The Role of Insect-Derived Cues in Eliciting Indirect Plant Defenses in Tobacco, Nicotiana tabacum

Abstract

In response to insect feeding, plants release complex volatile blends that are important host-location cues for natural enemies of herbivores. These induced volatile responses are mediated by insect-derived cues and differ significantly from responses to mechanical wounding. To improve understanding of the cues that elicit plant volatile responses, we explored the effects of Heliothis virescens saliva on volatile induction in tobacco, Nicotiana tabacum, using an ablation technique that prevents the release of saliva from the labial glands during feeding. Plants damaged by intact caterpillars released 11 volatile compounds. Ablated caterpillars induced these same 11 compounds plus an additional eight. Of the 11 shared compounds, plants damaged by ablated caterpillars released greater quantities of six, most notably volatile nicotine, compared to plants damaged by intact caterpillars. We further investigated the effects of H. virescens oral secretions on volatile induction through the collection and application of caterpillar regurgitant and saliva to mechanically wounded plants. Plants treated with H. virescens regurgitant released significantly more volatile nicotine than plants treated with saliva or those damaged by intact caterpillars. Additionally, application of a mixture of saliva and regurgitant induced less volatile nicotine compared to treatment with regurgitant alone. Our results suggest that saliva has an inhibitory effect on plant volatile responses to H. virescens feeding and that insect-derived cues originating from both regurgitant and saliva may interact to elicit the volatile “signature” of H. virescens.

Introduction

Plants actively synthesize and release complex volatile blends in response to attack by insect herbivores.^1–7 These induced volatiles serve as important host-location cues for foraging natural enemies of the herbivores and can convey information about herbivore species identity.^3,5,8,9 Several insect-derived cues involved in the mediation of plant responses to herbivore attack have been isolated and identified from both insect regurgitant^1,10–15 and saliva.^16,17

Regurgitant and saliva are distinct secretions differing in origin and composition.¹⁸ Regurgitant arises from the fore- and midgut and can be collected from caterpillars by gently squeezing them with forceps. It is unclear whether regurgitant is typically secreted during feeding, though indirect evidence suggests that it is.¹⁸ Saliva is released from the labial glands via the spinneret, external to the oral cavity, and is actively released during feeding.^18,19 For simplicity, we will refer to regurgitant and saliva collectively as oral secretions.

Three categories of elicitors have been identified in caterpillar regurgitant: enzymes, fatty acid-amino acid conjugates, and peptides. The first elicitor identified was β-glucosidase, an enzyme found in the regurgitant of Pieris brassicae caterpillars that induces cabbage plants to release a volatile blend similar to that released in response to caterpillar feeding.¹⁰ The second elicitor identified was volicitin, N-[17-hydroxylinolenoyl]-L-glutamine, a fatty acid-amino acid conjugate present in the regurgitant of the beet armyworm (BAW) Spodoptera exigua. When applied to mechanically wounded maize seedlings, volicitin induces the release of a volatile blend similar to that released in response to BAW feeding, but not in response to mechanical wounding alone.^11,20,21 Since the discovery of volicitin, several additional fatty acid amides have been found in the regurgitant of other caterpillar species.^14,20,22 More recently, the peptide elicitor inceptin has been isolated and identified from the oral secretions of fall armyworm, Spodoptera frugiperda.²³ Inceptins are proteolytic fragments of chloroplastic ATP synthase produced by the protease action of herbivore feeding. Application of inceptin or S. frugiperda oral secretions to wounded cowpea (Vigna unguiculata) leaves resulted in the induction of similar levels of several important plant defense-signaling hormones.

Insect saliva performs various functions from initial digestion and lubrication of mouthparts to the suppression of animal-host defenses.^18,24 Recently, caterpillar saliva has been shown to play a role in suppressing herbivore-induced plant defenses.^16,25,26 The first identified suppressor of plant defenses was glucose oxidase (GOX), an enzyme isolated from the saliva of Helicoverpa zea, and subsequently found in other caterpillar species.^16,27 Application of GOX or salivary-gland extract suppresses the wound-inducible production of foliar nicotine in tobacco, Nicotiana tabacum, and results in increased survival and growth of larvae fed GOX-treated leaves compared to those fed leaves treated with water alone.¹⁶ Glucose oxidase also suppresses transcription of genes involved in terpenoid biosynthesis in alfalfa, Medicago truncatula.¹⁷ In addition to suppressing plant defenses, saliva has been shown to induce the production of some plant volatiles.²⁸ In maize, application of salivary gland extracts of the western tarnished plant bug, Lygus hesperus, induces the release of the same blend of volatiles as that released in response to L. hesperus feeding.²⁸

The experiments described here explored the induction of plant volatiles of tobacco, N. tabacum, in response to Heliothis virescens oral secretions. Previous studies of plant volatile responses to insect-derived compounds have employed artificial wounding followed by the application of potential elicitors. Our study utilized an ablation method that involves cauterizing the spinneret, which prevents the secretion of saliva.^16,19 This allowed us to examine the volatile response of plants to insects that do not produce saliva, but are otherwise able to feed normally. In addition to examining plant volatile responses to ablated and intact caterpillars, we also examined the influence of H. virescens oral secretions on volatile induction through the application of regurgitant and saliva to mechanically wounded tobacco plants.

Materials and Methods

Plants and insects.

Tobacco seeds (Nicotiana tabacum strain K326) were germinated in a peat-based, general-purpose potting soil (Pro-Mix, Premier Horticulture Inc., Quakertown, PA), and transplanted as two-week-old seedlings into pots (16-cm tall × 16.5-cm diameter) with fertilizer (Osmocote/Hummert International, MO) in a growth chamber (16:8 L:D; 25°C:22°C Day: Night; 65% RH). Five-week-old plants with three fully-expanded leaves were used for all experiments. Eggs of H. virescens were reared on an artificial casein diet in a growth chamber under the same conditions. Insects in all experiments were reared on artificial diet except where otherwise specified.

Ablation of H. virescens spinneret.

Fourth-instar H. virescens were ablated using a previously described method.^16,19 Briefly, caterpillars were chilled on ice until flaccid and then the spinneret was cauterized by touching it with a heat pen (Electron Microscopy Sciences, Fort Washington, PA). The larvae were then returned to artificial diet for approximately two hours prior to use in experiments.

Collection of H. virescens salivary glands and regurgitant.

Labial salivary glands were dissected from artificial-diet-fed H. virescens larvae that had been fifth-instars for 48 hr and had been chilled on ice for approximately five minutes. Once removed, salivary glands were placed into microcentrifuge tubes, kept on ice in groups of ten pairs, and stored at −80°C until needed. Immediately prior to use, each ten-gland sample was extracted in 50 µl physiologically-buffered saline (pH 7.2) by homogenizing the salivary glands with a hand-held pestle in a microcentrifuge tube kept on ice. This allowed salivary-gland extracts to be applied to mechanically wounded leaves using a pipette. Protein concentrations were determined using the method of Bradford (1976).²⁹

Prior to collecting regurgitant, recently molted fifth-instar caterpillars were fed for 48 hr on tobacco leaves from approximately seven-week-old plants. Regurgitant was collected from these larvae by gently squeezing the caterpillars, collecting the resulting oral secretions using a pipette, and dispensing them into a microcentrifuge tube on ice. Regurgitant was pooled from approximately 50 larvae then separated into 50 µl aliquots and stored at −80°C until needed.

Induced plant volatiles.

Feeding by ablated versus intact H. virescens. To examine induced plant responses to herbivory by caterpillars that were unable to secrete saliva during feeding, we collected volatiles from tobacco plants that were damaged by H. virescens larvae with ablated (cauterized) spinnerets and from plants damaged by intact H. virescens. Because ablated caterpillars fed less than intact caterpillars, the numbers of ablated versus intact caterpillars were manipulated daily to equalize damage across treatments as previous research has demonstrated a correlation between the amount of damage and volatile production in other plant-herbivore systems.³⁰ Typically, between five and seven ablated caterpillars were feeding at a given time, compared to two intact caterpillars. Ablated and intact caterpillars were allowed to feed continuously throughout the four day experiment, and volatiles were collected on each of the four days. Undamaged plants were used as controls.

Treatment with oral secretions. To examine induced plant responses to the application of H. virescens oral secretions, we collected volatiles from tobacco plants receiving the following treatments: (1) control (i.e., undamaged) plants, (2) mechanical wounding, (3) mechanical wounding plus the application of 40 µl of H. virescens regurgitant distributed among three wounds/day, (4) mechanical wounding plus the application of 40 µl of H. virescens homogenized salivary glands distributed among three wounds/day, or (5) feeding by three third-instar H. virescens larvae. Plants were mechanically wounded by using a razor blade to scrape an approximately 1 cm² area on each of three leaves per plant. The same three leaves received new mechanical damage each day, so that at the end of the four-day experiment plants had three leaves with four wounds each. The amounts of oral secretions applied to plants were based on the amount of glucose oxidase secreted by Helicoverpa zea caterpillars during feeding, approximately 15 µg/day/caterpillar.¹⁹ Peiffer and Felton (2005) found that ∼10% of the total protein concentration of secreted saliva is glucose oxidase. Total protein concentrations for our samples of homogenized salivary glands were around 700 µg per 40 µl. Because we used homogenized salivary glands, not secreted saliva, the percentage of glucose oxidase in our samples was considerably less than 10% of the total protein concentration. We used 40 µl of regurgitant to standardize the volume of regurgitant with the volume of homogenized salivary glands applied to plants. Caterpillars were allowed to feed continuously throughout the four day experiment. Treatments were applied immediately prior to volatile collection and volatiles were collected on each of the four days.

We conducted an additional experiment to examine the combined effects of saliva and regurgitant on volatile responses. Plants were wounded with a razor blade (as above). We then applied 40 µl of a mixture of H. virescens homogenized salivary glands and regurgitant to the wounded sites (as above). The mixture of homogenized salivary glands and regurgitant was obtained by combining 25 µl of homogenized salivary glands to 25 µl of regurgitant. We applied 40 µl of the mixture to plants to standardize the volume of oral secretions applied to plants in the previous experiment. Treatments were applied immediately prior to volatile collections and volatiles were collected on each of the four days. It should be noted that this experiment was performed under higher ambient temperatures and longer day lengths relative to the previous experiments, which can influence plant responses.⁷

Collection and analysis of volatiles.

Volatiles were collected from tobacco plants using a closed push/pull system (Analytical Research Systems, Inc., Gainsville, FL). A two-piece Teflon® base with a hole for the plant stem rested on the pot. A glass dome chamber (15-cm tall × 16-cm diameter) enclosed plants and rested on the base. Filtered air was pushed through Teflon® tubing into the top of the chamber (3.0 L/min) and was pulled through side ports (1 L/min) across beds of adsorbent Super-Q® (25 mg, Alltech Associates, Deerfield, IL). Plant volatiles were collected between 10:00 and 22:00 (light period 06:00 to 22:00). Super-Q® traps were rinsed with 150 µl of dichloromethane; 5 µl of n-octane (40 ng/µl) and nonyl acetate (80 ng/µl) were added as internal standards. Samples were injected in 1-µl aliquots into an Agilent model 6890 gas chromatograph fitted with a flame ionization detector (FID), using a splitless injector held at 220°C. The column (OV 101 methyl silicon) was maintained at 35°C for 0.5 min and then increased 12°C per min to 180°C. Quantifications of compounds were made relative to the nonyl acetate internal standard using ChemStation software (Agilent Technologies, Wilmington, DE). Samples were also analyzed by GC-MS using electron ionization. Identifications of all volatile compounds were confirmed by comparing retention times and mass spectra to commercial standards. For further details on analytical methods see De Moraes and Mescher (2004).³¹

Leaf measurements.

To determine leaf area for plants used for volatile collections, we removed leaves from plants and photocopied them. Photocopies were digitized and leaf areas and amount of damage were determined using an imaging program (SigmaScan® Pro, SPSS Inc., Chicago, IL).

Statistical analyses.

We analyzed data collected on day four from both volatile collection experiments in order to allow sufficient induction in response to herbivory. We quantified all volatile compounds released in measurable amounts. The data from the ablated versus intact feeding experiment were analyzed by ANOVA with trial treated as a random effect. We used a one-sample t-test to determine whether compounds only released in response to damage by ablated caterpillars were significantly different from zero. The data from the oral secretions experiment were log_e (x + 1) transformed or square-root transformed to satisfy the assumptions of normality and homogeneity of variance among treatments and analyzed by ANOVA with trial treated as a random effect. We analyzed leaf damage by ANOVA with trial treated as a random effect. Pairwise comparisons were conducted using Tukey's HSD test. All statistical analyses were conducted using Minitab v. 14.1 (Minitab Inc., State College, PA).

Results

Induced plant volatiles.

Feeding by ablated versus intact H. virescens. In response to feeding by intact caterpillars, tobacco plants released a volatile profile that was qualitatively and quantitatively different from that released in response to feeding by ablated caterpillars (Fig. 1; Table 1). Feeding by intact caterpillars induced the release of 11 compounds, while feeding by ablated caterpillars induced the same 11 compounds plus an additional eight (Table 1). Of the 11 shared compounds, the production of six were significantly higher in response to feeding by ablated caterpillars (five terpenoids and nicotine; p < 0.05; Table 1). Of the eight compounds released only in response to feeding by ablated caterpillars, two, (Z)-3-hexenyl tiglate and β-farnesene, were produced in quantities significantly different from zero (p < 0.05; Table 1). Total volatile production was also significantly higher in response to feeding by ablated caterpillars (ANOVA: F_{1, 5} = 16.53, p = 0.01).

Figure 1.

Total ion chromatograms showing typical daytime emissions released by tobacco plants damaged by ablated versus intact H. virescens. Compounds are labeled as follows: 1, unknown 1; 2, unknown 2; 3, unknown 3; 4, myrcene; 5, (Z)-3-hexenyl acetate; 6, (E)-β-ocimene; 7, linalool; 8, (Z)-3-hexenyl isobutyrate; 9, (Z)-3-hexenyl butyrate; 10, indole; 11, (Z)-3-hexenyl tiglate; 12, nicotine; 13, β-elemene; 14, β-caryophyllene; 15, α-humulene; 16, β-farnesene, 17, unidentified sesquiterpene; 18, (E,E)-α-farnesene; 19, caryophyllene oxide; IS and IS₂, internal standards (n-octane and n-nonyl-acetate).

Table 1.

Amounts (ng/day; mean ± SE) of volatiles released on day four by tobacco plants damaged by ablated versus intact H. virescens (n = 4)

	Treatment
Compound	Ablated	Intact	P-value
1. unknown 1a	7.30 ± 4.28	0 ± 0	0.187
2. unknown 2a	21.2 ± 10.4	0 ± 0	0.135
3. unknown 3a	27.6 ± 10.4	0 ± 0	0.077
4. myrceneb	31.87 ± 3.47	18.86 ± 1.06	0.018
5. (Z)-3-hexenyl acetatea	61 ± 37	0 ± 0	0.198
6. (E)-β-ocimeneb	7886 ± 1654	3145 ± 382	0.018
7. linaloolb	265.7 ± 54.2	125.3 ± 16.9	0.068
8. (Z)-3-hexenyl isobutyratea	8.67 ± 8.67	0 ± 0	0.391
9. (Z)-3-hexenyl butyratea	17 ± 14.5	0 ± 0	0.325
10. indoleb	166.1 ± 63.7	92 ± 58.2	0.142
11. (Z)-3-hexenyl tiglatea	41.1 ± 11.6	0 ± 0	0.039
12. nicotineb	1017 ± 231	155.7 ± 58.5	0.008
13. β-elemeneb	124 ± 25	28.7 ± 12.9	0.008
14. β-caryophylleneb	6447 ± 1917	2957 ± 581	0.165
15. α-humuleneb	224.2 ± 70.6	91.9 ± 19	0.145
16. β-farnesenea	18.75 ± 2.99	0 ± 0	0.008
17. unidentified sesquiterpeneb	492 ± 142	139.8 ± 47.4	0.032
18. (E,E)-α-farneseneb	847 ± 281	173.6 ± 50.2	0.048
19. caryophyllene oxideb	188.3 ± 42.3	120.8 ± 16.9	0.231
Total volatilesb	17896 ± 2434	7049 ± 977	0.010

^aAnalyzed using one-sample t-test to test difference from zero, α = 0.05;

^bAnalyzed by ANOVA, α = 0.05 .

We manipulated the numbers of caterpillars daily to control for tissue loss. Thus, there were no significant differences in the amount of leaf tissue damage per day by intact versus ablated caterpillars (1619 ± 272, 1169 ± 296 mm² damage/day, respectively; ANOVA: F_1,5 = 4.16, p > 0.05), although the mean amount of damage appeared higher for intact caterpillars. We did, however, observe differences in how the caterpillars damaged the leaf tissue. Intact caterpillars fed in a more or less continuous manner, removing larger areas of tissue at a given time, compared to ablated caterpillars, which tended to produce many small feeding holes. Because these differences in feeding patterns could have affected the amount of contact caterpillars had with the leaf tissue during feeding, we measured the perimeters of the feeding holes. There were no significant differences in the combined perimeters of the feeding holes per day between intact and ablated caterpillars (623 ± 105, 745 ± 121 mm leaf edge/day, respectively; ANOVA: F_{1, 5} = 1.47, p > 0.05).

Treatment with oral secretions. The volatile profile released by tobacco plants in response to caterpillar feeding was distinctly different from that released in response to application of H. virescens regurgitant, saliva, or mechanical wounding (Fig. 2; Table 2). Heliothis virescens feeding consistently induced the release of 11 compounds. Mechanical wounding alone induced the release of very small concentrations of three (E)-β-ocimene, nicotine, and β-caryophyllene) of these 11 compounds (Fig. 2; Table 2). Plants that were mechanically wounded and treated with H. virescens regurgitant emitted seven of the 11 compounds released in response to caterpillar feeding, plus one additional compound, β-pinene (Table 2). Of these eight compounds, only four (E)-β-ocimene, linalool, nicotine, and β-caryophyllene) were consistently released (Table 2). The other four compounds (α-humulene, an unidentified sesquiterpene, (E,E)-α-farnesene, and β-pinene) were only released in one trial. Application of H. virescens saliva to mechanically wounded plants induced the release of five of the 11 compounds released in response to caterpillar feeding, plus two additional compounds, α-pinene and β-pinene (Table 2). Of these seven compounds, five (E)-β-ocimene, linalool, nicotine, β-caryophyllene, and β-pinene) were released consistently (Table 2).

Figure 2.

Total ion chromatograms showing typical daytime emissions released by tobacco plants receiving the following treatments: unwounded control plants (C), mechanical wounding (W), mechanical wounding plus the application of H. virescens regurgitant (R), mechanical wounding plus the application of H. virescens homogenized salivary glands (S), and H. virescens feeding (HV). Compounds are labeled as follows: 1, α-pinene; 2, β-pinene; 3, myrcene; 4, (E)-β-ocimene; 5, linalool; 6, indole; 7, nicotine; 8, β-elemene; 9, β-caryophyllene; 10, α-humulene; 11, unidentified sesquiterpene; 12, (E,E)-α-farnesene; 13, caryophyllene oxide; IS and IS₂, internal standards (n-octane and n-nonyl-acetate).

Table 2.

Amounts (ng/day; mean ± SE) of volatiles released on day four by tobacco plants^*

	Treatment
Compound	C	HV	R	S	W
1. α-pinene	0 ± 0	0 ± 0	0 ± 0	23.7 ± 15	4.11 ± 4.11
2. β-pinene	0 ± 0	0 ± 0	6.41 ± 6.41	14.93 ± 1.24	7.95 ± 4.87
3. myrcene	0 ± 0	9.63 ± 6.15	0 ± 0	0 ± 0	0 ± 0
4. (E)-β-ocimene	0 ± 0	3094 ± 1427	158.9 ± 48	141.9 ± 65.2	49 ± 8.31
5. linalool	0 ± 0	126.3 ± 66.3	12.2 ± 7.89	19.72 ± 5.88	0 ± 0
6. indole	0 ± 0	110 ± 61	0 ± 0	0 ± 0	0 ± 0
7. nicotine	0 ± 0	120.8 ± 74	878 ± 215	227.2 ± 72.2	91.8 ± 56.2
8. β-elemene	0 ± 0	74.6 ± 47.6	0 ± 0	0 ± 0	0 ± 0
9. β-caryophyllene	0 ± 0	1964 ± 661	365 ± 90.5	129.4 ± 45.9	110.4 ± 38.7
10. α-humulene	0 ± 0	63.5 ± 23.3	4.72 ± 4.72	0 ± 0	0 ± 0
11. unidentified sesquiterpene	0 ± 0	258 ± 151	7.11 ± 7.11	0 ± 0	0 ± 0
12. (E,E)-α-farnesene	0 ± 0	554 ± 338	3.86 ± 3.86	5.02 ± 5.02	0 ± 0
13. caryophyllene oxide	0 ± 0	77.8 ± 25.2	0 ± 0	0 ± 0	0 ± 0

^*Receiving the following treatments: unwounded control plants (C), mechanical wounding (W), mechanical wounding plus the application of H. virescens regurgitant (R), mechanical wounding plus the application of H. virescens homogenized salivary glands (S), and H. virescens feeding (HV) (n = 5).

The two dominant compounds released in response to H. virescens feeding were (E)-β-ocimene and β-caryophyllene (Table 2). Significantly more (E)-β-ocimene was released in response to H. virescens feeding compared to treatment with H. virescens regurgitant, saliva, or mechanical wounding (Fig. 3a; ANOVA: F_{4, 16} = 81.54, p < 0.001). Plants treated with either H. virescens regurgitant or saliva released significantly more (E)-β-ocimene than unwounded control plants (Tukey's HSD p < 0.001), but were not significantly different from one another or from plants that were mechanically wounded (Tukey's HSD p > 0.05; Fig. 3a). More β-caryophyllene was released by plants in response to H. virescens feeding compared to plants treated with H. virescens saliva (ANOVA: F_4,16 = 25.96, p < 0.001), but not in comparison to plants treated with H. virescens regurgitant (Tukey's HSD p > 0.05; Fig. 3b).

Figure 3.

Mean (± SE) amounts of (E)-β-ocimene (A) and β-caryophyllene (B) released on day four by plants receiving the following treatments: unwounded control plants (C), mechanical wounding (W), mechanical wounding plus the application of H. virescens regurgitant (R), mechanical wounding plus the application of H. virescens homogenized salivary glands (S), and H. virescens feeding (HV) (n = 5). Data are presented untransformed. Letters above each bar indicate significant differences as indicated by Tukey's HSD test on log_e (x ± 1) transformed data, α = 0.05.

In addition to (E)-β-ocimene and β-caryophyllene, nicotine was released by plants damaged by H. virescens, plants treated with H. virescens regurgitant or saliva, and mechanically wounded plants (Fig. 2; Table 2). Plants treated with H. virescens regurgitant released significantly more nicotine than any other treatment (ANOVA: F_4,16 = 14.94, p < 0.001; Fig. 4; Table 2).

Figure 4.

Mean (± SE) amounts of nicotine released on day four by plants receiving the following treatments: unwounded control plants (C), mechanical wounding (W), mechanical wounding plus the application of H. virescens regurgitant (R), mechanical wounding plus the application of H. virescens homogenized salivary glands (S), and H. virescens feeding (HV) (n = 5). Data are presented untransformed. Letters above each bar indicate significant differences as indicated by Tukey's HSD test on square-root transformed data, α = 0.05.

Application of both regurgitant and saliva to mechanically wounded plants elicited a volatile response qualitatively and quantitatively more similar to feeding by intact caterpillars than that induced by the application of either regurgitant or saliva alone (data not shown). However, this treatment still induced fewer compounds than caterpillar feeding and most compounds were produced in lesser amounts. Plants treated with saliva and regurgitant released nine of the 11 compounds released in response to H. virescens feeding (myrcene, (E)-β-ocimene, linalool, nicotine, β-elemene, β-caryophyllene, α-humulene, an unidentified sesquiterpene, and (E,E)-α-farnesene; data not shown). Because this experiment was conducted under different environmental conditions (i.e., longer day lengths and higher ambient temperatures than those used for other experiments), direct comparisons between treatments are not ideal, as higher temperatures may have resulted in a general increase in volatile responses.⁷ Treatments with saliva and regurgitant individually, run as controls, elicited responses that were qualitatively similar but quantitatively elevated compared to previous experiments. Nevertheless, application of both saliva and regurgitant to mechanically wounded plants induced less nicotine (302 ± 55.7 ng/day) in comparison to treatment with regurgitant alone (Table 2).

Discussion

Our results demonstrate that feeding by ablated caterpillars that do not produce saliva elicits a volatile profile that is qualitatively and quantitatively distinct from that released in response to feeding by intact caterpillars (Fig. 1; Table 1). These findings indicate that cues associated with H. virescens saliva play a role in mediating the release of herbivore-induced volatiles; in particular, saliva appears to suppress production of volatile nicotine as well as several other compounds that were released in response to feeding by ablated H. virescens (Table 1).

Feeding by ablated caterpillars induced the release of 19 compounds versus 11 for intact caterpillars (Table 1). Of the compounds released in response to both ablated and intact caterpillars, six were induced in significantly greater quantities in response to feeding by ablated caterpillars (Table 1). Five of these (myrcene, (E)-β-ocimene, β-elemene, an unidentified sesquiterpene, and (E,E)-α-farnesene) are products of terpenoid biosynthesis. Three additional terpenoids, (linalool, β-caryophyllene, and α-humulene) were also emitted in greater amounts in response to ablated-caterpillar feeding, though the differences were not statistically significant (Table 1). These results are consistent with the recent finding that factors associated with caterpillar saliva suppress the transcription of genes encoding early enzymes involved in the production of terpenoids.¹⁷ Notably, several of the compounds released only in response to ablated caterpillars have previously been shown to be released by tobacco during the dark phase of the photoperiod in response to feeding by intact H. virescens and to be repellent to female H. virescens moths.³²

A major difference in plant responses to feeding by ablated and intact caterpillars was the release of significantly greater amounts of nicotine in response to feeding by ablated caterpillars (Fig. 1; Table 1). Plants damaged by ablated caterpillars released more than six times as much nicotine as plants damaged by intact caterpillars (Table 1). Similarly, mechanically wounded plants treated with regurgitant released significantly more volatile nicotine than plants damaged by intact caterpillars or those treated with H. virescens saliva (Fig. 4; Table 2). Application of both saliva and regurgitant to mechanically wounded plants induced less volatile nicotine compared to treatment with regurgitant alone. These findings suggest that insect-derived cues associated with H. virescens regurgitant induce the production and release of large quantities of volatile nicotine and that factors present in H. virescens saliva suppresses this induction (Figs. 1 and 2). This is consistent with the observation that glucose oxidase, a compound present in the saliva of many caterpillars, including H. virescens (unpublished data), suppresses the wound-inducible production of foliar nicotine in N. tabacum.¹⁶ However, our findings contrast with previous reports that the application of Manduca sexta regurgitant to mechanically wounded tobacco, Nicotiana sylvestris, results in lower whole-plant nicotine levels than mechanical wounding.³³ Differences between these systems might be explained by differences in the profiles of elicitors in the regurgitant of H. virescens and M. sexta³⁴ and/or differences between Nicotiana species.

Previous research in maize has demonstrated a correlation between damage intensity and the amount of volatiles released.³⁰ Despite feeding by different numbers of larvae and/or different larval instars, Gouinguené et al. (2003)³⁰ found no differences in volatile emissions between plants with equal amounts of feeding damage. Moreover, only quantitative differences in volatile production were found between plants with different amounts of damage. Because we successfully controlled for the amount of feeding damage inflicted (i.e., area removed) by ablated and intact caterpillars, the observed qualitative and quantitative differences in volatile profiles cannot be attributed to differences in tissue loss. We further controlled for the perimeter of feeding holes, as Dean and De Moraes (2006)³⁵ found that differences in volatile emissions observed between Bt and non-Bt maize with similar amounts of tissue loss were no longer evident when they equalized the perimeters of feeding holes caused by H. zea feeding.

Although we successfully controlled the intensity of feeding damage, the observed feeding patterns for ablated and intact caterpillars were different. Intact caterpillars fed in a more or less continuous manner and removed larger areas of tissue. Ablated caterpillars tended to create many small feeding holes and more than double the number of intact caterpillars were needed to equalize the tissue loss between the two treatments. This behavioral difference might be explained by several factors. The absence of saliva may directly inhibit feeding by making it more difficult for the caterpillar to process plant tissues. Indirectly, the absence of saliva might inhibit feeding through an increase in the strength of plant defense mechanisms. Musser et al. (2002)¹⁶ demonstrated that feeding by ablated caterpillars causes increased production of foliar nicotine levels in N. tabacum. As we see induction of volatile nicotine in response to ablated caterpillar feeding, it is reasonable to assume that foliar nicotine, and possibly other compounds, are being induced in the absence of salivary secretions, making the plant less palatable and/or more toxic to the insect. Previous studies have demonstrated altered feeding behavior and avoidance in response to transgenic crops.^36,37 Dean and De Moraes (2006)³⁵ found that the feeding behavior of H. zea on Bt and non-Bt maize resulted in a feeding pattern similar to that observed in our study—H. zea caused many small feeding holes on Bt maize and required more larvae to produce amounts of tissue loss equal to that observed on non-Bt maize. Finally, it is possible that the ablation method we employed may have other unknown effects on caterpillars that alter their feeding patterns.

Apart from the pattern of volatile nicotine induction discussed above, we observed a lower volatile response to the application of oral secretions to mechanically wounded plants compared to feeding by intact or ablated caterpillars. The application of either regurgitant or saliva alone induced the release of far fewer compounds and in lower amounts than caterpillar feeding (Table 2). Application of both saliva and regurgitant induced a volatile response that was more similar to intact caterpillar feeding, but still induced the release of fewer compounds and in lower amounts.

The reduced volatile responses to the application of oral secretions to mechanically wounded plants compared to that elicited by caterpillars might be explained by differences in the wounding regime. A recent study suggests that plants may be able to distinguish between continuous, sustained damage caused by caterpillar herbivory compared to a single wounding event.³⁸ In lima bean, Phaseolus lunatus, a continuous, long-lasting, mechanical wounding protocol more closely approximating actual herbivore feeding damage elicited a volatile blend that was qualitatively similar to that released in response to caterpillar feeding, though quantitative differences for several compounds were still observed.³⁸ While we mechanically wounded plants once each day, it is possible that a different wounding regime and/or method of applying oral secretions would have produced different results. It is also important to note that we did not apply pure caterpillar saliva to plants, but rather homogenized salivary glands. We used homogenized salivary glands as a substitute for saliva as others have done due to the difficulty in obtaining pure saliva,¹⁶ but this preparation may contain some compounds and/or concentrations not normally associated with salivary secretions that may possibly have influenced plant responses. In contrast, our experiments with ablated caterpillars should mimic normal caterpillar feeding damage as accurately as possible. Our finding that application of caterpillar regurgitant was not sufficient to induce a volatile blend similar to caterpillar herbivory was contrary to expectations derived from previous studies. For example, in corn, the application of beet armyworm, Spodoptera exigua, regurgitant to mechanically wounded corn seedlings elicited a volatile profile similar to that elicited by beet armyworm (BAW) feeding.¹ Application of regurgitant from several other caterpillar species to corn seedlings also resulted in the release of the same blend of eight compounds in the same relative ratios as released in response to BAW regurgitant.³⁹ More recently, Dean and De Moraes (2006)³⁵ found that application of corn earworm, H. zea, regurgitant to mechanically wounded corn seedlings was sufficient to induce a volatile blend similar to that released in response to corn earworm feeding. These results suggest that insect-derived cues associated with herbivory may elicit a more generalized response in maize than in tobacco. They further suggest that different plant species may use different mechanisms or biochemical pathways to recognize herbivore attack and to regulate the production of plant volatiles.

In summary, our study offers insight into the induction of plant volatiles in response to insect-derived cues. We have shown that feeding by ablated caterpillars elicits a volatile profile that is qualitatively and quantitatively different from that induced by intact-caterpillar feeding and that neither H. virescens regurgitant nor saliva was sufficient to mimic the volatile response of tobacco plants to feeding by intact H. virescens. It appears that cues originating from both regurgitant and saliva are necessary and may interact to elicit a volatile response that is similar to that released in response to H. virescens feeding. It also appears that factors present in the saliva of H. virescens play a significant role in suppressing the induction of nicotine and other volatiles that are induced by feeding by ablated caterpillars or the application of regurgitant to mechanically wounded plants. To our knowledge, this is the first study to show a potential role for saliva in suppressing volatile nicotine and mediating the release of other herbivore-induced volatiles. Further investigations into natural plant defenses in response to insect herbivory and the specific elicitors mediating these complex interactions has potential relevance for the development of effective, biologically based pest management techniques.