Tomato plant [Solanum lycopersicum L., cv. ‘Gan Liang Mao Fen 802 F₁’ seeds (Xian Qinshu Seeds Company Limited, Xian, Shaanxi, China)] were purchased from a local market. The seeds were sown in small plastic pots (7 cm × 7 cm × 8 cm) containing perlite as a plant-growing medium. The plants were grown inside insectaries where environmental conditions maintained at 25 ± 1°C with 60 ± 5% RH and 16:8 h L:D at a light intensity of 1400–1725 lux. All plants were nurtured in insect-proof mesh cages (60 cm × 60 cm × 60 cm).

We used four levels of nitrogen for growing tomato plants namely, 5 mM (T₁), 10 mM (T₂), 15 mM (T₃), and 20 mM (T₄) nitrogen where 5 mM was considered as a below normal level, 10 mM a normal level, and 15 mM and 20 mM were as a high level. Wahle and Masiunas (2003) and Wang et al. (2007) reported that the growth and yield of tomato was highest at near about 10 mM nitrogen levels but the tomato growth rate was limited with below 5 mM of nitrogen solution. Nutrient solutions for growing tomato plants were prepared with 5 mM nitrogen from a commercial fertilizer, ‘Kang Pu Jin’ with 20-20-20 of N-P₂O₅-K₂O + Mg + TE (COMPO Expert GmbH, Krefeld, Germany), which was used as a standard control in all treatments. For other three treatments, i.e., T₂, T_3, and T₄, the source of additional nitrogen was urea (Sigma U 1250, purity > 99.5%). The nutrient combination of above mentioned fertilizer was N-P₂O₅-K₂O (20%-20%-20%), Mg (0.3%) and other necessary trace elements, including S (0.8%), B (0.01%), chelated Cu (0.04%), chelated Fe (0.1%), chelated Mn (0.1%), chelated Zn (0.04%), and Mo (0.003%). The pH of the nutrient solutions was corrected at 6.0–6.5 if needed to add concentrated HCl or NaOH. The nutrient solutions were prepared with de-ionized water (Millipore, Rios^TM 5).

Different levels of nitrogen containing fertilizer solutions were applied to tomato plants following the technique of Stout et al. (1998) with slight modification. Tomato seeds were sown in the small plastic pots (7 cm × 7 cm × 8 cm) containing perlite as a plant-growing medium and watered with de-ionized water (Millipore, Rios^TM 5) up to 8 days after seed sowing. The plants were randomly assigned as T₁, T₂, T_3, and T₄ treatments for 5, 10, 15, and 20 mM nitrogen, respectively. From 9th to 14th day, all plants were fertilized daily with 20 ml of 5 mM nitrogen prepared nutrient solution to avoid severe changes in the level of nitrogen applied to a plant. From days 15 to 20, all plants were fertilized daily with 20 ml of 10 mM nitrogen prepared nutrient solution except T₁ treatment, which were fertilized daily with 5 mM nitrogen prepared nutrient solution. From days 21 to completion of the experiment, T₃ and T₄ treatments were fertilized daily with 20 ml of 15 mM and 20 mM nitrogen prepared nutrient solutions, respectively. At that time, T₁ and T₂ treatments were fertilized daily with 20 ml of 5 mM and 10 mM nitrogen prepared nutrient solutions, respectively, as followed in previous fertilization. To avoid water stress, additional de-ionized water was added as necessary. Thirty-five days old intact tomato plants were used for all experiments, i.e., wind tunnel and Y-tube bioassays, determination of total nitrogen and volatile compounds containing four to five fully expanded leaves (Supplementary Figure S1).

The whiteflies, B. tabaci MEAM1 (Frohlich et al., 1999), were mass-reared in walk-in insectaries in large screen cages (60 cm × 60 cm × 60 cm) on eggplant Solanum melongena L. (Solanaceae) cv. ‘Zichangqie’ which was grown in 15 cm diameter plastic pots containing 5:1:1 by volume of peat moss, perlite and vermiculite. The insectaries were maintained at 25 ± 1°C with 60 ± 5% RH and a 16:8 h L:D at a light intensity of 1400–1725 lux, which was similar to the experimental environments in which tomato plants were grown.

Response of adult B. tabaci toward different levels of nitrogen applied tomato plants was tested in a plexiglass wind tunnel (200 cm × 70 cm × 70 cm) which was set up in a small controlled environmental room (temperature 25 ± 1°C, RH 60 ± 5%). A blower pulled the air, and the airflow rate was adjusted at 22 cm s⁻¹. Three 18 W fluorescent lamps were set up at the take-off point in the flight chamber. The plastic pot except the plant was covered with polythene bags (EasyOne Oven Bags, Reynolds Kitchens, Lake Forest, Illinois, USA) and aluminum foil to minimize plant-growing media produced volatiles before using the plants. In this test, at a time two different nitrogen doses applied intact tomato plants were placed next to each other on the upwind end of the arena maintaining 30 cm distance to observe that the male or female B. tabaci would fly upwind in the presence of both visual and volatile plant cues. A petri dish containing 100 male or female B. tabaci was kept inside the wind tunnel maintaining a short-distance (1-m) from the two plants. Thereafter, B. tabaci were released there to make a choice their suitable host. The B. tabaci were released at 12:00 h, and 24 h later, the individuals were carefully counted to investigate their preference between two nitrogen treatments in the presence of both visual and olfactory cues. Numbers of whitefly adults were counted 24 h later because it took the adults some time to make a choice and settle down in the wind tunnel against wind flow. Before mass releasing of the whiteflies, five individuals were pre-released to ensure that the mass release does not affect the responses of the whiteflies to the odor sources. Each experiment was conducted three times. The wind tunnel was wiped thoroughly with 70% ethanol and air was pulled through the wind tunnel for at least 2 h before it was re-used.

A Y-tube olfactometer was used to evaluate the behavioral response of B. tabaci to nitrogen applied tomato plant volatile compounds, following the procedure as previously described by Akol et al. (2003) and Saad et al. (2015) with some modifications. The transparent glass made Y-tube olfactometer consisted of 8 cm long base with 0.8 cm internal diameter, two 8 cm lateral branches at a 60° angle from each other. The lateral branches were individually connected to two 3 L glass container with odorless tubes and each glass container contained volatiles releasing one intact tomato plant from each nitrogen treatments. The plant was kept in the glass container for 1 h before sampling for exiting impure volatiles from the system. Plant containing glass containers were covered with thick paper so that B. tabaci individual cannot receive visual cues from the plants. Charcoal filtered humidified and purified air was provided at 100 ml min⁻¹ to both branches of the Y-tube via odor sources using a vacuum pump (Beijing BCHP Analytical Technology Institute, China) for circulating the volatile organic compounds. The air flow was adjusted and measured by an inline flowmeter (LZB-3WB, Changzhou, Jiangsu, China). Male and female B. tabaci were tested separately. Adult whiteflies were starved for 2 h and then released within 0.5–1.0 cm of the base of the Y-tube with a small PCR tube and their responses assessed for 10 min. A B. tabaci adult that walked into one of the lateral branches of the Y-tube at least 5 cm, stayed a minimum of half a minute and did not return was considered as a positive, responsive individual. If an individual did not make a decision within 10 min, they were excluded from the results, and considered as non-responsive insect. To eliminate lighting bias, a 20 W fluorescent light was placed vertically 0.5 m over the Y-tube olfactometer. The positions of two lateral branches of the Y-tube were inverted 180° after every five insects tested. Each olfactory test was repeated two times for each combination of stimulus pairs, and each replicate consisted of 25 B. tabaci adults tested individually, i.e., a total of 50 B. tabaci adults for each treatment were assayed. The experiment was carried out between 12.00 and 16.00 h in a controlled environment maintained at 25 ± 1°C and 60 ± 5% relative humidity. To minimize plant-growing media produced volatiles, the pot except the plant was covered with the same polythene bags and aluminum foil as described previously. The experimental Y-tube and all glassware were washed with soap with tap water, then distilled water, and finally sterilized with 70% ethyl alcohol before oven drying (120°C for 3 h) to reduce the contamination risk of previous tested odors.

Tomato plants (leaves with stem) were taken separately from each treatment and oven dried at 65°C for 4 days. The dried plant samples were ground with a mortar and a pestle, taken in poly bag and preserved it at room temperature before analysis. Total nitrogen from tomato plant tissues was measured using the Kjeldahl method. Exactly 0.1 g sample was digested with 4 ml of H₂SO₄ at 420°C for 1 h along with catalyst of K₂SO₄ and CuSO₄ at a ratio of 9:1. FOSS 8400 automatic Kjeldahl apparatus (FOSS Analytical AB, Sweden) was used to analyze total nitrogen from tomato plant tissues. Five samples from each treatment were used to measure the quantity of nitrogen.

Different levels of nitrogen-treated tomato plants released dynamic headspace volatiles were collected with solid phase micro extraction (SPME) fiber coated with poly dimethyl siloxane-divinyl benzene (PDMS-DVB, 65 μm) purchased from Supelco (Bellefonte, PA, USA). The volatile collection procedure is shown in Figure 1. Previously, many researchers collected VOCs in headspace static using the SPME method by placing a plant in a closed chamber, and this method consequently increased the temperature and relative humidity in the chamber, which could affect the normal physiological process in volatile chemicals emissions (Gouinguenè and Turlings, 2002; Egigu et al., 2014). In our experiment, VOCs were collected using the headspace dynamic SPME method. Here, a continuous airflow went through a 35-days old plant which was confined inside a glass jar (3 L) that minimized the variation of temperature and humidity inside the glass jar. Again, the same polythene bags and aluminum foil were used to avoid plant-growing media produced volatiles as described previously. On the top of the lid of the glass container was a large hole that closed with a glass made cover containing a bend opening (4 mm in diameter) which was used to insert the SPME fiber. At the bottom of the glass container was another hole for ventilation. A vacuum pump was used for circulating the air which was purified with activated charcoal and thereafter a Tenax TA adsorbent. The flow rate of the air was maintained at 200 ml min⁻¹ with a flow meter. The SPME fiber was conditioned at 250°C for 30 min in a gas chromatograph injection port according to the guideline of the manufacturer. The plant was kept in the glass container for 1 h before sampling for exiting impure volatiles from the system. The SPME needle was inserted into the opening of the glass container and extended the fiber to absorb dynamic headspace plant volatiles which escaped through the opening of the glass container for 1 h. After absorption of the volatiles, the SPME needle was directly inserted into a gas chromatograph-mass spectrometer (GC-MS) thermal desorption port as soon as possible, and thereafter the fiber was extended and kept 5 min for desorption of volatile substances.

The system consisted of a GC (TRACE 1310, Thermo Fisher Scientific, Waltham, MA, USA) that was used for the separation of volatile chemicals and an MS (ISQ Single Quadrupole MS, Thermo Fisher Scientific, Waltham, MA, USA) used for their detection, identification, and quantification. The thermally desorbed VOCs were separated in a 30 m × 0.25 mm × 0.25 μm film thickness fused silica capillary column (Zebron, ZB-5 MS Ui). Programming splitless injector temperature was maintained at 250°C whereas MS transfer line and ion source temperatures were maintained at 280°C. The purified helium (99.999%) was used as a carrier gas and the flow rate was maintained at 1.0 ml min⁻¹ with constant mode. The initial GC oven temperature was set to 40°C for 4 min. The oven temperature was increased from 40 to 250°C at a rate of 8°C min⁻¹ and held for 5 min. The MS was operated in an electron ionization (EI) mode. The ion energy and emission current were maintained at 70 eV and 25 μA, respectively. The Xcalibur program (Ver. 2.1, Thermo Electron Corporation, USA) was used for data acquisition which was performed in a total ion chromatogram (TIC) with mass range from 33 to 500 amu. The identification of separated compounds was carried out with NIST 2008 (National Institute of Standards and Technology, Washington, DC, USA) database. Kovats retention index (KI) was calculated for each constituent in relation to a mixture of n-alkanes standard (Van Den Dool and Kratz, 1963), and C₇–C₄₀ (Sigma–Aldrich, Louis, MO, USA). The data were matched to previously published data (Table 1). The peak area of each component of the volatiles was the relative quantity (Fang et al., 2013).

IBM SPSS statistics version 19 (Chicago, IL, USA) was used to conduct all statistical analyses. Data produced from Y-tube olfactometer choice bioassays were analyzed by X² test. Wind tunnel bioassay data were analyzed by paired t-test. Tomato plant volatiles and nitrogen were analyzed through one-way analysis of variance (ANOVA); means were separated by the Tukey test. Correlation was used to identify a possible relationship between different levels of nitrogen and volatile constituent production of tomato plants. In all cases, means were considered significant at P < 0.05 level.

Different levels of nitrogen-treated tomato plants were tested in the wind tunnel to investigate their attractiveness to B. tabaci in the presence of both visual and olfactory cues. The results of these dual choice bioassays are presented in Figure 2 for B. tabaci females and Figure 3 for B. tabaci males. The attractiveness of B. tabaci females was found different in 5, 10, 15, and 20 mM N treated tomato plants. The results revealed that significantly more B. tabaci females were found in 15 mM N (t = 7.317, P < 0.05; Figure 2B) and 20 mM N (t = 5.034, P < 0.05; Figure 2C) than in 5 mM N treated tomato plants. When given an option to choose 10 mM N versus 15 mM N treated plants_, B. tabaci females preferred 15 mM N to 10 mM N treated plant (t = 4.508, P < 0.05; Figure 2D). When 10 mM N and 20 mM N were offered, B. tabaci females preferred 20 mM N to 10 mM N treated plant (t = 6.263, P < 0.05; Figure 2E). However, B. tabaci females did not show significant preference between 5 mM N and 10 mM N (t = 1.199, P = 0.353; Figure 2A) and between 15 mM N and 20 mM N treated tomato plants (t = 1.609, P = 0.249; Figure 2F).

The data from the wind tunnel dual choice test showed that adult B. tabaci males did not have significant preference between 5 mM N and 10 mM N (t = 0.122, P = 0.914; Figure 3A), between 5 mM N and 15 mM N (t = 2.592, P = 0.122; Figure 3B), between 5 mM N and 20 mM N (t = 0.610, P = 0.604; Figure 3C), between 10 mM N and 15 mM N (t = 1.131, P = 0.375; Figure 3D), between 10 mM N and 20 mM N (t = 1.113, P = 0.382; Figure 3E), and between 15 mM N and 20 mM N (t = 0.355, P = 0.757; Figure 3F) treated tomato plants.

The preferences of B. tabaci were assayed in the Y-tube olfactometer toward volatile blends released from different levels of nitrogen-treated tomato plants, and the results are shown in Figure 4A for B. tabaci females and Figure 4B for B. tabaci males. Adult B. tabaci females showed significant preference among the VOCs of the plants treated with various levels of nitrogen including 5, 10, 15, and 20 mM N. Significantly more B. tabaci females were attracted to 15 mM N (χ² = 8.333; P < 0.01) than 5 mM N when 5 mM N and 15 mM N were offered, and to 20 mM N (χ² = 5.333; P < 0.05) when 5 mM N and 20 mM N were offered. Similarly, B. tabaci females preferred 15 mM N when 10 mM N and 15 mM N treated tomato plants were offered (χ² = 5.000; P < 0.05), and preferred 20 mM N (χ² = 5.488; P < 0.05) when 10 mM N and 20 mM N treated tomato plants were offered. However, B. tabaci females did not show significant preference between 5 mM N and 10 mM N (χ²= 0.556; P = 0.456), and between 15 mM N and 20 mM N (χ² = 1.043; P = 0.307) treated tomato plants (Figure 4A).

However, in the choice tests, B. tabaci male adults did not show any significant preference among the VOCs from the plants treated with different levels of nitrogen (5 mM N and 10 mM N: χ² = 1.333; P = 0.248; 5 mM N and 15 mM N: χ² = 0.364; P = 0.546; 5 mM N and 20 mM N: χ² = 1.140; P = 0.286; 10 mM N and 15 mM N: χ² = 0.000; P = 1.000; 10 mM N and 20 mM N: χ² = 2.689; P = 0.101; and 15 mM N and 20 mM N: χ² = 0.184; P = 0.668) (Figure 4B).

The results of total nitrogen in different levels of nitrogen-treated tomato plants are shown in Figure 5. Significant variation of total nitrogen in tomato plants (leaves with stem) were found due to different levels of nitrogen application. Tomato plants grown in 15 mM (T₃) and 20 mM (T₄) nitrogen levels had significantly higher percentage of total nitrogen than the plants grown in 5 mM (T₁) and 10 mM (T₂) nitrogen (F_3,16 = 12.163, P < 0.001). There was no significantly difference of percentages of total nitrogen in T₁ and T₂ treatments. Similarly, no significant difference of percentages of total nitrogen was observed between T₃ and T₄ treatments (Figure 5). However, different levels of nitrogen affected plant weight (Supplementary Figure S2).

Effects of nitrogen on tomato plant VOCs are shown in Figures 6A–P, and GC-MS chromatograms of volatiles are presented in Figures 7A–D. Sixteen VOCs were identified from the tomato plants with nitrogen fertilization treatments. Besides plant emitted VOCs, some compounds are generally related with the analytical system such as phthalates or siloxanes, as well as compounds related with earth’s atmosphere such as benzene and toluene (Warneke et al., 2001; Jansen et al., 2008) were not included in the list. The quantities of eight VOCs emitted from tomato plants, including heptanal (F_3,12 = 2.461, P = 0.113; Figure 6A), α-pinene (F_3,12 = 2.367, P = 0.122; Figure 6B), myrcene (F_3,12 = 1.993, P = 0.169; Figure 6D), nonanal (F_3,12 = 0.220, P = 0.881; Figure 6I), δ-elemene (F_3,12 = 2.007, P = 0.167; Figure 6J), longifolene (F_3,12 = 0.177, P = 0.910; Figure 6L), α-cedrene (F_3,12 = 0.733, P = 0.552; Figure 6M), and farnesan (F_3,12 = 0.178, P = 0.909; Figure 6P), were not significantly varied with increasing nitrogen treatments. The quantities in the remaining eight identified volatiles differed significantly among the N levels applied, including β-pinene (F_3,12 = 8.863, P < 0.01; Figure 6C), (+)-4-carene (F_3,12 = 7.853, P < 0.01; Figure 6E), α-terpinene (F_3,12 = 5.290, P < 0.05; Figure 6F), p-cymene (F_3,12 = 5.875, P < 0.05; Figure 6G), β-phellandrene (F_3,12 = 14.110, P < 0.001; Figure 6H), α-copaene (F_3,12 = 16.683, P < 0.001; Figure 6K), β-caryophyllene (F_3,12 = 29.783, P < 0.001; Figure 6N), and α-humulene (F_3,12 = 9.029, P < 0.01; Figure 6O). Of the eight significantly varied VOCs including β-pinene, (+)-4-carene, α-terpinene, p-cymene, β-phellandrene, α-copaene, β-caryophyllene, and α-humulene, no significant differences were found between 5 mM (T₁) and 10 mM (T₂) nitrogen-treated plant volatiles except for β-pinene, p-cymene, and α-humulene. The tomato plants at 15 mM (T₃) and 20 mM (T₄) nitrogen produced significantly lower levels of β-pinene, (+)-4-carene, α-terpinene, p-cymene, β-phellandrene, α-copaene, β-caryophyllene, and α-humulene as compared with the tomato plants grown at the 5 mM (T₁) and 10 mM (T₂) nitrogen. Of the eight significantly varied VOCs, no significant differences were observed between 15 mM (T₃) and 20 mM (T₄) nitrogen-treated plant volatiles except for α-terpinene and α-humulene. The amounts of significantly varied eight VOCs identified are well and negatively correlated with the N levels in the tomato plants, the higher the N levels, the lower of the VOCs identified (r = −0.883 to −0.967).