Defensive Function of Herbivore-Induced Plant Volatile Emissions in Nature

See allHide authors and affiliations

Science  16 Mar 2001:
Vol. 291, Issue 5511, pp. 2141-2144
DOI: 10.1126/science.291.5511.2141


Herbivore attack is known to increase the emission of volatiles, which attract predators to herbivore-damaged plants in the laboratory and agricultural systems. We quantified volatile emissions fromNicotiana attenuata plants growing in natural populations during attack by three species of leaf-feeding herbivores and mimicked the release of five commonly emitted volatiles individually. Three compounds (cis-3-hexen-1-ol, linalool, and cis-α-bergamotene) increased egg predation rates by a generalist predator; linalool and the complete blend decreased lepidopteran oviposition rates. As a consequence, a plant could reduce the number of herbivores by more than 90% by releasing volatiles. These results confirm that indirect defenses can operate in nature.

Plants defend themselves against herbivores with chemical and physical defenses that directly influence herbivore performance and indirectly through traits that attract the natural enemies of herbivores (1–3). One such indirect defense, the release of volatile organic compounds (VOCs) specifically after herbivory, is known to attract parasitoids and predators to actively feeding larvae in the laboratory (4,5), and evidence from agricultural systems suggests a role for herbivore-induced VOCs in increasing predation pressure (6–8). However, conclusive evidence has been lacking, and it is not even known whether plants growing in natural populations increase VOC emissions after herbivore attack. VOCs might be able to function as indirect defenses only in simplified agroecosystems, in which a single natural enemy species of a herbivore can act as an important biocontrol agent on an agricultural plant (9). In contrast, in natural systems, herbivore mortality is more commonly mediated by a suite of generalist enemies (10). Moreover, both the qualitative and quantitative characteristics of herbivore-induced plumes of VOCs are known to vary among plant genotypes (11–13); the genetic variation commonly found in natural populations may undermine the reliability of VOCs as a signal for natural enemies because prior exposure is often needed to associate plant VOCs with the occurrence of a feeding herbivore (5,14, 15). Herbivore-induced plant VOCs may also influence herbivore host-location behavior, potentially increasing herbivore attack on plants releasing VOCs (1, 2).

To evaluate the role of herbivore-induced VOCs in nature, we characterized the VOCs released from Nicotiana attenuataTorr. ex Wats (Solanaceae) plants growing in a native population (16–18) in the Great Basin desert of southwest Utah, which were under continuous attack by three numerically dominant folivores: the caterpillars of Manduca quinquemaculata (Lepidoptera, Sphingidae), the leaf bug Dicyphus minimus (Heteroptera, Miridae), and the flea beetle Epitrix hirtipennis(Coleoptera, Chrysomelidae) (19). We used an open-flow trapping design (13) to collect VOCs individually from 32 plants growing in a natural population that each had one leaf attacked by one of the three herbivore species or remained undamaged (13, 20) (Fig. 1, A and B). All plants were growing in a 150-m2 portion of the population (18) and were sampled simultaneously for 7 hours. Gas chromatography–mass spectrometry (GC-MS) analysis (21) of the trapped VOCs revealed that all three herbivore species elicited increases in the same suite of VOCs, although the odor profiles were not identical (Fig. 1B). The pattern and amount of herbivore-induced VOCs trapped from N. attenuata growing in the field were very similar to those found in laboratory studies with plants attacked by Manduca sexta larvae (13).

Figure 1

VOC release in response to herbivory in nature. (A) Representative total ion chromatograms of the headspace of an undamaged leaf from an undamaged plant (CTRL) in comparison with a leaf damaged by a M. quinquemaculata larva (hornworm). (B) Comparison of the mean (±SEM) emissions of VOCs from undamaged plants (CTRL) and plants damaged by M. quinquemaculata larvae (hornworm), D. minimus (leaf bug), and E. hirtipennis (flea beetle) (with eight replicate plants per treatment in the same native population). Because herbivore-induced VOCs from N. attenuata are emitted with different kinetics depending on the compound (13), data from the 7-hour trapping, 24 hours after herbivore infestation, are presented. Each VOC trap was spiked with 300 ng of tetraline as an internal standard (ISTD), eluted with 750 μl of dichloromethane, and analyzed by GC-MS (21). The labels in both (A) and (B) represent the following: 1,cis-3-hexen-1-ol; 2, trans-β-ocimene; 3,cis-3-hexenyl acetate; 4, linalool; 5, terpineol; 6,cis-3-hexenyl butyrate; 7, methyl salicylate; 8,cis-α-bergamotene; and 9, trans-β-farnesene. Asterisks designate compounds whose emission was significantly [P < 0.05; Bonferroni-corrected Fisher's protected least significant difference (LSD) test of an analysis of variance (ANOVA)] elevated in comparison with undamaged control plants.

The emitted VOCs common to all three herbivores are derived from three biosynthetic pathways. Green leaf volatiles (cis-3-hexene-1-ol, cis-3-hexenyl acetate, andcis-3-hexenyl butyrate; 1, 3, and 6, respectively, in Fig. 1) derived from the octadecanoid pathway are known to be emitted rapidly after damage (11, 13) but are not specific to plants attacked by herbivores (2). The terpenoids (trans-β-ocimene,cis-α-bergamotene, and trans-β-farnesene; 2, 8, and 9, respectively, in Fig. 1) are emitted more slowly, typically 24 hours after attack (11, 13). Only 5 of the 24 attacked plants emitted linalool (4 in Fig. 1), which is consistent with previous findings that only some N. attenuata genotypes are linalool emitters (13). All attacked plants showed significantly elevated emissions of cis-α-bergamotene, which is known to be elicited in N. attenuata when a suite of seven to eight fatty acid–amino acid conjugates that occur in the oral secretions of both M. sexta and M. quinquemaculata larvae are introduced to leaf wounds (22) as well as when plants are treated with methyl jasmonate (MeJA) (13). Finally, the emission of the shikimate-derived methyl salicylate (MeSA; 7 in Fig. 1) was significantly elevated in the headspace of all attacked plants. Because all of these compounds have been identified in the headspace of other herbivore-infested plants (2, 5, 23,24), they may function as universal signs of herbivore damage, and we hypothesized that increasing the emission of single compounds in the context of the plants' natural background emission should attract natural enemies in nature.

To test this hypothesis, we mimicked the herbivore-induced emission of individual compounds from each biosynthetic class. To mimic the volatile release, we applied 200 μg of either MeJA, MeSA,cis-3-hexene-1-ol, trans-β-ocimene,racemic linalool, or cis-α-bergamotene in 20 μl of lanolin paste to the stems of flowering plants in a natural population (18). Controls were treated with 20 μl of pure lanolin. To determine the fidelity of the mimicry, we trapped the VOCs from whole plants and found the emission to be very similar to those of herbivore-infested or MeJA-treated plants (Fig. 2A: data shown for control, MeJA-, and cis-α-bergamotene–treated plants). Application of individual compounds (with the exception of MeJA, which elicited a majority of herbivore-induced VOCs) resulted in increased emissions of only the applied compound, which was released in quantities within the range of emissions observed in herbivore-infested plants (Fig. 2A) (25).

Figure 2

VOCs and their influence on the survivorship of hornworm eggs and oviposition rates of adult M. quinquemaculata in nature. (A) Mean (±SEM) whole-plant emissions of nine VOCs from plants (four per treatment) treated with lanoline paste (CTRL), methyl jasmonate (MeJA), andcis-α-bergamotene (bergamotene) 24 hours after application. Plants were enclosed in an open 2-liter conical plastic chamber, and the headspace air was pulled through a charcoal trap for 7 hours at a flow rate of 450 to 500 ml min−1 by a vacuum pump (13). The “bergamotene” treatment (as well as thecis-3-hexene-1-ol, trans-β-ocimene, linalool, and methyl salicylate treatments) (25) only increased the emission of the applied compound. The dotted line denotes the highest natural cis-α-bergamotene emission rate measured from a plant damaged by a hornworm in this study. (B) Mean (±SEM) percentage survival of M. sexta eggs on plants releasing elevated amounts of single VOCs: 1, cis-3-hexene-1-ol; 2,trans-β-ocimene; 4, linalool; 7, methyl salicylate; and 8, cis-α-bergamotene. In the “clumped” experimental design, plants treated with the same VOC were within 3 to 5 m of each other, whereas in the “transect” experimental design, similarly treated plants were within 21 to 35 m of each other. (C) Mean (±SEM) M. quinquemaculataoviposition rates on MeJA-treated plants, plants under previous attack by hornworms, and plants treated with cis-3-hexenyl butyrate (6) and linalool (4). Asterisks represent significant differences from control (CTRL) plants (*, P < 0.05; **,P < 0.01; ***, P < 0.001) as determined by Fisher's protected LSD test of an ANOVA.

We used M. sexta eggs to measure predation rates to avoid the confounding influence of direct defenses elicited by herbivore feeding. A single hornworm larva can completely defoliate 1 to 10 reproductively mature N. attenuata plants in the course of its development, and hornworm larvae have been responsible for most of the leaf area lost to insect herbivores in six N. attenuatapopulations monitored in southwest Utah over the past 2 years.Manduca eggs are typically laid singly on plants, but clusters of four to seven eggs on a single leaf are occasionally found. To measure predation rates, we fixed five M. sexta eggs to the underside of both the second and third stem leaves (26) of 105 unattacked N. attenuataplants, using a neutral α-cellulose glue that was known not to effect changes in VOC emissions (27). Lanolin paste with and without the individual VOCs was placed on the stem between the leaves.

Plants were of the same size (30 to 40 cm), unattacked, and 3 to 5 m apart in a linear 500-m transect across a population of more than 100,000 plants. Treatments were applied so that no two plants of the same treatment were within 21 to 35 m from each other. This distance was sufficient to prevent the plumes from similarly treated plants from interacting and provided a conservative measure of the ecological function of VOC emission. In another experiment in the same population of plants, in which distances between similarly treated plants were only 3 to 5 m of each other, predation rates were 13-fold higher (Fig. 2B; data for MeJA and control treatments shown), probably because of the lack of independence of replicates within a treatment.

During the experiment, only one predator (Geocoris pallens; Heteroptera, Geocoridae) was observed feeding on the eggs and hatching larvae. Attacked eggs are emptied and hence easy to distinguish. Moreover, this predator was repeatedly observed preying on leaf bugs and flea beetles. During the previous field season, G. pallens was responsible for 95% of the M. quinquemaculata and M. sexta larvae mortality (26); no Manduca eggs on N. attenuata were parasitized. Survival of eggs and neonatant larvae was monitored for 48 hours. After 24 hours, the mortality was already significantly higher on plants treated with MeJA (36 ± 5.4%),cis-3-hexene-1-ol (33.8 ± 6.7%), linalool (37.5 ± 5.3%), and cis-α-bergamotene (33.3 ± 5.1%), whereas mortality on plants treated with MeSA (21.2 ± 5.8%) andtrans-β-ocimene (28.7 ± 5.8%) did not differ significantly from mortality on untreated control plants (16.7 ± 4.4%; Fig. 2B).

To determine if the herbivore-induced release of VOCs influenced the oviposition behavior of native adult Manduca moths, we established a new transect with unattacked plants across the same populations with the following treatments (28): one to four foraging M. quinquemaculata first to third instar larvae per plant, MeJA to elicit a majority of the herbivore-induced VOCs, one green leaf VOC (cis-3-hexenyl butyrate) and one terpenoid (linalool), and a control lanolin treatment. At the time of the experiment, M. quinquemaculata adults were ovipositing, and the number of eggs deposited on each plant was counted and removed every second day for 14 days. Lanolin treatments were refreshed every 48 hours to maintain VOC emissions. Over a 1-week period, moths laid fewer eggs per plant on plants under attack by caterpillars (0.31 ± 0.08), on plants treated with MeJA (0.53 ± 0.1), and on plants treated with and releasing linalool (0.42 ± 0.2). Plants treated with cis-3-hexenyl butyrate (0.64 ± 0.2) received the same number of eggs as control plants (1.0 ± 0.2) (Fig. 2C).

By releasing VOCs after herbivore attack, a plant can profoundly influence both oviposition and predation rates in nature and thereby influence both “bottom-up” as well as “top-down” control over its herbivore populations. The emission of linalool alone caused a 2.4-fold reduction in oviposition rate. Daily predation rates on plants releasing VOCs, extrapolated to 1 week, were 4.9 to 7.5 times higher than those observed on control plants; hence, the top-down effects in this experiment were more strongly influenced by VOC emissions. The multiplicative effect, calculated from both the bottom-up and the top-down components of this indirect defense (25) could reduce the numbers of the plant's most significant insect folivore,M. quinquemaculata, by 91.7 (MeJA treatment) and 94.5% (linalool treatment). Herbivores as well as predators appear to use the same volatile signals, suggesting that plants are under strong selection to release them.

  • * To whom correspondence should be addressed. E-mail: baldwin{at}


View Abstract

Stay Connected to Science

Navigate This Article