Volatile fragrances associated with flowers mediate host plant alternation of a polyphagous mirid bug

Apolygus lucorum (Hemiptera: Miridae) is an important insect pest of cotton and fruit trees in China. The adults prefer host plants at the flowering stage, and their populations track flowering plants both spatially and temporally. In this study, we examine whether flower preference of its adults is mediated by plant volatiles, and which volatile compositions play an important role in attracting them. In olfactometer tests with 18 key host species, the adults preferred flowering plants over non-flowering plants of each species. Coupled gas chromatography-electroantennography revealed the presence of seven electrophysiologically active compounds from flowering plants. Although the adults responded to all seven synthetic plant volatiles in electroantennography tests, only four (m-xylene, butyl acrylate, butyl propionate and butyl butyrate) elicited positive behavioral responses in Y-tube olfactometer bioassays. The adults were strongly attracted to these four active volatiles in multi-year laboratory and field trials. Our results suggest that these four fragrant volatiles, which are emitted in greater amounts once plants begin to flower, mediate A. lucorum’s preference to flowering host plants. We proved that the use of commonly occurring plant volatiles to recognize a large range of plant species can facilitate host selection and preference of polyphagous insect herbivore.


Results
Behavioral responses of A. lucorum adults to plants. In Y-tube olfactometer assays of 18 plant species, male and female A. lucorum adults preferred odors of flowering plants (FP) versus those of blank controls (CK) or non-flowering plants (NFP) (P < 0.05). Comparison of P values between genders for each plant species showed that males tended to be more responsive than females (CK vs FP: males = 11, females = 7; NFP vs FP: males = 10, females = 8). Additionally, for both genders, the magnitude of responses (as indicated by P values) was usually stronger for trials comparing the blank control versus flowering plants than for trials comparing non-flowering plants versus flowering plants (females: CK vs FP = 12, NFP vs FP = 6; males: CK vs FP = 14, NFP vs FP = 4) ( Table 1).  values generally increased as concentration increased. For some concentrations, EAG responses of males to butyl acrylate, butyl propionate, butyl butyrate, (Z)-3-hexenyl acetate and 3-ethylbenzaldehyde were significantly greater than those of females (P < 0.05) (Fig. 2).
Greenhouse evaluation of different plant volatiles attractive to adult A. lucorum. For the four active synthetic volatiles identified in the behavioral study above (m-xylene, butyl acrylate, butyl  propionate and butyl butyrate), the number of trapped bugs was significantly higher than that of the control in each trial (100 mg/ml: P < 0.001; 300 mg/ml: P < 0.001). The sex ratio (female/male) of A. lucorum adults trapped by these volatiles was 1:1.55, and no significant difference was found between sex ratios of tested individuals (1:1) and trapped individuals (P = 0.107), and between those trapped by different volatiles (P = 0.847). At 100 mg/ml, trap catches for butyl acrylate and butyl propionate were not different from each other, but were each greater than m-xylene. Catches for butyl butyrate were not different from m-xylene, butyl acrylate or butyl propionate. At 300 mg/ml, trap catches for the three esters, butyl acrylate, butyl propionate, and butyl butyrate, were not different from each other, but were significantly greater than for m-xylene. Furthermore, significantly more A. lucorum adults were trapped with 300 mg/ml volatiles than with 100 mg/ml (butyl acrylate: P = 0.002, butyl propionate: P < 0.001, and butyl butyrate: P = 0.001), but not for m-xylene (P = 0.356) (Fig. 4).

Discussion
Results of this study are consistent with previous work that indicates that A. lucorum shows a marked preference for host plants at the flowering stage, and that host plant phenology plays an important role in seasonal dispersal between hosts by A. lucorum in agricultural landscapes 23,25,27 . Dong et al. 26 noted that A. lucorum adults and nymphs had higher survival rates and population fitness when on flowering plant species compared to related non-flowering plants. This positive relationship between adult flower preference and offspring performance further highlights the significance of flowers in seasonal dynamics and populations of the generalist A. lucorum. This phenomenon has also been recorded for other generalist herbivores, such as Lygus bugs 11,28 and lepidopteran insects 29 . Hence, exploring the mechanisms underlying the flower preferences of these generalist herbivores may help researchers to fully understand the population dynamics, interplant movements and the interactions between generalist herbivores and host plants.
Previous studies with other mirid bugs suggest gender-specific responses to volatiles of different chemical classes. Specifically, electrophysiological and behavioral studies suggest that male mirid bugs are usually more sensitive to acetates, either insect-or plant-produced, than female individuals, and that females are more responsive to plant-produced compounds in other classes 16,18,[30][31][32][33] . In the present study, we found that both sexes were responsive to host plant odors but male A. lucorum antennae were more The data on the y-axis are mean ± SE. For each release concentration, significant differences (P < 0.05) between compounds are denoted by different letters (uppercase letters, 300 mg/ml; lowercase letters, 100 mg/ml). *denotes a significant difference between two doses (P < 0.05), and ns shows no difference (P > 0.05). Four replicates were conducted for each dose.
responsive than female antennae to esters, and the electrophysiologically active compounds elicited a behavioural response involving locomotion towards these compounds (butyl acrylate, butyl propionate and butyl butyrate) in the laboratory, greenhouse, and field trials. Our results are consistent with the hypothesis that male mirid bugs are generally more responsive to acetates, while female bugs are more sensitive to other compounds. However, this hypothesis is based on studies of relatively few mirid bug species; thus, additional work on other species is warranted.
Although each plant species shows a unique volatile profile, there is often considerable overlap between species. For example, M. sexta visited flowers, although they differ qualitatively and quantitatively in their scent profiles, emit characteristic compounds, such as methyl benzoate, benzyl alcohol, and benzaldhyde 10 . In this study, each plant species showed a large difference in size and architecture; however, they all emitted at least four volatiles that elicited EAG and positive behavioral responses in adult A. lucorum. These four volatiles were present in significantly greater levels in each plant species during the flowering period compared to non-flowering periods. These data indicated that the increased level of these four volatiles was the major reason underlying the marked preference of A. lucorum adults for flowering host plants. From an evolutionary perspective, the use of commonly occurring plant volatiles to recognize a large range of plant species may also facilitate learning behavior and preference for the host that is more abundant in local agricultural landscapes. The finding supported our hypothesis but was not consistent with the common viewpoint that generalist insects typically utilize many different plant-derived volatiles to find different host plants 4 .
In this study, we conducted a series of experiments on the electrophysiological and behavioral responses of A. lucorum to plant volatiles in order to better understand host recognition and selection by the mirid bug. Our results demonstrate that A. lucorum perceives and orients toward several plant-derived volatiles that are emitted at higher levels when plants are in the flowering stage, and that this relationship plays an important role in host-plant alternation by the mirid bug. Our use of several types of behavioral experiments was a matter of necessity because of the great variability that is sometimes observed between different types of arenas, as well as laboratory and field settings. In particular, our data set from the Y-tube olfactometer study is a good example of why it is valuable to include odors for all choices in an olfactometer assay, i.e., "something vs something" as opposed to an odor compared to a clean air control, i.e., "something vs nothing", as well as following-up with field studies to verify (or refute) the results found in the laboratory 34 . As would be expected, our Y-tube data showed stronger significance, as measured by P values, for the clean air vs flower odor tests, "something vs nothing", than for the vegetative odor vs flower odor tests, "something vs something". The nearly 3-fold difference that we observed reflected the fact that "something vs something" tests require the insect to make a more discerning choice than tests of an odor vs clean air. Reliance solely on the tests of clean air vs flower odors would have presented a more liberal perspective of our results. Thus, behavioral studies may require a variety of approaches that use different strategies in the laboratory (e.g., Y-tube olfactometer) and field to understand the effect of volatile compounds on insect behavior 35 . In particular, in the complex field environment that is characterized by a multitude of potential cues, volatile and otherwise, use of traps baited with volatile compounds may reveal important information on the actual host-plant preferences of phytophagous insects. This powerful assessment tool can also be applied to the further development of attractants of insect pests, which are widely used in pest behavioral manipulation.
Our improved knowledge of A. lucorum-plant volatile interactions can be incorporated into existing management strategies. For example, trap crop (mungbean) and a repellent compound (dimethyl disulfide) have been successfully selected and applied in the management of A. lucorum 36,37 . These studies indicated that behavioral manipulation is a viable option for preventing the damage of A. lucorum. In the present study, the four bio-active plant volatiles that were identified provide a basis for further development of attractants for A. lucorum. Additionally, the combination of attractant and trap crop for enhancing attractiveness 38 , and attractant and repellent for developing push-pull strategy 39,40 must be further assessed. These alternative environmentally-friendly measures are needed for controlling A. lucorum 41 .
Our study provides chemistry-based insights into flower preference by A. lucorum adults, which supplies basic information for further developing behavioral manipulation measures for control. It also lays a solid foundation for exploring a new pattern of host selection and volatile identification of generalist herbivores, which is available for fully understanding the interaction and co-evolution between polyphagous herbivores and their host plants.   Supplementary Table S2 online). All plants were planted on two separate occasions (early May and late June) by direct seeding or transplanting into pots at the Langfang Experiment Station of CAAS. Plots were managed using identical agronomic practices, and no insecticides were used 23 . All plants were caged (80 mesh) to prevent damage by herbivorous insects and mites. Vegetative and flowering periods of each plant species were selected for this study. We sampled plants at the vegetative growth stage ca. 2 weeks before the first emergence of flower buds, and the flowering growth stage was sampled at the time of peak bloom.

Behavioral responses of A. lucorum adults to different plants.
Insect behavioral responses to plant odors are commonly analyzed using a Y-tube olfactometer 43 . The olfactometer had the following specifications: a 3-cm inner diameter clear glass tube with a 15-cm-long central tube and two 15-cm-long lateral arms with a 60° angle at the Y-junction. The apparatus was placed in a 100 × 100 × 60-cm chamber that was illuminated with two 40-W fluorescent lamps (light intensity 2000 lux), maintained at 25 ± 1 °C, 60 ± 5% RH. A vacuum pressure pump (Beijing Institute of Labor Instrument, Beijing, China) pushed air through activated charcoal and an Erlenmeyer flask filled with distilled water. Airflow through each of the olfactometer arms was maintained at 300 ml/min and entered the apparatus via Teflon tubing. When a given test plant was in full bloom, the branches containing the flowers were removed with scissors and immediately introduced into one holding chamber for the olfactometer assay. In a second chamber, we introduced the branches cut from the same plant species at the vegetative period. The branches of some plant species were large enough so that only one was used for each assay; branches of other species were small, which allowed several to be used together. Due to great variability in size and architecture among plant species, the total amount of plant tissue was not standardized between all plant species; instead the same weight of plant tissue was used in the two chambers for each trial. The excised ends of the branches were wrapped with moist cotton towels and enclosed inside a plastic bag before they were placed in the holding chambers. Prior to behavioral bioassays, insects were starved for 4 h. Each insect was used only once. Plant tissue was replaced every hour, and the Y-tube olfactometer was replaced with a clean one after four individuals had been tested. For each treatment, we tested 60 insects per gender. All bioassays were conducted between 08:00 h and 18:00 h.

Collection of plant volatiles.
Headspace sampling was used to collect the volatiles from different plant species 44 . More specifically, once a plant reached the vegetative period or flowering period, pots were moved to the laboratory for volatile collection. The plant material samples were the same as those used in the above olfactometer assay. The selected plant parts were sealed in a polyester cooking bag (unprinted 45 × 55-cm bags, Terinex, Bedford, UK). Clean air was introduced from the lower part of the bag, and allowed out of the top through a volatile collection trap, i.e., a plastic tube containing 40 mg of 80-100 mesh Porapak Type Q adsorbent (Bulk Packing Material, Altech Assoc. USA). The collection of volatiles was replicated six times for a 4 h time period (16:00-20:00 h), which coincides with the period of the greatest activity of A. lucorum adults in the field 45 . Volatiles were eluted with 400 μ l HPLC-grade dichloromethane (Fisher, Fair Lawn, NJ). The solutions were stored at − 20 °C in 1.5 ml glass vials (Agilent, USA).

Coupled gas chromatography-electroantennography detection (GC-EAD).
A coupled GC-EAD system was used, in which the effluent from the GC column was simultaneously directed to both the antennal preparation and the GC detector 44 . The heads of female or male A. lucorum adult were excised using a scalpel, and tips of the antennae were removed to ensure good contact with the electrode. The reference electrode was inserted into the head cavity, and the recording electrode was attached to both antennae. The reference and recording electrodes were silver wires enclosed in drawn glass capillary tubes filled with phosphate buffered saline 18 . Ten coupled runs were completed for each gender × plant species × plant growth stage combination. Only flame ionization detector (FID) peaks that corresponded to an electroantennography (EAG) peak in five or more replicates were considered to show electrophysiological activity.

Identification of electrophysiologically active volatiles. Identification of FID peaks showing elec-
trophysiological activity was performed by positive ion electron impact gas chromatography-mass spectrometry (EI GC-MS) on an HP 6890 GC coupled to an HP 5973 MS detector. One microliter samples were injected at 250 °C onto the HP-5MS column. These volatiles were tentatively identified by comparing mass spectra with those of authentic samples in the NIST 2005 database and were further confirmed by co-injecting the collection of plant volatiles with commercially available authentic standards on both non-polar HP-5MS and polar DB-WAX columns (30 m × 0.25 mm × 0.25 μ m, Agilent Technologies, Palo Alto, CA, USA), with peak enhancement indicating co-elution. Quantification of identified compounds was achieved by injecting authentic standards at four known concentrations onto the HP-5MS column and recording the peak area. Each standard was injected four times, and a calibration curve was plotted for each compound, which was used to determine the amounts of each active volatile from the 18 tested plant species during both flowering and non-flowering (= vegetative) periods. To compare the amounts of each active volatile, six samples were analyzed at flowering and non-flowering stages of each plant species, respectively. Electroantennography. The EAG apparatus (Syntech Ltd., Hilversum, The Netherlands) was linked to a desktop computer with an IDAC-02 data acquisition interface board, on which EAG responses were recorded, stored and quantified. Electrical signals from the antennae were amplified using a high-impedance DC amplifier and a signal connection interface box (Auto Spike, IDAC2/3; Syntech). The antennae of female and male A. lucorum adults were prepared as above, and were used as detectors to identify potential activity of given volatiles. The reference and recording electrodes were prepared according to the GC-EAD methods described above. Seven authentic chemical standards were used for EAG: (Z)-3-hexen-1-ol (98%), m-xylene (99%), butyl acrylate (99%), butyl propionate (99%), butyl butyrate (98%), (Z)-3-hexenyl acetate (97%) and 3-ethylbenzaldehyde (98%) (Sigma Aldrich, Gillingham, UK).
For the EAG trials, serial dilutions (0.01, 0.1, 1.0, 10 and 100 μ M) of each chemical were made using paraffin oil (Sigma Aldrich, UK). Stimulus applicators were prepared by pipetting 10 μ l of the test dilution onto a 5 cm × 0.5 cm strip of filter paper, after which the filter paper was placed inside a 15-cm-long glass Pasteur pipette. Four controls were used: 1) an empty pipette; 2) a pipette containing only a strip of filter paper; 3) a pipette containing 10 μ l of paraffin oil on filter paper; and 4) a pipette containing 10 μ l of standard (100 μ g/μ l paraffin oil-dissolved nonanal, 97%, Sigma Aldrich, UK) on filter paper. EAG recording began 3 min after the antennal preparation and used the following test protocol. The controls were tested in the following order (1,2,3,4), after which low-to-high serial dilutions (0.01, 0.1, 1.0, 10 and 100 μ M) of the given chemical were tested, then the controls 4 and 3. Presentation of controls throughout the recording session permitted standardization of antennal responses and correction for antennal fatigue. Tested chemicals and controls were applied at 0.5-sec pulses, at 30-sec intervals separated by purges of filtered, humidified air. Each antennal preparation only tested one chemical with five dilutions. EAGs were measured as the maximum amplitude of depolarization (mV) and were analyzed using a customized software package (Syntech). For each chemical, 10 A. lucorum adults of each gender were tested.
Behavioral responses of A. lucorum adults to synthetic volatiles. The Y-tube olfactometer was used to test behavioral responses of A. lucorum adults to each of the seven synthetic volatiles. The protocols and methods, as well as the setup of the olfactometer, are described as above. During behavioral trials, a single chemical was diluted in amounts equivalent to those used in the EAG assays. We applied 10 μ l (1 μ g/μ l) of the dilution onto a 5 cm × 0.5 cm piece of filter paper and placed the paper into a flask (10 ml, Beijing Kenin Industrial Biotechnology Co. LTD, Beijing, China). In these experiments, female and male A. lucorum adults were offered a choice between a given chemical and mineral oil, which was used as solvent for the chemical. New filter papers were used for each insect. For each treatment, we tested 60 replicates per gender.
Greenhouse evaluation of A. lucorum attraction. Cage experiments were conducted to determine the level of attractiveness of four active volatiles (i.e., m-xylene, butyl acrylate, butyl propionate and butyl butyrate) selected from the Y-tube olfactometer assays to A. lucorum adults in the greenhouse. These compounds were chosen based on their activity during the Y-tube assays. During the experiments, two pots of cotton plants (20-25-cm height; 8-10 leaves) were placed in the center of a 2 × 2 × 2-m screen cage (80 mesh). A cotton wick soaked in 5% honey water was placed at the center of the cage for feeding. A bucket trap (Pherobio Technology Co., LTD., China; 15 cm in diameter, 12.5 cm in height) was placed in each corner of the cage, and another one was placed in the center of the cage. All five traps were hung 10-15 cm above the plant tops, and four randomly-selected traps were baited with synthetic volatiles (one volatile per trap), with the remaining trap containing the solvent-only control. Synthetic volatiles dissolved in the solvent lanolin or the single solvent were placed in open 1.5-ml glass vials, and fixed to the middle of the trap. Next, 200 A. lucorum adults (1:1 sex ratio) were released into the cage. At 3 d post-release, we counted all trapped individuals and determined their gender. Two experiments using different volatile rates were conducted, (A) 1 ml of 100 mg/ml volatile per trap; and (B) 1 ml of 300 mg/ ml volatile per trap. Each set of experiments was repeated four times. Experiments were conducted at 25 ± 1 °C and 60 ± 5% RH.
Field evaluation of A. lucorum attraction. Field trials were performed to evaluate the behavioral response of A. lucorum adults to four active volatiles (m-xylene, butyl acrylate, butyl propionate and butyl butyrate) at the Langfang Experimental Station of the CAAS, during 2012-2014. Bucket traps were placed in a cotton (Gossypium hirsutum) field in August of 2012-2013 and in a weed (mainly Artemisia argyi) field in early-September of 2014, at a height of 1.8 m above the ground (i.e., ca. 20 cm above at the plant canopy level). At this height, A. lucorum was found to most abundant and active 45 . Within the field, traps were spaced at least 15 m apart, arranged randomly and oriented uniformly in an east-west fashion in the field. Synthetic volatiles dissolved in the solvent lanolin (1 ml, 100 μ g/μ l) or the single solvent (i.e., as a control) were placed in open 1.5-ml glass vials, and stuck to the middle of sticky card. Traps were Scientific RepoRts | 5:14805 | DOi: 10.1038/srep14805 collected 3 d later, and trapped A. lucorum adults of each gender were recorded. Each volatile treatment was replicated 12, 6 and 8 times (one trap for each replicate) in 2012, 2013 and 2014, respectively.

Statistical analyses.
For Y-tube olfactometer bioassays, the null hypothesis that A. lucorum showed no preference for either olfactometer arm (i.e., 50:50 response) was analyzed using χ 2 goodness-of-fit test. The individuals that remained unresponsive were not included in the analysis. EAG amplitudes (mean ± SE) were control-adjusted and presented as responses relative to the standard, which was 100 μ M nonanal. Trap data were log-transformed (n + 1) to fit the assumption of homogeneity of variance for analysis of variance (ANOVA). Means were compared using ANOVAs followed by Tukey's honestly significant differences (HSD) test, whereas the Kruskal-Wallis test was conducted on data sets that were not normally distributed. The difference in sex ratio between trapped and tested individuals in greenhouse evaluation and between trapped and field-collected individuals in the field trial was analyzed using χ 2 goodness-of-fit test. Differences in volatile amounts between flowering and non-flowering plants were determined using analyses of variance (ANOVA) followed by Tukey's HSD test. SAS statistical software package was used for data analysis 46 .