Innate positive chemotaxis to pollen from crops and banker plants in predaceous biological control agents: towards new field lures?

Predator-prey interactions form the core of biological control of arthropod pests. Which tools can be used to monitor and collect carnivorous arthropods in natural habitats and targeted crops? Eco-friendly and effective field lures are urgently needed. In this research, we carried out olfactometer experiments assess innate positive chemotaxis to pollen of seven crop and banker plant by two important predatory biological control agents: the coccinellid Propylea japonica (Thunberg) and the anthocorid Orius sauteri (Poppius). We compared the attractiveness of pollens from crops and banker plants to that of common prey homogenates (aphids and thrips, respectively). Attractiveness of the tested odor sources was checked via field trapping experiments conducted in organic apple orchards and by release-recapture assays in organic greenhouse tomato crops. Maize and canola pollen were attractive to both P. japonica and O. sauteri, in laboratory and field assays. P. japonica was highly attracted by balm mint pollen, whereas O. sauteri was attracted by alfalfa pollen. Our results encourage the use of pollen from crops and banker plants as low-cost and eco-friendly attractors to enhance the monitoring and attraction of arthropod predators in biological control programs.

Olfactory cues produced by herbivore-free plants, such as volatile organic compounds from pollen and nectar also play a role in predatory arthropod attraction 33 . This phenomenon has been studied in many parasitoid species 37,38 , however little knowledge is available about the attractiveness of floral origin olfactory cues to predatory species [39][40][41] . In addition, a number of flowering plant species support predator arthropod populations by providing supplementary food (e.g. nectar and pollen) 37,40 , shelter for mating activity 9,42,43 and protection from pesticides [44][45][46] . Food web enhancements using banker plants may help limit pest populations by re-attracting natural enemies, and diminishing the cost [47][48][49] .
Recent research on the harlequin ladybird, Harmonia axyridis (Pallas) showed that individuals tend to aggregate in close proximity to plant volatiles in their natural habitat [50][51][52][53][54] , and that abundant ladybird populations can be found in alfalfa and white clover fields during the entire growing season, seeking aphids or other Homoptera species 55,56 . Preliminary our observations have indicated that pollen of different crops (e.g. alfalfa, maize, and canola) attracts a number of predatory ladybirds and flower bugs during the early spring. Conserved native grassland flora located nearby orchards has been associated with increased predator biodiversity and enhanced biological control effectiveness 9,57,58 .
Based on these findings, we hypothesized that substances such as pollen from crops and banker plants will attract predator. We performed olfactometer experiments to assess innate positive chemotaxis toward seven crop and banker plant pollens by two important predators, the coccinellid Propylea japonica (Thunberg) and the anthocorid Orius sauteri (Poppius). In their native regions, both P. japonica and O. sauteri help to suppress outbreaks of various arthropod pests, such as aphids, thrips, and spider mites in orchards, fields and greenhouses [59][60][61][62][63] . We compared the attractiveness of different pollens from crops and banker plants to that of prey homogenates of P. japonica and O. sauteri (i.e. aphids and thrips, respectively). Attractiveness of the tested odor sources under field conditions was studied by field trapping experiments conducted in organic apple orchards and release-recapture assays in organic greenhouse tomato crops. Olfactometer. A custom-made ten-arm olfactometer (Jingyi, Beijing, China) was used to evaluate innate -chemotaxis of P. japonica and O. sauteri adults to odor sources of plant or insect origin. The olfactometer is described in Fig. 1. The entrance hole (diameter 3.0 cm) was located in the center of an insect release chamber (diameter = 45.0 cm, height = 1.5 cm), which was connected to 10 odor-source arms made of glass tubes (diameter = 0.5 cm, length = 3.0 cm). Air flow into each arm passed through a series of four chambers, included a distilled water filter chamber, an activated carbon filter chamber, an attractor chamber and a sampling chamber (for collecting attracted predators) ( Fig. 1). At the sample movement chamber, the airflow meter was connected to the sampling chamber in each odor-source arm. An air pump (Sunny Industry, AirPower 6-CL, Beijing, China) was connected to the insect release chamber and set for constant airflow of 1,500 ml min −164 . To avoid any bias from surroundings, the 10-arm olfactometer was placed inside a white plastic box (120 × 120 × 80 cm) illuminated by four white, overhead LED white lights (800 lux each). Preliminary trials using empty odor-sources did not show directional bias for either predator (P. japonica χ 2 = 7.131, P = 0.117; O. sauteri: χ 2 = 8.379, P = 0.1196).

Predatory biocontrol agents. From
Pollen collection and insect-borne odor-sources. Nine odor sources were evaluated for innate positive chemotaxis in P. japonica and O. sauteri. We tested seven floral-borne odor sources: canola pollen (Brassica campestris L. cv. Jingyou-1), maize pollen (Zea mays L. cv. Denghai-605), rose pollen (Rosa chinensis Jacq.), amaranth pollen (Myosotis sylvatica L.), white clover pollen (Trifolium repens L.), balm mint pollen (Mentha haplocalyx Briq.) and alfalfa pollen (Medicago sativa L.). The pollens were produced and purified at Zhongnong Company from China academy of agricultural science. All pollens were collected daily by shaking the flower in a plastic bag. The collected pollen was air-dried at room temperature for 48 h and subsequently passed through a screen to remove anthers and contaminants. Pollen collected from each flower was pooled and stored at − 80 °C until used.
The two insect-borne odor-sources were homogenates of 3 rd instar nymphs of M. japonica and F. occidentalis, common preys of P. japonica and O. sauteri, respectively. Control odor source was clean air.
Olfactometer experiments. In olfactometer experiments, 65 P. japonica female adults or 65 O. sauteri female adults were gently transferred into the release chamber before starved 12 hours. The nine odor sources, balanced for weight (50.0 mg/odor source) were placed into attractor chambers positioned at the end of each odor arm, with the last attractor chamber empty after 30 min, the insects found in the sampling chamber of each arm were counted. The test was replicated 10 times with each predator. In each replicate, odor sources were randomized and all olfactometer arms were rotated by 36° to avoid experimental bias. After each replicate, olfactometer components were cleaned by the soap-water rinse hexane method described in Benelli et al. 34 .
Field attraction assays in organic apple orchard. We used the sampling trap described by Wang et al. 52 to evaluate the attraction of P. japonica and O. sauteri toplant pollens (canola, maize, rose corolla, amaranth corolla, white clover, balm mint and alfalfa) in organic apple orchards (a natural habitat for both species). Prey homogenates of M. japonica and F. occidentalis were also tested in the orchard. A potential attractor, balanced for weight (100 g/trap), was placed into each sampling trap. Moist cotton (100 g) served as control 52 . The field survey was carried out in an organic apple orchard (WANG J-Y organic apple Co. Ltd. N40°10′ , E116°2′ , WANG J-Y village, Changping, Beijing) during August and September 2012. For each predator species, the test was repeated 10 times. In each replicate, field attraction was evaluated by placing 3 traps per treatment in the orchard. Sampling traps for each attractor or control were randomly distributed in the orchard at a height of 1.5 m above the ground. The minimum distance separating traps is 12 m that not to interfere with each other. After 24 h, for each species, the number of trapped adults was recorded.
Release -recapture assays in organic greenhouse tomato crops. These experiments were carried out from May to July 2013 in organic tomato greenhouses (7500 cm × 600 cm each) at NOYA Organic Farm (N40°108′ , E116°99′ ) in Pinggu County, Beijing. Initially, 1500 tomato plants, each with 6 main leaves, were transplanted into each greenhouse. After 5 days, over 500,000 spider mite adults, Tetranychus cinabarinus were released into the greenhouse. Eleven days later, randomly selected unmated P. japonica (35 adults) or O. sauteri (90 adults) were released in the greenhouse, when the spider mites get even distribution. We used the same sampling traps described above for our field attraction assays in organic apple orchards 52 . Ten sampling traps baited with different attractors (100 g each, control: moistened

Data analysis
Statistical analyses from both data form olfactometer and field tests were based on a log-linear model for inter-comparison group difference, in difference with the method described by Turling et al. 65 , Tamo et al. 66 and Davison & Ricard 67,68 . We used on a stochastic model developed from Tamo et al. 66 , to allowing for data on the multinomial distribution. The number of predators choosing the ith arm (i = 1,…,10) indicates the relative attractiveness of the corresponding odor source, which is parameterized by λ i . If only the odor source affects predator's choice, and this effect is the same for all predators, the corresponding model is where β p measures the attractive of either an empty arm or of the odor source in the ith arm. The ration of residual deviance to degree of freedom was less than 1, indicated therefore no data over dispersion.

Results
Innate positive chemotaxis in the olfactometer. There were significant differences in the attractiveness of crop and banker plant pollens for predatory. In P. japonica experiments, canola pollen, maize pollen and balm mint pollen were more attractive compared to other treatments (P < 0.01). Rose corolla pollen, amaranth corolla pollen, white clover pollen and alfalfa pollen where also more attractive than the control (Fig. 2A). In O. sauteri experiments, canola pollen, maize pollen and alfalfa pollen were more attractive compared to other treatments (P < 0.01). Rose corolla pollen, amaranth corolla pollen, white clover pollen and balm mint pollen were more attractive than the control (Fig. 2B). Prey homogenate odors were not attractive.
Field attraction in organic apple orchards. Field attraction assays confirmed the behavioral responses seen in olfactometer experiments (Fig. 3). Balm mint pollen and maize pollen attracted more P. japonica than other treatments (P < 0.01), while rose corolla pollen, amaranth corolla pollen, white clover pollen and alfalfa pollen were more attractive than the control (Fig. 3A). Alfalfa pollen, maize pollen and canola pollen attracted more O. sauteri adults over other treatments (P < 0.01). Rose corolla pollen, amaranth corolla pollen, white clover pollen and balm mint pollen were more attractive than the control (Fig. 3B). Attraction to prey homogenates did not differ from the control.
Release-recapture in organic greenhouse tomato crops. Release-recapture assays carried out with P. japonica and O. sauteri in greenhouse tomato crops showed results comparable to our previous experiments (Fig. 4). In P. japonica trials, balm mint pollen was the best lure, followed by maize pollen, canola pollen and rose corolla pollen (P < 0.001) (Fig. 4A). P. japonica attraction to amaranth pollen, aphid or thrip homogenates was not different from the control (Fig. 4A). In O. sauteri trials, maize pollen was the most effective lure, followed by canola pollen and alfalfa pollen (P < 0.001) (Fig. 4B). Rose corolla pollen, amaranth corolla pollen and white clover showed higher attraction than control or the prey homogenates (Fig. 4B). The majority of P. japonica (82%) and O. sauteri (70%) were recaptured.

Discussion
For decades, researchers have sought effective lures to attract entomophagous insects from natural habitats into crops, since this could help reduce the need for natural enemy augmentation that relies on artificial mass rearing 8,20,52 . Traps enhanced with low-cost lures within an agro-ecosystem may help to monitor the population dynamics of released predatory insects, and could lead to decreased ecological risks resulting from their over-aggregation. However, experimental evidence of these dynamics for predaceous biological control agents is hard to find. Furthermore, mass rearing of natural enemies are not only high cost in China, but also faced quality control challenges 69 . However, inundative and seasonal inoculative releases of natural enemies are mass applied in greenhouse. The number of natural enemies available has increased dramatically in future. These programs benefit from tools that can be used to attract, collect and monitor predatory arthropods in natural habitats and targeted crops.
Crop and banker plant pollens have been investigated as alternative foods for predatory arthropods [39][40][41] . Although many studies about available on the cues guiding predatory biocontrol species towards these food sources, our results point out the important role for olfactory stimuli. We observed that crop and banker plant pollens are able to induce positive chemotaxis by Coccinellidae and Anthocoridae species in laboratory conditions, field orchard assays and release-recapture greenhouse experiments. On the other hand, crushed preys were poorly attractive to predaceous arthropods, this may be due to the fact that predators may have adapted to easier detect plant-borne olfactory cues associated with the herbivore's presence rather than to the cues from the herbivore itself.
Interestingly, maize and canola pollen show good attraction rates for P. japonica and O. sauteri, in both laboratory and field assays. P. japonica is also attracted by balm mint pollen, while O. sauteri is attracted by alfalfa pollen 45,[70][71][72][73] . We found that maize pollen is more attractive than canola pollen for both P. japonica and O. sauteri. This could be due to the high sugar content of maize pollen, which far exceeds that of canola pollen 38 . Indeed, sugar-rich food can enhance flight and the host foraging ability of predatory insects 74 , and may positively affect lifespan 73,75,76 . In the field, many species of predators can be less abundant in nectar-less cotton fields than in nectared ones 77 . Previous research has shown that, in addition to sugars, VOC profiles from corn plants contain complex odor components including green leaf alcohols, aldehydes, derivate esters, terpenes, and sesquiterpenes 78 . Predatory insects may use these compounds as cues when foraging for prey 79,80 . For example, the 12-spotted ladybird Coleomegilla maculata (DeGeer) and the green lacewing Chrysoperla carnea (Stephens) showed electroantennographic responses to VOCs from corn pollen and leaves 81 . Also, balm mint plants are attractive to a number of herbivorous arthropods, including aphids, whiteflies and thrips 82 . Electrophysiological assays are required to shed light on which compounds are the principal sources of this attractiveness. Alfalfa pollen has been observed as one of the best lures for O. sauteri. Alfalfa has been already tested as an intercrop in cotton fields to aggregate predatory arthropods. When compared with pure cotton cultivation, the species richness index and diversity of natural enemies is often higher in intercropped fields 83 . At the time of alfalfa cutting, ladybirds, spiders and lacewings may move to cotton plants, helping to control outbreaks of Aphis gossypii Glover 84 . In addition, alfalfa fields frequently provide a favorable environment for the reproduction of aphidophagous coccinellids (e.g. Coccinella septempunctata L.) 85 and flower bugs 83 . In the greenhouse release-recapture experiments, pollen attractiveness was lower compared to the results in the olfactometer and field assays in apple orchards. This may be due to other factors which affect the predator's searching for non-prey foods. For instance, if greenhouse conditions presented higher prey densities than the other two experiment; this may deter the predators from utilizing non-prey food sources 41 . However, balm mint and maize pollen can still be proposed as good lures for P. japonica, and O. sauteri, respectively in greenhouses, since our results show positive chemotaxis to these two lures.
Overall, this research suggests that pollen from four plant species can serve as an eco-friendly and effective field lure for predatory biological control agents, both in orchard trapping surveys and in greenhouse release-reattraction assays. Maize, canola and balm mint pollens can be used for the coccinellid P. japonica, while maize, canola and alfalfa pollens are attractive to the flower bug O. sauteri. These results encourage the use of pollen from crops and banker plants as low-cost and eco-friendly attractors to enhance monitoring and attraction strategies in predatory arthropod-based biocontrol programs. The appropriate placement of suitably designed and improved sampling traps may prove to be a viable tactic for collecting predators.