Significance of terpenoids in induced indirect plant defence against herbivorous arthropods


R. Mumm. Fax: 0031317484821; e-mail:


Many plants respond to herbivory by arthropods with an induced emission of volatiles such as green leaf volatiles and terpenoids. These herbivore-induced plant volatiles (HIPVs) can attract carnivores, for example, predators and parasitoids. We investigated the significance of terpenoids in attracting herbivores and carnivores in two tritrophic systems where we manipulated the terpenoid emission by treating the plants with fosmidomycin, which inhibits one of the terpenoid biosynthetic pathways and consequently terpenoid emission.

In the ‘lima bean’ system, volatiles from spider-mite-infested fosmidomycin-treated plants were less attractive to the predatory mite Phytoseiulus persimilis than from infested control plants. In the ‘cabbage’ system, fosmidomycin treatment did not alter the attractiveness of Brussels sprouts to two Pieris butterflies for oviposition. The parasitoid Cotesia glomerata did not discriminate between the volatiles of fosmidomycin-treated and water-treated caterpillar-infested cabbage. Both P. persimilis and C. glomerata preferred volatiles from infested plants to uninfested ones when both were treated with fosmidomycin.

Chemical analysis showed that terpenoid emission was inhibited more strongly in infested lima bean plants than in Brussels sprouts plants after fosmidomycin treatment.

This study shows an important role of terpenoids in the indirect defence of lima bean, which is discussed relative to the role of other HIPVs.


Plants produce a plethora of volatile compounds, which are involved in various interactions with their environment. Terpenoids (isoprenoids) represent the largest and the most diverse group of volatiles that are emitted by plants, sometimes in substantial amounts (Pichersky & Gershenzon 2002; Dudareva & Pichersky 2006; Rodríguez-Concepcion 2006). The volatile fraction of terpenoids predominantly consists of the hemiterpene isoprene (C5), monoterpenes (C10) and sesquiterpenes (C15) and their derivatives such as homoterpenes (C11 and C16). Monoterpenes and sesquiterpenes are synthesized by the condensation of two or three isoprene units, respectively (Cheng et al. 2007). Because of their physicochemical properties, such as volatility, reactivity, toxicity and aroma, many protective functions against abiotic and biotic factors have been ascribed to terpenoids (Holopainen 2004). Terpenoids are involved in plant–pollinator interactions and have important functions in plant defence against herbivores (Dicke 1994; Paré & Tumlinson 1999; Pichersky & Gershenzon 2002; Dudareva et al. 2006; Hilker & Meiners 2006; Keeling & Bohlmann 2006; Mumm & Hilker 2006; Van Schie, Haring & Schuurink 2006; Cheng et al. 2007).

It has been demonstrated in numerous studies that the egg deposition or feeding by herbivorous arthropods, such as insects and mites, can induce alterations in the emission of volatile blends in many plant species (reviewed by Dicke & van Loon 2000; Dicke & Van Poecke 2002; D'Alessandro & Turlings 2006; Hilker & Meiners 2006; Mumm & Hilker 2006). These herbivore-induced changes in the volatile composition can be exploited by natural enemies of the herbivores, such as predators and parasitoids, to locate plants infested with their prey/host species. The attraction of natural enemies to herbivore-induced volatiles is accepted to be an important mechanism on how plants can indirectly defend themselves (e.g. Dicke & van Loon 2000; Arimura, Kost & Boland 2005; Hilker & Meiners 2006). Induced plant volatiles can also act directly against herbivores by repelling them, but alternatively, herbivores can also use these to locate suitable host plants (e.g. De Moraes, Mescher & Tumlinson 2001; Shiojiri et al. 2002; Horiuchi et al. 2003; Bruinsma et al. 2007).

The herbivore-induced changes in the volatile blends can be quantitative, that is, volatiles that are also present in non-induced plants are emitted in larger total amounts, their relative abundance changes, or both (e.g. Mumm et al. 2003; Bukovinszky et al. 2005). On the other hand, herbivory can also induce de novo production of compounds in many plant species, resulting in qualitative changes in the composition of the emitted blend (Turlings et al. 1998; Dicke et al. 1999; Krips et al. 1999; Leitner, Boland & Mithöfer 2005). Typical volatile plant compounds that are induced by herbivory are C6-alcohols, -aldehydes and -acetates (so-called green leaf volatiles, GLVs); methyl salicylate (MeSA); phenylpropanoids; and various monoterpenes and sesquiterpenes, as well as two homoterpenes, (E)-4,8-dimethyl-1,3,7-nonatriene (DMNT) and (E,E)-4,8,12-trimethyltrideca-1,3,7,11-tetraene (TMTT) (Arimura et al. 2005; Dudareva et al. 2006). GLVs are fatty acid derivatives resulting from the conversion of linolenic and linoleic acid released from damaged plasma membranes through the lipoxygenase pathway (Arimura et al. 2005; D'Auria et al. 2007). All terpenoids are synthesized via the cytosolic mevalonate (MVA) pathway or the methylerythritol 4-phosphate (MEP) pathway, which is located in the plastids (Aharoni, Jongsma & Bouwmeester 2005; Rodríguez-Concepcion 2006; Cheng et al. 2007). Both terpenoid pathways can synthesize isopentenyl diphosphate (IPP) and its isomer dimethylallyl diphosphate (DMAPP), which are the central intermediates for the biosynthesis of terpenes. In a wide range of cyclization and rearrangement steps, precursors are converted into the parent skeletons of the respective terpenes. Finally, the parent skeletons are converted into the myriad of different terpenoids by a big variety of transformations like oxidations, isomerizations and conjugations (Gershenzon & Kreis 1999). Monoterpenes and diterpenes are synthesized via the MEP pathway, whereas sesquiterpenes are produced by the MVA pathway. Recently, it was demonstrated that there is some exchange of IPP and DMAPP between the two pathways, indicating that the pathways are not strictly separated (Rodríguez-Concepcion 2006).

Much attention has been paid to the role of monoterpenes in plant defence against herbivores, because this group of terpenes is the most abundant among volatile terpenoids. Many plants, such as conifers, show a high constitutive emission of monoterpenes; others, however, show a strong induced emission of monoterpenes after herbivory. Constitutive and inducible defences are supposed to be negatively correlated (Dicke & Van Poecke 2002; Koricheva, Nykänen & Gianoli 2004). Terpenes may act as feeding deterrents for insects, but also, the oviposition behaviour of herbivorous arthropods can be influenced by the presence or absence of terpenes (Aharoni et al. 2003; Tripathi et al. 2003; Isman 2006). Furthermore, it has been shown that terpenes can attract predators and parasitoids. Many of these studies used single terpenes or mixtures, which are then applied to some sort of dispensing material. Although this approach might work well in some cases, in others, it failed. One reason is that by applying single synthetic compounds, one neglects that background odours of the plants play an important role in bringing terpenes into the right ‘context’ (Pettersson 2001; Pettersson, Birgersson & Witzgall 2001; Buitenhuis et al. 2005; Mumm & Hilker 2005). Other studies showed that terpenes can mask the attractiveness of other plant volatiles to herbivorous insects and parasitoids (Yamasaki, Sato & Sakoguchi 1997; Turlings & Fritzsche 1999). This shows that many questions regarding the role of terpenes in the interplay with other plant volatiles in direct and indirect plant defence against herbivorous arthropods still remain to be answered.

Here, we address the role of terpenes in induced indirect defence by applying fosmidomycin, an inhibitor of the MEP pathway, to plants. The antibiotic fosmidomycin blocks 1-deoxy-D-xylulose 5-phosphate reductoisomerase, an enzyme catalysing an early step in the MEP pathway (Zeidler et al. 1998). Fosmidomycin has been shown to effectively inhibit the biosynthesis and emission of volatile monoterpenes and partly also sesquiterpenes (Jux, Gleixner & Boland 2001; Copolovici et al. 2005; Hampel, Mosandl & Wüst 2005; Bartram et al. 2006). This makes it an interesting tool to study the role of monoterpenes in direct and indirect plant defence.

We used two tritrophic model systems to study the effect of treating plants with fosmidomycin on the behaviour of herbivores and their natural enemies. In the ‘lima bean’ system, lima bean plants, Phaseolus lunatus (Fabaceae), are used as host plants for the herbivorous two-spotted spider mite Tetranychus urticae (Acari, Tetranychidae). The predatory mite Phytoseiulus persimilis (Acari, Phytoseiidae) was used as carnivorous species. The ‘cabbage’ system consisted of Brussels sprouts plants, Brassica oleracea (Brassicaceae), which are frequently attacked by the specialized large cabbage white butterfly Pieris brassicae and the small cabbage white butterfly Pieris rapae (Lepidoptera, Pieridae). P. brassicae and P. rapae caterpillars are parasitized by the larval parasitoid Cotesia glomerata (Hymenoptera, Braconidae), a gregarious endoparasitoid that prefers to parasitize young larval instars of particularly P. brassicae (Geervliet et al. 2000).

It has previously been demonstrated that C. glomerata is attracted to caterpillar-infested Brussels sprouts, and P. persimilis to volatiles from spider-mite-infested lima bean, respectively (Dicke et al. 1990a; Mattiacci, Dicke & Posthumus 1994; Geervliet 1997; De Boer 2004). In both plant species, volatiles are induced after herbivory. However, in Brussels sprouts, the changes in terpene emission after herbivore feeding are mainly quantitative (Blaakmeer et al. 1994; Mattiacci et al. 1994; Bukovinszky et al. 2005), whereas in lima bean plants, spider mites induce a distinct de novo production of terpenoids such as (E)-β-ocimene, linalool, DMNT and TMTT (Dicke et al. 1990b, 1999; De Boer, Posthumus & Dicke 2004; Shimoda et al. 2005).

Recently, Smid et al. (2002) showed that the antennae of C. glomerata respond to limonene, a monoterpene that was identified in the headspace of P. brassicae-infested Brussels sprouts. Furthermore, the predatory mite P. persimilis is known to behaviourally respond to terpenoids, but MeSA, which is also strongly induced in lima bean by spider mite feeding, is also strongly attractive to the predatory mites, making it difficult to estimate the relative contribution of terpenoids (Dicke et al. 1990b; De Boer & Dicke 2004a,b).

By applying fosmidomycin, we expected that particularly the emission of monoterpenes and the homoterpene TMTT is inhibited. Other inducible compounds, like GLVs and MeSA, should not be affected by the fosmidomycin treatment. We expected that if terpenoids play a significant role in indirect plant defence, the infested fosmidomycin-treated plants should be less attractive to predators and parasitoids than the untreated ones. On the other hand, herbivores are expected to prefer to lay eggs on plants with less terpenoid emission. We also collected the headspace volatiles from infested Brussels sprouts and lima bean plants after fosmidomycin treatment to assess treatment effects on the chemical changes in headspace composition for the two plant species.


Plants and arthropods


Lima bean plants, P. lunatus L. cv. Sieva (Fabaceae), and Brussels sprouts plants, B. oleracea var. gemmifera L. cv. Cyrus (Brassicaceae), were grown from seed in separate greenhouse compartments in plastic pots (11 × 11 × 11 cm) at 24 ± 4 °C, 60 ± 20% relative humidity (RH) and a 16 h light/8 h dark photoperiod. Lima bean plants were used in experiments when the primary leaves had fully expanded, which was 12–16 d after sowing. Experiments were conducted with 6- to 8-week-old Brussels sprout plants.


A colony of the two-spotted spider mite T. urticae Koch (Acari, Tetranychidae) was maintained on lima bean plants in another greenhouse compartment under the same conditions as described for the lima bean culture. A continuous rearing of the large cabbage white, Pieris brassicae L. (Lepidoptera, Pieridae), and the small cabbage white, Pieris rapae L. (Lepidoptera, Pieridae), was maintained on Brussels sprouts in a climatized room at 21 ± 1 °C, 60 ± 10% RH and a 16 h light/8 h dark photoperiod.


A colony of the predatory mite P. persimilis Athias-Henriot (Acari, Phytoseiidae) was kept on spider-mite-infested lima bean leaves in a climate cabinet at 23 ± 1 °C, 60 ± 10% RH and continuous light. The parasitoid C. glomerata L. (Hymenoptera, Braconidae) was reared on P. brassicae caterpillars feeding on Brussels sprouts under similar environmental conditions as the uninfested cabbage plants. For experiments, C. glomerata pupae were collected and kept in a cage in a climate cabinet (23 ± 1 °C, 60 ± 10% RH and a 16 h light/8 h dark photoperiod). Emerging wasps were provided with water and honey. Male and female wasps were kept together until the experiment.

Behavioural bioassays

Olfactometer and windtunnel tests

A Y-tube olfactometer set-up similar to the one described by Takabayashi & Dicke (1992) was used to investigate the olfactory choice of predatory mites to two odour sources. A metal wire was positioned at the middle of the olfactometer tube. The side arms were both connected to 5 L Duran glass jars (Duran, Mainz, Germany) containing the odour sources. Pressurized air was filtered over activated charcoal, and was led into the jars at the top, and the jars were left at the bottom towards the olfactometer. An airflow in each arm of 4 L min−1 was controlled by a flowmeter (Brooks Instr., Veenendaal, the Netherlands). Air was extracted at the base of the Y-tube at 8 L min−1. Experiments were conducted at 23 ± 2 °C and 60 ± 5 µmol m−2 s−1 photosynthetically active radiation (PAR). In order to correct for unforeseen asymmetry in the set-up, the position of the odour sources was switched after five tested predators. After testing 10 predators, the odour sources were replaced with new ones. Every experiment was replicated on four different days.

To enhance the responsiveness of predatory mites to plant volatiles, all predators had been starved for 2 h prior to release in the olfactometer, by confining them individually in Eppendorf tubes (Sarstedt, Nümbrecht, Germany). The tubes with the predators were placed into the experimental room to acclimatize to the new environment. Predators were individually released on the iron wire in the basal tube, and their behaviour was observed for 10 min. Predatory mites that reached a gauze mesh at the middle of the side arm within this period were recorded as having made a ‘choice’. Mites that did not make a choice were recorded as ‘no choice’. Each predator was used only once.

Behavioural choice experiments with the parasitoid C. glomerata were carried out with a windtunnel set-up (25 ± 1 °C, 60 ± 10% RH and 35 µmol m−2 s−1 PAR) as described by Geervliet, Vet & Dicke (1994). The wind speed was adjusted to 0.2 m s−1. Two-choice experiments were conducted by placing a treated leaf and a respective control leaf at the upwind end of the tunnel.

Naïve C. glomerata females were separated from males 3 h prior to the experiment and were transferred to another cage, which was placed in the experimental room to acclimatize to the new environment. The wasps were individually introduced into the windtunnel on an infested cabbage leaf piece from which the caterpillars and their products had been carefully removed. The wasps were allowed to walk onto the leaf pieces themselves. The leaf piece with the wasp was placed at the middle of the release cylinder, which was 60 cm downwind from the two odour sources. As soon as the parasitoid had left the leaf for some seconds, the leaf was carefully removed using tweezers without disturbing the parasitoid. The flight behaviour of the wasps was observed. Flights that resulted in a landing on one of the two odour sources were recorded as a ‘choice’. Parasitoids that did not leave the release cylinder or landed on other parts of the windtunnel within 10 min were recorded as ‘no choice’. Every parasitoid was used only once. In order to correct for unforeseen asymmetry in the set-up, the position of the odour sources was switched after five tested parasitoids. After testing 10 parasitoids, the odour sources were replaced with new ones. Every experiment was replicated on four different days.

Oviposition preference test

Freshly emerged P. brassicae and P. rapae adults were each transferred to a large cage (67 × 100 × 75 cm) in a greenhouse compartment at 24 ± 4 °C, 60 ± 20% RH and a 16 h light/8 h dark photoperiod. Butterflies were provided with a 10% sucrose solution. Three to five days after emergence, one male and one female butterfly were transferred to oviposition cages (67 × 50 × 75 cm) in the same greenhouse compartment. In addition to natural daylight, the cages were illuminated by sodium vapour lamps (SON-T, 500 W, Philips, Eindhoven, the Netherlands) from 1000 to 1600 h. At 48 h prior to the experiment, a single untreated Brussels sprouts leaf was placed in each cage as an oviposition substrate. After 6 h, the leaf was removed. On the experimental day between 0900 and 1000 h in the morning, a fosmidomycin-treated Brussels sprouts leaf and a respective control leaf were introduced into every cage. The treated and the control leaf always originated from the same plant to minimize intraspecific variation between treatment and control. Each butterfly couple was also provided with a 10% sucrose solution. The leaves were placed in an upright position approximately 40 cm apart from each other. P. brassicae was allowed to lay eggs on the two leaves for 24 h, whereas P. rapae could lay eggs for 6 h. The leaves were then removed, and the eggs layed were counted. P. rapae lays individual eggs, which are distributed over the leaves. In general, P. rapae starts to deposit eggs soon after being exposed to leaves. Therefore, 6 h was sufficient to obtain sufficient eggs. On the other hand, P. brassicae lays egg batches, which can last for several hours. Therefore, we chose to confine P. brassicae for 24 h with the cabbage leaves. The experiments were conducted in several cages at the same time and on 4–5 d per treatment.

Plant treatments

For all experiments, lima bean leaves or Brussels sprouts leaves were cut under water. When leaves are cut under water, a small water droplet normally forms around the petiole, and prevents air invading the vascular system through the petiole. In this way, the leaves could be transferred in glass vials filled either with water or a 50 µm aqueous fosmidomycin solution. Fosmidomycin was purchased from Invitrogen Molecular Probes (Breda, the Netherlands). The openings of the glass vials with the leaves were thoroughly covered with Parafilm (Pekhiney Plastic Packing, Chicago, IL, USA). Only leaves that were in visually good condition after the treatment period were used for the experiments.

Lima bean leaves were treated either with a 50 µm aqueous fosmidomycin solution or with water, and then were either immediately infested with 20 adult T. urticae or left uninfested. Subsequently, the lima bean plants were kept in a climate chamber for 48 h at 23 ± 1 °C, 60 ± 5% RH and a 16 h light/8 h dark photoperiod before being used in the experiment. The leaves used for the behavioural experiments with C. glomerata were treated with fosmidomycin or with water. The fosmidomycin-treated and water-treated leaves were then immediately either infested with 50 L1 P. brassicae caterpillars or were left uninfested. The leaves were kept in a climate chamber at 23 ± 1 °C, 60 ± 5% RH and a 16 h light/8 h dark photoperiod. For oviposition experiments with butterflies, both fosmidomycin-treated and control (water-treated) cabbage leaves were kept for 24 h in the greenhouse compartment at 24 ± 4 °C, 60 ± 20% RH and a 16 h light/8 h dark photoperiod before being used in the bioassay.

Collection of headspace volatiles

Volatiles emitted from fosmidomycin-treated lima bean and Brussels sprouts leaves were collected using a dynamic headspace collection system. The leaves of one treatment were transferred to a 5 L Duran glass jar (Duran). The jar was tightly closed with a glass lid that was pressed on the jar with a metal clamp with a viton O-ring in between. The lid had an air inlet and an air outlet. Pressurized air was filtered over activated charcoal and was led into the jar with a constant flow of 70 mL min−1. Air was sucked out of the jar with 50 mL min−1 by passing through a glass tube filled with 90 mg Tenax TA (Grace-Alltech, Deerfield, IL, USA) connected to the air outlet of the jar. The slight overpressure was created to prevent that unfiltered air could invade the system. The system was purged for 30 min with cleaned air before the volatiles were trapped onto the Tenax. The flow through the jar was controlled by flowmeters (Brooks Instr.). Teflon tubing was used for all connections. Headspace collections were made in a climate chamber at 23 ± 1 °C, 60 ± 5% RH, and 90 ± 5 µmol m−2 s−1 PAR. The volatiles of one treatment and its respective control were collected simultaneously. Lima bean plants were infested with 20 spider mites per plant. Brussels sprouts were infested with 50 L1 P. brassicae caterpillars per leaf. Both the infested lima bean leaves and Brussels sprouts plants were then either placed in vials containing fosmidomycin or water as described previously. The volatiles emitted from lima bean leaves were trapped for 1 h, and for Brussels sprouts leaves, trapping time was 5 h.

Chemical analysis of headspace samples

Headspace samples were analysed with a Varian 3400 gas chromatograph (GC) (Varian, Palo Alto, CA, USA) connected to a Finnigan 95 mass spectrometer (MS) (Thermo Scientific, Waltham, MA, USA). The collected volatiles were released from the Tenax by heating the trap in a Thermodesorption Cold Trap Unit (Chrompack, Middelburg, the Netherlands) at 250 °C for 10 min, and flushing with helium at 14 mL min−1. The released compounds were cryofocused in a cold trap 0.52 mm [inner diametre (ID)] deactivated fused silica at a temperature of −85 °C. By ballistic heating of the cold trap to 220 °C, the volatiles were transferred to the analytical column (60 m × 0.25 mm ID, 0.25 µm film thickness, DB-5 ms J&W, Folsom, CA, USA). The temperature programme started at 40 °C (4 min hold) and rose at the rate of 4 °C min−1 (lima bean) or 5 °C min−1 (Brussels sprouts) to 280 °C (4 min hold). The column effluent was ionized by electron impact ionization at 70 eV. Mass scanning was carried out from 24 to 300 m/z with a scan time of 0.7 s/d and an interscan delay of 0.2 s. The compounds were identified by comparison of the mass spectra with those in the Wiley library and in the Wageningen Mass Spectral Database of Natural Products, and by checking the retention index. One peak area unit represents approximately 0.17 ± 0.05 ng.

Statistical analyses

A two-sided binomial test was used to analyse whether the behavioural choices of predatory mites and parasitoids differed from a 50:50 distribution over the two odour sources. Predatory mites and parasitoids that did not make a choice were excluded from the statistical analysis. Most individuals of both Pieris species laid eggs on both the control and the fosmidomycin-treated leaves. The number of eggs on each treatment per individual was considered as a paired sample and was analysed with the Wilcoxon signed ranks test. Amounts of headspace volatiles trapped were analysed on the basis of normalized peak area units, as determined by GC-MS analysis. Mann–Whitney U-tests were applied to test for differences between the fosmidomycin-treated and control plants for groups of compounds. P values were adjusted by the sequential Bonferroni method to correct for the family wise error rate (Holm 1979).


Behavioural response of predatory mites and parasitoids to fosmidomycin-treated plants

The predatory mite P. persimilis significantly preferred the volatiles from spider-mite-infested lima bean leaves, which where not treated with fosmidomycin, to those from spider-mite-infested fosmidomycin-treated lima bean leaves (Fig. 1). Thus, treatment with fosmidomycin reduces the attractiveness of spider-mite-induced lima bean volatiles to predatory mites. Yet, predatory mites were strongly attracted to the volatiles from fosmidomycin-treated, spider-mite-infested lima bean when offered against the volatiles from fosmidomycin-treated uninfested lima bean plants (Fig. 1). This shows that although fosmidomycin reduces the attraction of predators to spider-mite-infested lima bean leaves, it does not eliminate the emission of attractive volatiles altogether.

Figure 1.

Effect of fosmidomycin treatment of lima bean plants on the response of Phytoseiulus persimilis to lima bean volatiles. Upper panel: response to spider-mite-infested lima bean treated with 50 µm fosmidomycin solution (grey bar) or spider-mite-infested lima bean (white bar) n = 117. Lower panel: response to spider-mite-infested lima bean treated with 50 µm fosmidomycin solution (grey bar) or uninfested lima bean treated with 50 µm fosmidomycin solution (white bar) n = 80. The number behind each bar is the percentage of predatory mites that did not make a choice. Choices between odour sources were analysed with a two-sided binomial test (**P < 0.01, ***P < 0.001).

The parasitoid C. glomerata did not discriminate between the volatiles from fosmidomycin-treated and P. brassicae-infested cabbage leaves, and the volatiles from control leaves, that is, water-treated P. brassicae-infested cabbage leaves (Fig. 2). Thus, fosmidomycin treatment did not change the attractiveness of caterpillar-induced cabbage volatiles to those from the untreated cabbage. C. glomerata females significantly preferred the volatiles from fosmidomycin-treated P. brassicae-infested leaves to those from fosmidomycin-treated uninfested plants (Fig. 2). This shows that the volatiles that are used by the parasitoids to discriminate between infested and uninfested plants are still present after the treatment with fosmidomycin.

Figure 2.

Effect of fosmidomycin treatment of Brussels sprouts on the response of Cotesia glomerata to Brussels sprouts odours. Upper panel: response to caterpillar-infested Brussels sprouts treated with 50 µm fosmidomycin solution (grey bar) or caterpillar-infested Brussels sprouts (white bar) n = 45. Lower panel: response to caterpillar-infested Brussels sprouts treated with 50 µm fosmidomycin solution (grey bar) or uninfested Brussels sprouts treated with 50 µm fosmidomycin solution (white bar) n = 36. The number behind each bar is the percentage of parasitoids that did not make a choice. Choices between odour sources were analysed with a two-sided binomial test (n.s., P > 0.05; ***P < 0.001).

Oviposition preference of cabbage white butterflies after application of fosmidomycin

The average number of eggs deposited by the large cabbage white butterfly P. brassicae was not significantly different between fosmidomycin-treated cabbage leaves and untreated controls (Wilcoxon signed ranks test, Z = −0.812, P > 0.05, n = 22) (Fig. 3). Neither did the small cabbage white butterfly P. rapae discriminate between fosmidomycin-treated and untreated cabbage leaves (Wilcoxon signed ranks test, Z = −0.975, P > 0.05, n = 23) (Fig. 3). The differences in the average number of eggs laid by the two species probably result from the differences in the oviposition behaviour and the differences in time the butterflies were allowed to oviposit.

Figure 3.

Oviposition of Pieris brassicae and Pieris rapae on Brussels sprout plants either treated with 50 µm fosmidomycin (black bars) or left untreated (white bars). Mean number of eggs per female + standard error (n.s., P > 0.05, Wilcoxon signed ranks test).

Analysis of headspace volatiles

In the headspace of caterpillar-infested Brussels sprouts plants, 31 compounds were detected, and 12 compounds were identified in the headspace of spider-mite-infested lima bean leaves. Both plant species were treated either with fosmidomycin or with water as control (Table 1). All compounds that were emitted by fosmidomycin-treated leaves were also detected in the headspace of control leaves. Therefore, the compounds that were detected in at least half of the control samples are depicted in Table 1. Herbivore-induced volatiles of excised leaves of lima bean and Brussels sprouts resembled those emitted by whole plants (Bukovinszky et al. 2005; Pinto et al. 2007).

Table 1.  List of volatile compounds detected in the headspace of Brussels sprouts and lima bean after different treatments
 Brussels sproutsLima bean
Fosmidomycin (n = 3)Control (n = 5)Fosmidomycin (n = 5)Control (n = 5)
  1. Median and interquartile range (in parentheses) of normalized peak area units are given.

  2. nd, Compounds were neither detected in fosmidomycin-treated plants nor in the controls; tr, trace amounts in single samples.

 11-Penten-3-ol68 (54–103)105 (21–206)ndnd
 23-Pentanone35 (20–207)131 (27–249)ndnd
 33-Pentanol56 (40–123)128 (83–157)ndnd
 42-Methylbutanalndnd0 (0–3)5 (0–8)
 53-Methylbutanalndnd9 (6–12)14 (6–20)
 6(Z)/(E)-2-Penten-1-ol2 (1–4)7 (4–14)ndnd
 7(Z)-3-Hexenal12 (6–49)34 (34–95)ndnd
 8Hexanal8 (7–9)12 (4–16)1 (0–4)8 (2–8)
 9(E)-2-Hexenal0 (0–6)7 (0–45)ndnd
10(Z)-3-Hexen-1-ol237 (203–884)650 (263–899)1 (0–5)1 (1–2)
112-Methyl-2-cyclopenten-1-one0 (0–2)5 (0–8)ndnd
12(Z)-2-Penten-1-ol acetate79 (71–85)117 (71–199)ndnd
13Pentyl acetate3 (2–3)4 (0–5)ndnd
14α-Thujene110 (67–130)158 (54–178)ndnd
15α-Pinene25 (20–37)44 (12–83)ndnd
163-Ethyl-1,5-octadiene6 (6–7)10 (8–13)ndnd
17Sabinene437 (271–546)716 (207–841)ndnd
18β-Pinene27 (17–33)39 (10–80)ndnd
19Myrcene103 (62–116)151 (94–201)ndnd
20(Z)-3-Hexen-1-ol acetate2630 (2500–2855)3680 (2930–5320)20 (0–41)33 (0–83)
21Hexyl acetate42 (21–48)36 (33–79)ndnd
22Limonene258 (158–284)343 (269–547)ndnd
23β-Phellandrene4 (2–4)10 (5–12)ndnd
241,8-Cineole174 (98–181)170 (104–249)ndnd
25(Z)-β-Ocimenendnd0 (0–0)18 (14–59)
26Phenylacetaldehyde0 (0–1)7 (0–12)ndnd
27(E)-β-Ocimenendnd0 (0–0)280 (210–958)
28Salicylaldehyde0 (0–0)3 (0–6)ndnd
29γ-Terpinene0 (0–0)5 (3–17)ndnd
30trans-4-Thujanol0 (0–5)11 (0–19)ndnd
31Linaloolndnd0 (0–0)31 (11–32)
32Nonanal16 (8–18)58 (17–81)ndnd
33(E)-4,8-dimethyl-1,3,7-nonatrienendtr57 (15–84)573 (405–820)
342-Methyl-6-methylene-1,7-octadiene-3-one0 (0–2)5 (0–13)ndnd
35Methyl salicylate5 (4–7)5 (3–12)3 (2–77)103 (30–123)
36α-Gurjunene4 (3–5)3 (0–30)ndnd
37Longifolene7 (6–8)6 (0–10)ndnd
38(E)-β-Caryophyllenendnd2 (0–3)10 (8–18)
39(E,E)-4,8,12-trimethyltrideca-1,3,7,11-tetraenendnd0 (0–0)51 (42–67)

The emission of the monoterpene hydrocarbons (Z)- and (E)-β-ocimene, the oxygenated monoterpene linalool and the homoterpene TMTT was completely inhibited by fosmidomycin in lima bean (Table 1, Fig. 4). The second homoterpene DMNT was emitted in significantly lower amounts by fosmidomycin-treated lima bean leaves compared with the water-treated controls (Fig. 4, Table 1). The only sesquiterpene, that is, (E)-β-caryophyllene, was emitted in lower amounts by fosmidomycin-treated lima bean leaves than by water-treated leaves, but this was not statistically significant. The emission of GLVs (hexanal, (Z)-3-hexen-1-ol, (Z)-3-hexen-1-ol acetate) and MeSA was not significantly affected by fosmidomycin treatment (Fig. 4, Table 1).

Figure 4.

Emission of volatile compound classes by lima bean plants infested with 50 Tetranychus urticae caterpillars and that were either treated with 50 µm fosmidomycin (grey bars) or left untreated (white bars). Weighted mean and standard error of normalized peak area units are given. n.s., P > 0.05; *P > 0.05, Mann–Whitney U-test, P values are corrected by sequential Bonferroni method (Holm 1979).

In contrast to the situation in lima bean, fosmidomycin did not inhibit the emission of monoterpenes in Brussels sprouts leaves significantly (Fig. 5, Table 1). However, fosmidomycin-treated Brussels sprouts leaves emitted lower amounts of monoterpenes than the water-treated leaves. Furthermore, the emission of GLVs was reduced in fosmidomycin-treated plants although there was a strong variability in the emission of these compounds (Fig. 5, Table 1). The homoterpene DMNT was only detected in trace amounts in a few samples, and TMTT was not found in the headspace of Brussels sprouts (Table 1).

Figure 5.

Emission of volatiles compound classes by Brussels sprouts plants infested with 50 Pieris brassicae caterpillars and that were either treated with 50 µm fosmidomycin (grey bars) or left untreated (white bars). Weighted mean and standard error of normalized peak area units are given. n.s., P > 0.05, Mann–Whitney U-test, P values are corrected by sequential Bonferroni method (Holm 1979).


The treatment with fosmidomycin affected the terpene emissions in lima bean and to a lesser extent in Brussels sprouts (Figs 4 and 5). Likewise, the effects on carnivore behaviour were more pronounced in the lima bean than in Brussels sprouts (Figs 1 and 2). We used fosmidomycin to block the emission of terpenoids that are synthesized via the plastidial MEP pathway. Therefore, we expected especially the emission of monoterpenes and the homoterpene TMTT, which is a derivative of the diterpene geranyllinalool, to be inhibited (Boland et al. 1998).

The concentration of fosmidomycin we used was shown to be effective in inhibiting isoprenoid emission when fed through the petiole (Barta & Loreto 2006; Bartram et al. 2006). The incubation period with fosmidomycin differed between 24 h in Brussels sprouts and 48 h in lima bean, because we knew from previous experiments that parasitoids and predatory mites are attracted to induced volatiles after 24 and 48 h of infestation, respectively. It is not likely that this difference in incubation time is responsible for the differences in inhibition because fosmidomycin can effectively inhibit the isoprenoid emission already after a few hours (Loreto & Velikova 2001; Barta & Loreto 2006; Bartram et al. 2006). In lima bean leaves, the emission of no other compound except for terpenoids was significantly affected by fosmidomycin treatment compared with the controls, suggesting that the plants did not severely suffer from the treatment.

The headspace data for lima bean plants indicate that the herbivore-induced de novo production of monoterpenes and TMTT relies completely on the MEP pathway. Interestingly, the allocation of precursors of DMNT (produced through the cytosolic MVA pathway) is more plastic (Bartram et al. 2006). When plants are not stressed, precursors of mainly the MVA pathway but partly also from the MEP pathway are assembled into DMNT. If the supply of precursors from the MEP pathway was blocked, this did not reduce the constitutive emission of DMNT because of an increasing assemblage of MVA-derived precursors (Bartram et al. 2006). However, when the emission of DMNT is induced in lima bean plants by herbivory or treatment with jasmonic acid (JA), the increased demand of precursors is covered by the MEP pathway (Piel et al. 1998; Jux et al. 2001). Our results show that in spider-mite-infested lima bean plants, the MVA pathway is not able to fully compensate for a lack of precursors from the MEP pathway after fosmidomycin treatment, resulting in a reduced emission of DMNT and (E)-β-caryophyllene (Fig. 4). In turn, this indicates that spider mite feeding on lima bean does not induce the MVA pathway, but particularly the MEP pathway, similar to what was demonstrated for JA treatment. However, Jux et al. (2001) found no significant reduction in DMNT emission after JA and fosmidomycin treatment, suggesting that the mechanisms in terpenoid induction and regulation in response to spider-mite infestation and JA application are similar but not identical (see also Dicke et al. 1999).

The lima bean system

The predatory mite P. persimilis significantly preferred the volatiles of water-treated, infested lima bean plants to those of fosmidomycin-treated ones (Fig. 1). On the other hand, the volatiles from spider-mite-infested plants were still more attractive to predatory mites compared with the uninfested ones when both were treated with fosmidomycin (Fig. 1). This shows that the absence of certain terpenoids in the headspace makes infested lima bean clearly less attractive to predatory mites, but still other chemical cues than terpenoids are apparently used to discriminate between infested and uninfested lima beans. It has been shown that MeSA plays a crucial role in the attraction of predatory mites (Dicke et al. 1999; De Boer & Dicke 2004a,b). MeSA is induced after spider-mite infestation (e.g. Dicke et al. 1999), but as expected, fosmidomycin did not affect its emission (Fig. 4). De Boer et al. (2004) showed that offering MeSA as an alternative odour source reduced the preference of predatory mites to spider-mite-infested lima bean volatiles. In addition, when MeSA was added to the volatile blends induced by JA or feeding by Spodoptera exigua, which were similar to spider-mite-induced volatiles but lacking MeSA, they became more attractive to predatory mites than those of spider-mite-induced plants. The attractive effect of MeSA was thereby based on qualitative differences rather than quantitative changes (De Boer & Dicke 2004a,b, 2005; De Boer et al. 2004).

Despite the presence of MeSA, predatory mites preferred the volatiles of untreated lima bean leaves to fosmidomycin-treated ones, thus demonstrating the relative importance of terpenoids for the predatory mites. P. persimilis and other predatory mites are attracted to several synthetic terpenoids, such as linalool, (E)-β-ocimene or DMNT, when offered individually (e.g. Dicke et al. 1990b; Shimoda et al. 2005). Interestingly, TMTT, although not attractive to predatory mites as a pure synthetic compound, did affect the attractiveness of a mixture of herbivore-induced lima bean plants (De Boer et al. 2004). This suggests that the role of terpenoids in indirect plant defence depends on the presence and composition of certain background volatiles as was also demonstrated for other tritrophic systems. For example, Mumm & Hilker (2005) showed that the egg parasitoid Chrysonotomyia ruforum (Hymenoptera, Eulophidae) only responded to the combination of the sesquiterpene (E)-β-farnesene and background volatiles of pines. The parasitoid's response was dependent on the concentration of (E)-β-farnesene, suggesting that this parasitoid uses the contrast between (E)-β-farnesene and the background odour, that is, when the compound was experienced in the ‘right’ chemical context (Mumm & Hilker 2005, 2006; Hilker & Meiners 2006).

The cabbage system

In oviposition choice experiments, we tested whether P. brassicae or P. rapae avoided or preferred intact Brussels sprout leaves treated with fosmidomycin to water-treated control leaves. Both P. brassicae and P. rapae did not discriminate between fosmidomycin-treated and water-treated cabbage leaves (Fig. 3). Neither did the treatment with fosmidomycin reduce the attractiveness of P. brassicae-induced cabbage volatiles to C. glomerata (Fig. 2). In contrast to lima bean plants, the emission of terpenoids in cabbage leaves was not significantly inhibited by fosmidomycin. Thus, a role of terpenoids in the direct or indirect defence of Brussels sprouts against Pieris butterflies cannot be excluded by this study. Future studies should elucidate the regulation of constitutive and induced terpenoid production, for example, by applying combinations of different inhibitors, for example, fosmidomycin and cerivastatin.


In conclusion, this study demonstrates that inhibitors like fosmidomycin can be used to investigate the role of terpenoid infochemicals in plant defence mechanisms against herbivores. In comparison with compounds inducing plant defence responses, such as coronalon or jasmonates (Schüler, Mithöfer & Baldwin 2004; Wasternack et al. 2006), inhibitors have been applied far less in studies addressing indirect plant defence. Although many of the inhibitors are specific for a certain biosynthetic pathway, they may not specifically inhibit particular chemical compounds as biochemical pathways generally have more than one final product. In the case of fosmidomycin, one should be aware that not only volatile terpenoids are inhibited but also other ‘essential’ terpenoids, which are important membrane components, photosynthetic pigments, or antioxidants, such carotenoids, sterols and gibberellins, might be affected as well (Owen & Peñuelas 2005). Therefore, experiments studying the significance of volatile terpenoids should use rather short incubation times.

Future studies need to elucidate what the role of particular compounds is in attracting carnivorous arthropods. One approach would be to combine ecological and molecular tools by using plants that have been genetically modified in the emission of certain terpenoids. This new field of ecogenomics has been successfully developed in the last few years and provides promising opportunities on the way to understand how indirect plant defence mechanisms function (Dicke, van Loon & de Jong 2004; Dicke 2006; Ouborg & Vriezen 2007; Snoeren, De Jong & Dicke 2007).


We would like to thank Rieta Gols and Gabriella Bukovinszkine'Kiss for the help with the experiments. Many thanks to Leo Koopman, Frans van Aggelen and André Gidding for culturing the insects and mites, and the experimental farm of Wageningen University (Unifarm) for rearing the Brussels sprout plants. We also thank two anonymous reviewers for their highly valuable comments. The study was financially supported by the European Commission contract MC-RTN-CT-2003-504720 ‘ISONET’ and by a VICI grant (nr 865.03.002) from the Earth and Life Sciences Foundation, which is subsidized by the Netherlands Organization for Scientific Research.