Lycopersicon esculentum (variety “No. 9 Zhongza”) tomato plants were grown in plastic pots (7 cm diameter) containing a mixture of peat moss, vermiculite, organic fertilizer, and perlite in a 10:10:10:1 ratio by volume (Shi et al., 2013). Tomato plants with three to four expanded leaves were inoculated with a TYLCV infectious clone kindly provided by Professor Xueping Zhou (Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China); infection by TYLCV was confirmed as previously described (Ghanim et al., 2007; Shi et al., 2013; Liu et al., 2014). TYLCV-free plants were not inoculated with the virus. Plants with six to eight fully expanded leaves were used for the assays and experiments described in the following sections.

Laboratory colonies of B. tabaci biotype Q and B, whose identities was periodically confirmed based on the cleavage amplified polymorphic sequence (CAPS) and mitochondrial cytochrome oxidase I genes (mtCOI) (Chu et al., 2010a), were separately maintained on virus-free tomato plants in insect-proof cages in a greenhouse. Encarsia formosa was provided by the Beneficial Insects Research Center, Shandong Academy of Agricultural Sciences in China; the parasitoid had been reared on nymphs of Trialeurodes vaporariorum (Westwood) for >10 years.

Selection and oviposition of E. formosa on HQ vs. HB or VQ vs. VB were compared under in vitro conditions in the laboratory. HQ and HB refer to virus-free plants infested with 3rd-instar nymphs of B. tabaci Q or B. VQ and VB refer to virus-infected plants infested with 3rd-instar nymphs of B. tabaci Q or B. TYLCV-infected or TYLCV-free tomato leaves were infested with 30 3rd-instar nymphs of B. tabaci biotype Q or B, and in vitro parasitism was assessed as previously described (Liu et al., 2016). In brief, the leaves infested by 3rd-instar nymphs of B. tabaci biotype Q or B were detached from the TYLCV-free or TYLCV-infected plants. The leaf petioles were wrapped in cotton wool saturated with distilled water. Each treatment pair (HQ vs. HB and VQ vs. VB) was placed in plastic Petri dishes (diameter 8.5 cm, height 1.5 cm, two leaves per dish). A filter paper disk saturated with distilled water was also placed beneath each leaf. Newly emerged E. formosa females (1 day old; wasps: nymphs = 1: 30) were subsequently released in the center of each Petri dishes between the two leaves and were carefully removed after 24 h. About 8–10 days later, the number of nymphs parasitized by wasps was recorded. Each comparison was carried out for 19 replicates.

To confirm the results obtained from the in vitro experiment, a greenhouse experiment that simulated field conditions was carried out as described by Zhang et al. (2013b). Bemisia tabaci biotype Q or B adults were placed in clip-cages on the leaves of TYLCV-infected or TYLCV-free plants (four clip-cages per plant, 20 adults per clip-cage) and allowed to oviposit; after 24 h, all of the adults were removed. About 20 days later, when most eggs had hatched and developed into 3rd-instar nymphs, one plant of each treatment pair (HQ vs. HB and VQ vs.VB) was placed in a ventilated cage (88 × 60 × 79 cm); the two plants in each cage were 50 cm apart. Newly emerged E. formosa females (1 day old; wasps: nymphs = 1: 30) were subsequently released in the center of each cage between the two plants. After 24 h, we carefully checked the tomato leaves and removed the parasitoids by using a small suction trap. The viruliferous and TYLCV-free tomato plants carrying nymphs were separately maintained in a climate chamber at 26 ± 1°C with 70 ± 5% relative humidity and a 16 h/8 h light/dark photoperiod. About 8–10 days later, the number of nymphs parasitized by wasps was recorded. Each comparison was replicated four times, and replications were conducted on different days to account for day-to-day variation (Zhang et al., 2013b).

A Y-tube olfactometer was used to assess the behavioral responses of E. formosa to the following pairs of odor sources: TYLCV-free tomato plants infested with 3rd-instar nymphs of B. tabaci biotype Q (HQ) or B (HB); TYLCV-infected tomato plants infested with 3rd-instar nymphs of B. tabaci biotype Q (VQ) or B (VB). The following combinations were also carried out to verify the absence of differences within each treatment: TYLCV-infected tomato plants without B. tabaci infestation (V) vs. V; TYLCV-free tomato plants without B. tabaci infestation (H) vs. H; VQ vs. VQ; VB vs.VB; HQ vs. HQ; and HB vs. HB. In all cases involving infestation, plants were infested with approximately 20 nymphs per leaf.

The olfactometer assays were carried out as previously described (Zhang et al., 2013b). In brief, an adult female of E. formosa (1.0–2.0 days old) was released at the base of the Y-tube and then observed continuously for 5 min. Each female was tested only once. The selection of the two odor sources was recorded when the female had moved into one arm of the Y-tube for at least one-third of the arm's length and remained for at least 15 s; a female that walked to the far end of the arm was also recorded as a selection. If the female did not choose within 5 min, “no choice” was recorded. After five females were tested, odor sources were interchanged to avoid any influence of asymmetries in the set-up system. The Y-tube was replaced with a new one after it had been used to test 10 wasps. To avoid the influence of residual odor, each used tube was rinsed with 75% alcohol and kept in an oven at 180°C overnight. Each comparison was replicated three times, with 20 females per replicate.

Volatiles emitted from TYLCV-infected or TYLCV-free tomato plants infested by nymphs of B. tabaci biotype Q or B were collected and analyzed as described previously (Wei et al., 2007; Chen et al., 2017). The potted plants, which were carefully wrapped in aluminum foil with only the above-ground green part exposed, were placed in collection bags (one plant per bag) with a gas inlet and a gas outlet. Purified and humidified air at 300 mL min⁻¹ was pumped into each bag through the inlet. A glass tube filled with 0.1 g of PoraPak Q (80/100-mesh; Waters, USA) was used to trap plant volatiles for 4 h at the outlet under continuous high-intensity sodium-halide light. Volatiles were collected from four replicate plants per treatment.

The trapped volatile samples were eluted from the PoraPak Q with 800 μL of high-performance liquid chromatography (HPLC)-grade methylene chloride (Tedia Company, Fairfield, Ohio, USA); an internal standard (10 μL of 20 ng/μL of n-octane) was added to each sample for quantification of relative compound amounts. A 1-μL sample of the solution was subjected to gas chromatography–mass spectrometry (GC-MS); the gas chromatograph (GC-2010 Shimadzu, Japan) was equipped with an Agilent Technologies capillary column DB-5MS (30 m × 0.25 mm ID × 0.25 μm film thickness). The temperature program was as follows: the initial oven temperature was kept at 40°C for 4 min and was then increased to 200°C at a programmed rate of 5°C min⁻¹, followed by a rate at 20°C min⁻¹ to 280°C. The injector temperature was maintained at 250°C with a constant flow rate of 1.0 mL min⁻¹. Compounds were identified by comparison of retention times and mass spectra (NIST database and synthetic standards). The peak area of the volatile expressed as a percentage of the peak area of the internal standard was used to analysis the change of volatiles (Chen et al., 2017).

A Y-tube olfactometer was used to assess the attractions of candidate volatile compounds to E. formosa females as previously described (Zhang et al., 2013b). The candidate compounds (β-Myrcene, β-Ocimene, β-Caryophyllene, α-Humulene, and β-Elemene) were selected based on data obtained with TYLCV-infected tomato plants infested with B. tabaci nymphs (see Results). Nine pairs of treatments were compared: Paraffin oil (CK) vs. CK, β-Myrcene (4.350 ng/μl) vs. CK, β-Ocimene (3.525 ng/μl) vs. CK, β-Caryophyllene (3.350 ng/μl) vs. CK, α-Humulene (1.625 ng/μl) vs. CK, β-Elemene (0.675 ng/μl) vs. CK, and mixture of monoterpenes (β-Myrcene and β-Ocimene) vs. CK, mixture of sequiterpenes (β-Caryophyllene, α-Humulene, and β-Elemene) vs. CK, mixture of total candidate compounds vs. CK. As indicated, paraffin oil (CK) was used as the solvent. Two streams of purified and humidified air at 300 ml min⁻¹ were separately passed through two glass containers containing the test volatile in paraffin oil or containing only paraffin oil as a control and into the olfactometer arms (Chen et al., 2017). The experimental method and data collection were the same as described earlier for the olfactometer assays.

Chi-squared tests were used to analyze the Y-tube olfactometer assays (significance level: P < 0.05) (Martini et al., 2014; Mauck et al., 2015). Parasitoids that did not make a choice were excluded from the analysis. Independent-sample t-tests were used to compare the percentage of B. tabaci nymphs parasitized by E. formosa in each treatment pair and the quantities of volatile compounds collected from TYLCV-infected or TYLCV-free tomato plants infested with nymphs of B. tabaci biotype Q or B. All proportional data were first checked for normality and were transformed when necessary to meet the assumption of normal distribution before analysis. SPSS version 19.0 (SPSS Inc., Chicago, IL, USA) was used for all statistical analysis (Liu et al., 2016).

On tomato leaves in the laboratory, parasitism of 3rd-instar nymphs of B. tabaci by E. formosa females did not differ between biotype Q and B on TYLCV-free tomato plants (P = 0.834, Figure 1A). On TYLCV-infected tomato plants, however, parasitism was significantly higher for biotype Q than B (P = 0.031, Figure 1B).

The results of the semi-natural experiment showed the same trends as that in the laboratory experiment. The E. formosa did not show a preference for Q or B biotype on TYLCV-free tomato plants (P = 0.797, Figure 1C) but preferred biotype Q to biotype B on TYLCV-infected tomato plants (P = 0.027, Figure 1D).

When tomato plants were not infected with TYLCV in Y-tube olfactometer assays, the naive E. formosa females showed no significant preference for plants infested with 3rd-instar nymphs of biotype Q vs. biotype B of B. tabaci (P = 0.424, Figure 2). When tomato plants were infected by TYLCV, however, E. formosa significantly preferred tomato plants infested with biotype Q rather than biotype B (P = 0.046, Figure 2).

According to the analysis of the plant headspace volatiles by gas chromatography-mass spectrometry (GC-MS), all treatments produced the same 32 compounds, including monoterpenes, sequiterpenes, aldehydes, alcohols, and esters (Table 1), and differences between treatments were quantitative rather than qualitative. Quantitative analysis revealed that emissions of the monoterpenes including β-Myrcene and β-Ocimene, and the sesquiterpenes including β-Caryophyllene, α-Humulene, and β-Elemene were higher from TYLCV-infected plants infested with 3rd-instar nymphs of B. tabaci Q than with nymphs of B. tabaci B. The amounts of these compounds emitted from TYLCV-free plants did not significantly differ when the plants were infested with 3rd-instar nymphs of B. tabaci Q vs. B (Table 1).

Results from olfactory test indicated that the mixture of monoterpenes, mixture of sesquiterpenes and mixture of all synthetic volatiles were all strongly attractive for E. formosa. In the meanwhile, synthetic β-Myrcene, β-Ocimene, β-Caryophyllene, and α-Humulene attracted E. formosa females (Figure 3). β-Elemene did not attract significantly E. formosa females (Figure 3).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.