulate B. tabaci and attract natural enemies of this pest to establish a push-pull strategy in tomato greenhouses.

Materials and Methods

Insect rearing

Tomato (Solanum lycopersicum L.) was used as a host plant for rearing B. tabaci in the laboratory. One-month-old tomato seedlings were placed in a mesh cage (30 × 30 × 30 cm). The temperature was maintained at 25 ± 2℃, the relative humidity at 60-80%, and the light condition was maintained at 16 h of light and 8 h of darkness Occurrence pattern of B. tabaci in the greenhouse.

Occurrence pattern of B. tabaci in the greenhouse

To establish the optimal time for implementing the push-pull strategy, yellow sticky traps were placed at nine locations in a tomato (‘Superdotaerang’) greenhouse in Hwaseong area according to the method detailed in Song et al. (2014). The traps were collected daily to investigate the density of adult B. tabaci.

Selection of companion plants

Nine species including gypsophila (Gypsophila elegans), rosemary (Rosmarinus officinalis), marigold (Tagetes erecta L.), petunia (Petunia hybrida Vilm), spearmint (Mentha arvensis L.), lavender (Lavandula angustifolia L.), peppermint (Mentha piperita), geranium (Pelargonium inquinans), and sweet basil (Ocimum basillicum) were tested for their effectiveness as a companion plant in the tomato green house.

Each of these companion plants were placed in the tomato greenhouse in three replicates of three pots each. To ensure minimum interference from the other treatments, the distance between plants were maintained at 5 m. The density of B. tabaci was determined by replacing the yellow sticky traps on the top of the tomato plants every day and counting the trapped individuals.

Selection of attractant color

To determine which colors attracted and which repelled the pest species, we used red, black, white, green, blue, and yellow colored papers (21 × 15 cm each) sandwiched between two transparent sticky traps. The traps were placed at a distance of approximately 1 m parallel to the B. tabaci habitat. At each sampling location, a different colored trap was installed. The density of B. tabaci per trap was investigated by sequentially changing the position of the traps daily for seven days to minimize the effect of trap location.

Selection of volatile compound

Twelve volatile organic compounds (Table 2) were used in the tomato planted greenhouse (1200 m²) to investigate their effect as either an attractant or repellent of B. tabaci. Commercially available insect pheromone lures were used. Each lure was treated by adding 30 μl of one of the twelve compounds, sealing it, and allowing it to be absorbed in a refrigerated condition (4℃) for 24 hours. A volatile compoundtreated lure was attached to the center of the yellow sticky trap and placed on the top of the tomato plants. The distance between the traps was maintained at more than 5 m to reduce interference between treatments. The density of the trapped B. tabaci adults was examined after three days.

Selection of LED source

Based on literature, we tested 520, 660, and 730 nm LEDs (Fig. 1) and a composite light of 450 + 660 nm to select the light sources that can be packaged with companion plants. The test was carried out using a container (3 m × 3 m × 3 m) that prevented light from outside and in which the temperature could be maintained at 25 ± 2℃. Inside the container, one LED was placed on each wall and a tomato seedling was fixed with a colorless sticky trap, placed under each LED source. Then, approximately 350 adults were irradiated from the central part of the container and the number of B. tabaci adults captured per light source was investigated after 24 hours. All tests were conducted in three replicates.

Selection of banker plant

Based on literature and preliminary tests, we identified buckwheat (Fagopyrum esculentum Moench) as a potential banker plant to maintain natural enemies. Two pots (40.5 × 18.5 × 17.0 cm) planted with buckwheat (20 days after sowing) were placed per point at three points in the tomato planted greenhouse (1200 m²). The distance between the points was maintained at 5 m or more to minimize the effect between treatments. The density of B. tabaci was investigated daily using a yellow sticky trap and six tomato plants around the banker plant were examined for the natural enemies.

Integrated push-pull strategy effect test

The results from each of the above experiments were synthesized and tested for their combined effect in controlling B. tabaci in a tomato greenhouse (120 m²). Sticky roll traps (20 cm × 100 m) were placed in two rows 15 cm above the tomato plants. Three carvacrol treated lures were hung above a tomato plant in the center of greenhouse as a repellent, and two methyl isonicotinate treated lures were placed at each greenhouse entrance areas as an attractant. Yellow sticky traps were set at two sites in the tomato greenhouse and the density of adult B. tabaci was investigated at weekly intervals. The detailed set up for this integrated push-pull strategy is shown in Table 1. Table 2

Statistical analysis

The relationship between volatile organic compounds and B. tabaci density was analyzed using SAS PROC ANOVA. The incidence rate of B. tabaci by LED source was analyzed using SAS PROC ANOVA after arcsine transformation (SAS Institute, 2013).

Results and Discussion

Occurrence pattern of B. tabaci in the greenhouse

Tomatoes in Gyeonggi area were cultivated twice a year. The B. tabaci density was significantly higher in the autumn season than in the spring season. In the spring season, B. tabaci increased greatly from early June, and in the autumn season, it increased from mid-September. Therefore, it can be surmised that the appropriate period for controlling B. tabaci using natural enemies is late May for the spring season and early September for the autumn season. Spatially, B. tabaci density gradually increased from the entrance to the inside of the greenhouse (Fig. 2). If B. tabaci can be efficiently attracted to a point near the greenhouse entrance, and if its natural enemy is also present at this location, not only can B. tabaci be detected quickly but the natural enemy can also be established efficiently. In addition, if a banker plant for the natural enemy is added at this location, it is possible to increase the effect of the natural enemy and minimize the pest density.

Selection of companion plants

Among the companion plants used in this test, geranium showed the most repelling effect for B. tabaci, lowering the density by 52% in comparison to untreated replicates. However, when spearmint, rosemary, and gypsophila were planted with tomato, B. tabaci density was, respectively, 265, 229, and 204% higher than when no companion plants were present (Fig. 3). Geranium may be used for the purpose of repelling B. tabaci using the plant itself or its extracts. Additionally, this species can be planted around the tomato greenhouse to inhibit B. tabaci infestation. On the other hand, plants attracting B. tabaci can be planted outside the tomato greenhouse along with treating the soil with systemic insecticides (Choi et al., 2016) to control B. tabaci around the tomato greenhouse. However, one of the most important aspects for consideration is whether the companion plant can play a lasting role. The companion plants selected in this study often showed other pest outbreaks and abnormal growth during summer. Therefore, in this study, when developing and testing an integrated system using the various individual approaches studied here, eggplant seedlings that have been reported to strongly attract B. tabaci (Hasanuzzaman et al., 2017) were used as companion plants.

Selection of attractant color

The rate of attraction of B. tabaci was highest for yellow color at 62.4%. For white it was 1.6%, while it was lower than 0.8% for blue, red, and black was (Fig. 4). There are several studies that maximize the attraction of pests by using contrasting colors (Kim and Lim, 2011;Vernon and Gillespie, 1995). Vernon and Gillespie (1995) showed that western Frankliniella occidentalis was noticeably less attracted to yellow colored traps than those that were purple or yellow on blue backgrounds. Chu et al. (2000) reported that major sucking insects such as whitefly, thrips, and leaf hoppers preferred a color spectrum of 490 to 600 nm, and this spectrum was similar to the spectral reflectance curve on the back surface of green leaves. Our results suggest that it is possible to maximally attract B. tabaci by using a yellow sticky trap with a black or red background, but further studies will be needed on this. Also, using red and black colored nets will be useful in blocking the influx of B. tabaci from outside.

Selection of volatile compound

Treatments where ocimene and carvacrol lures were used showed the lowest density of B. tabaci of 26% and 33%, respectively (Table 2). This suggests that these compounds acted as a B. tabaci repellant. On the other hand, the density of B. tabaci when methyl isonicotinate was used was 179% higher than when no VOC treated lure was used. The effect of volatile organic compounds in attracting insect pests as well as natural enemies has been studied (Koschier et al., 2002;Tan and Liu, 2014). The aim of this study was to improve the efficacy of natural enemies by using the push-pull strategy for controlling B. tabaci efficiently. Therefore, it may be desirable to install lures with methyl isonicotinate, which was highly effective in attracting B. tabaci, at the entrance of the tomato greenhouse where this pest first occurs. In addition, lures with carvacrol, which was the most efficient repellent, could be installed inside the tomato greenhouse.

Selection of LED source

We selected an LED source that could be packaged with the banker plant for the attraction and early detection of B. tabaci in the tomato greenhouse. The 520 nm light source was the most effective in attracting B. tabaci, at 66.5%, while the lowest attraction rate was 3.7% for a combined light source of 450 + 660 nm (Table 3). Similar to the results from our study, Jahan et al. (2014) reported that B. tabaci adults were highly attracted to green light source (526 nm) regardless of the biotype, and they were least attracted to the blue light source. Meanwhile, Zheng et al. (2014) reported that the attraction of whitefly (Aleurodicus dispersus) adults to purple (405 nm) LED traps was higher than to blue (460 nm), green (520 nm), and yellow (570 nm) ones. LED source of 405 nm wavelength was not tested in this study and further studies on its effect on B. tabaci will be needed. The intensity of radiation differed with the light source. However, the intensity of radiation and the attraction of B. tabaci adults did not seem to have a positive correlation (Table 3). The effect of LED source on the natural enemies of B. tabaci is shown in Table 4. Two natural enemies, Cyrtopeltis tenuis and Orius laevigatus, were highly attracted to the 520 nm light source. Thus, the light source of 520 nm, which not only attracted B. tabaci but also two of its natural enemies at the same time, could be installed at the entrance of the tomato greenhouse along with planting the selected banker plant, at a time when B. tabaci occurrence begins.

Selection of banker plant

The first study in the use of a banker plant involved growing tomato plants that were pre-inoculated with Encarsia formosa along with the crop to control greenhouse whitefly, T. vaporariorum, in tomato greenhouse (Stacey, 1977). Of the studies undertaken on banker plants, 92% focus on the control of aphids, while less than 15% of the studies focus on the control of whitefly and thrips (Frank, 2010).

Figs. 5, 6 show the results of the density of B. tabaci and its natural enemy, Cyrtopeltis tenuis, when buckwheat was grown along with the tomato in the greenhouse. A high density of C. tenuis was maintained with the buckwheat treatment when compared to treatments where buckwheat was not used. Therefore, planting of buckwheat with C. tenuis in tomato greenhouse will be effective in increasing the control of B. tabaci and the additional input of C. tenuis can be reduced.

Presence of buckwheat and C. tenuis resulted in a 50% reduction in the density of B. tabaci in tomato greenhouse (Fig. 6). It is believed that C. tenuis, which was in the buckwheat moved to the tomato plants and effectively reduced the density of B. tabaci. On the other hand, another natural enemy, O. laevigatus, remained too low in density and eventually disappeared within seven days (Fig. 7). Therefore, it will be necessary to identify suitable banker plants to maintain O. laevigatus.

Integrated push-pull strategy effect

To effectively control B. tabaci in a tomato greenhouse, it is meaningful to integrate our results from testing the various push-pull approaches. The changes in the density of B. tabaci adults under an integrated push-pull strategy were investigated in the tomato greenhouse (Fig. 8). After 50 days of treatment, the density of B. tabaci adults was 250.8 adults per trap, which was three times lower than that when no treatment was applied. Control of the pest infestation increased with time after treatment and showed a maximum value of 68.7%. Until now, many studies on pest control have focused on the reduction of pest density, but it is meaningful that the study also verifies an increase in the yield, the ultimate goal of pest control. Table 5 compares the tomato yield when the integrated push-pull strategy for controlling B. tabaci control was applied with yield obtained when no treatments were used. In the case of non-treatment, the quantity of tomato produced per 990 m² was 3648 kg excluding 20.8% of non-commodities, while that produced after the integrated strategy treatment was 5230 kg, which was 143.4% higher, and the non-commodities rate decreased by about 5%. This result cannot be regarded as an effect exclusively through the control of B. tabaci. However, this strategy can be used as a means to further expand biological control because it has the advantage of maximizing the control effect, sustainability, and minimizing the adverse effects on the environment in the absence of the use of insecticides. Also, efforts must be made to create conducive environmental conditions before applying the push-pull strategy developed in this study to commercial greenhouse, which may incur costs. However, if there are long-term investments supported by further studies to reduce the cost incurred, this could become the best push-pull strategy to control B. tabaci in tomato greenhouses in Korea.