A specialist root herbivore reduces plant resistance and uses an induced plant volatile to aggregate in a density-dependent manner


  • Christelle A. M. Robert,

    1. Laboratory for Fundamental and Applied Research in Chemical Ecology (FARCE), University of Neuchâtel, 2000, Neuchâtel, Switzerland
    2. Root-Herbivore Interactions Group, Max Planck Institute for Chemical Ecology, Beutenberg Campus, 07745, Jena, Germany
    Search for more papers by this author
  • Matthias Erb,

    1. Root-Herbivore Interactions Group, Max Planck Institute for Chemical Ecology, Beutenberg Campus, 07745, Jena, Germany
    Search for more papers by this author
  • Bruce E. Hibbard,

    1. United States Department of Agriculture, Agricultural Research Service, Plant Genetics Research Unit, University of Missouri, Columbia, MO, USA
    Search for more papers by this author
  • B. Wade French,

    1. United States Department of Agriculture, Agricultural Research Service, North Central Agricultural Research Laboratory, Brookings, SD, USA
    Search for more papers by this author
  • Claudia Zwahlen,

    1. Laboratory for Fundamental and Applied Research in Chemical Ecology (FARCE), University of Neuchâtel, 2000, Neuchâtel, Switzerland
    Search for more papers by this author
  • Ted C. J. Turlings

    Corresponding author
    1. Laboratory for Fundamental and Applied Research in Chemical Ecology (FARCE), University of Neuchâtel, 2000, Neuchâtel, Switzerland
    Search for more papers by this author


1.Leaf-herbivore attack often triggers induced resistance in plants. However, certain specialist herbivores can also take advantage of the induced metabolic changes. In some cases, they even manipulate plant resistance, leading to a phenomenon called induced susceptibility. Compared to above-ground plant-insect interactions, little is known about the prevalence and consequences of induced responses below-ground.

2.A recent study suggested that feeding by the specialist root herbivore Diabrotica virgifera virgifera makes maize roots more susceptible to conspecifics. To better understand this phenomenon, we conducted a series of experiments to study the behavioural responses and elucidate the underlying biochemical mechanisms.

3.We found that D. virgifera benefitted from feeding on a root system in groups of intermediate size (3–9 larvae/plant in the laboratory), whereas its performance was reduced in large groups (12 larvae/plant). Interestingly, the herbivore was able to select host plants with a suitable density of conspecifics by using the induced plant volatile (E)-β-caryophyllene in a dose-dependent manner. Using a split root experiment, we show that the plant-induced susceptibility is systemic and, therefore, plant mediated. Chemical analyses on plant resource reallocation and defences upon herbivory showed that the systemic induced-susceptibility is likely to stem from a combination of (i) increased free amino acid concentrations and (ii) relaxation of defence inducibility.

4.These findings show that herbivores can use induced plant volatiles in a density-dependent manner to aggregate on a host plant and change its metabolism to their own benefit. Our study furthermore helps to explain the remarkable ecological success of D. virgifera in maize fields around the world.


To withstand herbivory, plants reconfigure their metabolism (Karban & Baldwin 1997; Walling 2000; Schwachtje & Baldwin 2008). This reconfiguration includes the production of toxic secondary metabolites (Steppuhn et al. 2004; Glauser et al. 2011), as well as reallocation of primary compounds (Babst et al. 2005; Orians, Thorn & Gomez 2011). In many cases, the induced changes increase the plant's resistance against the attacking herbivore (Steppuhn et al. 2004; Erb et al. 2009; Glauser et al. 2011). However, in some cases, herbivore attack can also reduce plant resistance. Most of the time, such plant susceptibility is induced by specialist herbivores, which have, over evolutionary time, adapted to specific host plants. Several mechanisms have been proposed to contribute to induced susceptibility (Karban & Agrawal 2002). First, herbivores may be able to suppress plant defences, by physically severing defensive structures (Berryman et al. 1989; Raffa 2001; Wallin & Raffa 2001; Kane & Kolb 2010) or interference with defensive signalling (Musser et al. 2002; Bede et al. 2006; Sarmento et al. 2011). Second, herbivores can evolve resistance to induced secondary metabolites (Ehrlich & Raven 1964; Rausher 1996; Berenbaum & Zangerl 1998; Stout & Bostock 1999; Glauser et al. 2011; Robert et al. 2012b) and even use them to their own advantage (Hopkins, Ekbom & Henkow 1998; Agrawal & Sherriffs 2001; Smallegange et al. 2007; Howe & Jander 2008; Robert et al. 2012b). Third, herbivores can induce reallocation of primary metabolites to their feeding site (Way & Cammell 1970; Larson & Whitham 1991; Giron et al. 2007; Kaiser et al. 2010; Compson et al. 2011).

One important aspect of induced susceptibility is density dependence. Most herbivores can only benefit from induced changes in their host plant as long as resources are sufficiently abundant (Katano et al. 2007). High densities of herbivores invariably lead to intraspecific competition and resource overexploitation, which reduces herbivore fitness (Ellner et al. 2001). Yet, it remains unclear to what extent specialist herbivores can select host plants on the basis of an optimal density of attacking conspecifics. Herbivore-induced plant volatiles that are emitted in a density-dependent manner (Shiojiri et al. 2010) can provide information about the status of the host plant and may be used by herbivore to optimize host selection in this context.

While it is generally accepted that induced resistance is more common than induced susceptibility in above-ground plant–insect interactions (Karban & Agrawal 2002), even for specialist herbivores (Agrawal & Kurashige 2003), little is known about induced changes in resistance below-ground, despite the fact that root herbivores are common in many ecosystems and are among the most important agricultural pests (Hunter 2001). In a recent study, feeding by larvae of the vine weevil, Otiorhynchus sulcatus, on raspberry plants caused a 19% reduction in growth of subsequently attacking weevil larvae (Clark, Hartley & Johnson 2011). On the other hand, slightly damaged onion bulbs support higher survival of Delia antiqua larvae (Hausmann & Miller 1989), and D. radicum larvae grow better on previously attacked turnip plants (Pierre et al. 2011). It has been speculated that induced resistance below-ground may be less frequent than above-ground, as alternative trading-off strategies like induced tolerance, such as induced resource sequestration and root regrowth after herbivory, may provide a greater benefit to the plant (Erb et al. 2012).

We previously found that the specialist root herbivore Diabrotica virgifera virgifera LeConte (Coleoptera: Chrysomelidae) induces susceptibility in its host plant, Zea mays L. (Poaceae), and that the herbivore uses induced volatiles to find infested host plants (Robert et al. 2012a). Diabrotica virgifera larvae remain highly mobile during their development, and several studies show that they change host plants and redistribute at later instars (Strnad & Bergman 1987; Hibbard et al. 2003, 2004), possibly upon overexploitation of their initial food source (Hibbard et al. 2004). In the current study, we tested whether D. virgifera can use the induced volatiles to select plants with an optimal density of attacking conspecifics. We further investigated whether the increase in growth on infested plants is attributable to a stimulating effect of the induced volatiles, but also if possible changes in primary metabolism may be responsible for the increase in performance. For instance, D. virgifera has been shown to induce water-stress in maize (Godfrey, Meinke & Wright 1993; Dunn & Frommelt 1998; Erb et al. 2011b), a condition that can lead to an increase in shoot–root assimilate flow (Farooq et al. 2009). To test this, we quantified the levels of free amino acids and sugars as well as the expression of marker genes associated with carbon transport and partitioning. Finally, we investigated whether attack by high densities of D. virgifera reduces the capacity of maize plants to mobilize defences in response to subsequent herbivory by quantifying the expression of maize defence marker genes.

Materials and methods

Plants and Insects

Maize plants (Zea mays, variety Delprim) were sown in plastic pots (11 cm high, 4 cm diameter) by placing them on moist washed sand (0–4 mm; Jumbo, Marin-Epagnier, Switzerland) and covering them with 2 cm of commercial potting soil (Aussaaterde, Ricoter, Aarberg, Switzerland). Seedlings were grown in a climate chamber (23 ± 2 °C, 60% relative humidity, 16:8 h L/D and 350 μmol m−2 s−1), and MioPlant Vegetable and Herbal Fertilizer (Migros, Neuchâtel, Switzerland) was added every 2 days after plant emergence. Twelve-day-old plants with two fully developed leaves were used for the experiments. Diabrotica virgifera eggs were obtained from the USDA-ARS-NCARL (Brookings, SD, USA) and kept on freshly germinated maize until use. Second, instar larvae were used in all laboratory experiments.

Density-Dependent Performance of D. virgifera

To assess the effect of egg density on D. virgifera under natural conditions, a field study was conducted at the University of Missouri Bradford Research and Extension Center, 9 km east of Columbia, MO, USA in 2005. Field design and soil conditions were described elsewhere (Hibbard et al. 2010). For this field experiment, the line DKC 60–17 (RR) was used. Briefly, plots of 64 maize plants were planted using at 76·2 cm row spacing and 17 cm seed spacing. Densities of 25, 50, 100, 300, 600, 1200 and 2400 viable D. virgifera eggs per 30·5 cm of maize row were applied into the soil and left to develop. Plots were covered with a screen tent (3·05 × 3·66 m; Coleman, Rye, NY, USA) prior adult emergence. Emerging beetles were collected two to three times per week using either mouth aspirators (BioQuip, Rancho Dominguez, CA, USA) or battery-operated aspirators (BioQuip). Adults were immediately transferred into 95% ethanol, and head capsule widths of the collected beetles were measured. Each egg density was replicated four to six times. To evaluate the effect of larval density on D. virgifera in the laboratory, maize seedlings were infested with one, three, six, nine or twelve larvae (n = 7), pre-weighed using a XP2U micro-scale (Maximum capacity 6·1 g, readability: 0·1 μg; Mettler-Toledo International Inc., LLC, Columbus, OH, USA). After 2 days, all larvae were collected by hand sorting and weighted again to determine their performance.

Density-Dependent Attraction of D. virgifera

To evaluate the attraction of D. virgifera to infested plants at different densities, healthy plants and plants infested with one, three, six, nine or twelve larvae were potted in two-arm below-ground olfactometers as described elsewhere (Robert et al. 2012a). All pots were filled using moist white sand (10% water; Migros) and covered with aluminium foil to avoid light stress and soil desiccation. After 2 days, the two pots were connected via an empty glass tube with a vertically connected access in the middle and one Teflon connector at both sides of the glass tube as previously described. The Teflon connectors contained a fine metal screen (2300 mesh; Small Parts Inc., Miami Lakes, FL, USA) that allowed volatile dispersion but avoided any visual or liquid cues for the herbivore. Furthermore, Teflon connectors prevented the larvae from reaching the roots. A group of six larvae was released in the central glass connector of the olfactometer, and the first choice of the insects towards one or the other plant was recorded. In total, 346 larvae were used for the choice experiments. Larvae that did not choose after 15 min were noted as;no choice'. Leaf wilting was recorded for all plants and scored from zero (no symptoms) to four (complete loss of turgidity) as previously described (Erb et al. 2011a). The experiment was repeated twice (n I1, I3, I6, I12 = 17, n I9 = 8). At the end of the first assay, CO2 emissions were evaluated as described below and root systems were collected and gently washed with tap water to determine their fresh biomass. In the second repetition, root systems were collected at the end of the choice experiment, washed with tap water and immediately frozen in liquid nitrogen to determine induced volatiles as described below.

Quantification of Volatiles

To investigate the potential signals that D. virgifera uses to assess the infestation level of a plant, volatile emissions of seedlings that were used in the above two-arm olfactometers assays were determined. CO2 emission was evaluated from the plants of the first choice experiment (see above) by connecting the below-ground glass pots to an additional glass vessel (28 cm long, 5 cm diameter) via the connector and a glass joint (n = 8). The glass vessel was closed using parafilm and left to stabilize for 1 h. A CO2 gas meter (Voltcraft, CM-100; Conrad Electronics, Dietlikon, Switzerland) was then introduced into the connected vessel for 3 min, and CO2 levels were recorded. Induced volatiles were determined from the second repetition of the choice experiment (see above) using SPME GC-MS following a previously described protocol (Erb et al. 2011a) (n = 8). The obtained peaks were analysed and identified by comparing volatile retention times and mass spectra with those of the NIST05 Mass Spectral Library (Agilent Technologies Life Sciences and Chemical Analysis Group, Santa Clara, CA, USA) and those of pure compounds.

Dose-Dependent Responses of D. virgifera to (E)-β-caryophyllene

As (E)-β-caryophyllene can be detected in vivo in the headspace of D. virgifera-induced roots (Hiltpold et al. 2011), has superior diffusion properties in the soil (Hiltpold & Turlings 2008) and was previously reported to be attractive for D. virgifera larvae (Robert et al. 2012a), we focused on this compound and investigated whether D. virgifera can use it in a dose-dependent manner to detect infested plants. Two-arm olfactometers were used as described above. Healthy plants were potted in both arms of the system. Synthetic (E)-β-caryophyllene (Sigma Aldrich Chemie GmbH, Buchs SG, Switzerland) was continuously released into the rhizosphere of one of the healthy plants using slow-release capillary dispensers as described by von Mérey et al.(2011). Capillaries of 0·5, 1, 2, 3, 6 and 25 μL were used (n 0·5 and 2 μL = 11, n 1, 3, 6 and 25 μL = 10). One microlitre capillary dispensers continuously release up to 40 ng h−1 (E)-β-caryophyllene, which corresponds to the emission of a D. virgifera infested root system (Robert et al. 2012a). Thus, the amount of (E)-β-caryophyllene released by the 0·5-, 1- and 2-μL dispensers are within the physiological range of infested maize seedlings, while the dose of (E)-β-caryophyllene released by bigger capillaries would be aberrant in nature. All pots were wrapped in aluminium foil to avoid light and desiccation of the system. After 48 h, the two pots were connected using an empty glass tube and Teflon connectors. After 30 min, six D. virgifera larvae were inserted into the below-ground olfactometers central part and their choice pattern was recorded.

D. virgifera Performance on Systemic Roots of Infested Plants

To investigate whether the increased growth of D. virgifera on previously infested plants is attributable to plant-mediated effects or direct, physical interaction with conspecifics, a split root system was designed. Maize root systems were gently washed with tap water and potted in two separate glass vials (15 cm long, 2 cm diameter) filled with moist white sand (10% water). Five D. virgifera larvae were added to one of the glass tubes (local roots). Control plants were left uninfested. After 4 days, a second batch of five D. virgifera larvae was weighted as described above and placed on the second (systemic) side of the root system of all plants. Six hours later, larvae were recovered and re-weighed.

Volatile-Mediated Feeding Stimulation

To determine whether the increased growth of D. virgifera is attributable to volatile-mediated feeding stimulation, we measured the growth of the herbivore upon exposure to induced volatiles. For this, two glass pots (5 cm diameter, 11 cm deep) were connected as described above. The first pot contained the odour source: a healthy plant, a plant infested with five D. virgifera larvae or five larvae feeding on artificial diet (Pleau et al. 2002). After 24 h, five D. virgifera larvae were weighted and allowed to feed on artificial diet in the second pot while exposed to the different odour sources at the same time (n diet = 6, n healthy plants = 9, n infested plants = 12). One day later, the larvae were collected and re-weighed.

Induced Changes in Primary Metabolism

To evaluate whether changes in plant primary metabolism benefits D. virgifera, the local and systemic response of maize roots following infestation was investigated using the split-root system as described above. Five D. virgifera larvae were added to one of the glass tubes (local roots). Control plants were left uninfested (n = 7). The local and systemic parts of the root system were collected separately 4 days after infestation, washed with tap water and immediately frozen in liquid nitrogen and stored at −80 °C. To ensure enough material for analyses, roots from three plants were pooled together. Roots were ground into a fine powder under liquid nitrogen. Free amino acids were determined as previously described (Knill et al. 2008). Sucrose and hexose contents were determined enzymatically using a Sucrose, d-fructose and d-glucose assay kit (Megazyme International Ireland Limited, Bray Business Park, Bray, Co. Wicklow, Ireland) following the manufacturer's instructions (Beutler 1988; Kunst, Draeger & Ziegernhorn 1988; Outlaw & Mitchell 1988). The concentrations of d-glucose, d-fructose and sucrose were calculated using the megazyme Mega-Calc™ software (Megazyme International Ireland Limited, Bray Business Park, Bray Co. Wicklow, Ireland). The expression of marker genes involved in carbohydrate transport and metabolism was assessed using previously established methods and primers listed in Table S1 (Supporting information; Erb et al. 2010).

Induced Changes in Plant Defences

To test whether D. virgifera suppresses plant defences in maize roots, we analysed the expression of marker genes involved in plant direct defences, hormonal signalling and volatile production using previously established methods and primers (Erb et al. 2009). For this experiment, the same cDNA as for the experiment above was used.

Root Defence Response to Subsequent Attack

To test whether the root infestation impacts the ability of healthy roots from the same plant to respond to subsequent attack, maize root systems were split by washing and transplanting them in two glass tubes as described above. Five D. virgifera larvae were added in one of the tubes. Control plants remained uninfested on both sides. As D. virgifera feeding would be influenced by the previous infestation treatment (see above), we used jasmonic acid (JA) as a second inducer. JA is produced in the roots after D. virgifera attack (Erb et al. 2009) and induces root volatiles in a similar manner as the herbivore (Erb et al. 2011a), which is why we used this hormone as a herbivore-mimic for this particular experiment. Four days after infestation, 10 mL aqueous solution of 100 μm jasmonic acid or 10 mL of water only were added to the systemic side of the root system (n = 8). Twelve hours later, roots of the two treatments (C→JA, D. virgifera →JA) were collected, washed with tap water, frozen in liquid nitrogen and ground into a fine powder as described above. The expression of marker genes that are directly involved in the production of secondary metabolites and defensive proteins and, consequently, are expected to contribute to induced resistance, was determined.

Statistical Analyses

Analyses were performed on the software package R, version 2·8·1 and sas Statistical Package (2004; SAS Institute Inc., Cary, NC, USA). All data were first analysed with a Levene's and a Kolmogorov–Smirnov test to determine heteroscedasticity of error variance and normality. The effect of D. virgifera density on head capsule width in the field and performance in laboratory were analysed using the proc Mixed model of the sas Statistical Package. Diabrotica virgifera choice was evaluated using a log linear model (glm) using R. As the data did not fit to simple variance assumptions implied in using a binomial distribution, quasi-likelihood functions were used to compensate for the over-dispersion of the larvae in the system. The two repetitions of the experiment were included as a co-factor in the analysis. As the repetition of the experiment had no effect on the model, the factor ‘experiment’ was removed from the analysis. Root biomasses and volatile production of healthy and infested plants were compared using one-way anovas followed by post hoc Tukey's HSD tests. If the data did not pass the two tests, Kruskal-Wallis one-way anovas on ranks were performed, followed by pairwise Dunn's tests. The effect of the emitted volatiles on the larval choice was evaluated by performing multiple one-way ancovas with one volatile as a co-factor in each analysis, after testing the assumptions of heteroscedasticity of error variance and normality, significance of the linear regressions (P < 0.10) and the independency of the variables. Diabrotica virgifera performance on healthy or infested plants in the split root design was compared using Student's t-tests. The effect of volatiles on D. virgifera growth on artificial diet was analysed using a Kruskal-Wallis on ranks test (H-test), and the comparison between the growth of larvae exposed to plant odours with the growth of larvae exposed to the volatile bouquet of conspecifics feeding on artificial diet was compared using a t-contrast on ranks test. The effect of D. virgifera feeding on amino acids and carbohydrate contents as well as marker gene expression in maize roots were investigated using t-tests when the data fulfilled the heteroscedasticity of error variance and normality conditions, otherwise, Mann–Whitney rank sum tests (U-tests) were conducted.


Induced Susceptibility by D. virgifera is Density-Dependent

In the field, D. virgifera head capsule width, which can be used as a fitness indicator (French & Hammack 2010), was found to be density-dependent: D. virgifera adults tended to have smaller head capsules when feeding together in small densities than in medium densities. Furthermore, adults that fed in medium densities had larger head capsules than those feeding in large densities (n 25 = 6, n 50, 100 = 5, n 300, 600, 1200, 2400 = 4; proc mixed, P = 0·018; Fig. 1a). Diabrotica virgifera larval performance was also found to be density-dependent in the laboratory, where larvae grew better when feeding in groups of three, six or nine larvae than alone (n = 7; proc mixed, P = 0·197; Fig. 1b).

Figure 1.

Diabrotica virgifera performance is density-dependent. (a) D. virgifera adult head capsule width after developing on plants infested with different egg densities. (b) D. virgifera larvae performance when feeding on plants with different larval densities in laboratory. Mean ± SE are presented. Different letters indicate significant differences (P < 0·05) within each larval density using the differences of least squares means.

D. virgifera is Attracted to Plants with a Medium Density of Conspecifics

Diabrotica virgifera larvae preferentially oriented towards healthy plants rather than plants that were infested with low (1 larvae) or high (12 larvae) density of conspecifics, but they were significantly attracted to plants that were infested with a medium density of 6 larvae [n I1, I3, I6, I12 = 17, n I9 = 8; glm, Control (C) vs. I1, I6 and I12: P < 0·05; Fig. 2]. High density D. virgifera infestation led to clear wilting symptoms in the leaves (see Fig. S1a, Supporting information). When we compared the choice of the larvae according to the severity of wilting symptoms induced by conspecifics (using leaf-wilting as a grouping factor rather than infestation number), we found that the extent of water-stress did not influence D. virgifera preference, with the exception of plants that had completely lost their turgidity, which were avoided by the larvae (see Fig. S1b, Supporting information).

Figure 2.

Diabrotica virgifera selectively orients towards suitable host plants. Number of larvae (mean ± SE) that oriented towards a healthy plant or a plant infested with different densities of D. virgifera larvae. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).

(E)-β-Caryophyllene and α-Humulene, but not CO2 Emissions Correlate with Larval Choice Patterns

At high infestation density (9 and 12 larvae per plant), root biomass was decreased significantly, while low to medium densities had no measurable impact (see Fig. S2, Supporting information).

High infestation also increased the CO2 production per gram of fresh roots (see Fig. S3a, Supporting information). However, the total amount of emitted CO2 was not influenced by infestation density (see Fig. S3b, Supporting information). Diabrotica virgifera attack induced the production of (E)-β-caryophyllene, α-humulene, α-copaene, tetradecane, heptadecane, 4-methyl nonane, 4-methyl heptane and tetradecene (n = 8; Kruskal-Wallis one-way analysis of variance on ranks: P < 0·05). Among those compounds, only (E)-β-caryophyllene, α-humulene and α-copaene are actually released into the rhizosphere (data not shown). When included as co-variates into the choice pattern model, only (E)-β-caryophyllene and α-humulene improved the model fit of the ancova compared to a simple anova (one-way anova: P = 0·021; one-way ancovas, (E)-β-caryophyllene: P = 0·008; α-humulene: P = 0·03; α-copaene: P = 0·603; Fig. 3), indicating that they are most likely to explain the density-dependent D. virgifera attraction. Only the model for (E)-β-caryophyllene and α-humulene resulted in linear regressions, indicating that the P-value for α-copaene should be interpreted cautiously.

Figure 3.

Diabrotica virgifera induces plant volatiles. SPME GC-MS peak areas (xE06) (mean ± SE) of plant-induced volatiles upon D. virgifera attack. Different letters indicate significant differences within each larval density using a post hoc Dunn's test.

The Attraction of D. virgifera to (E)-β-Caryophyllene is Dose-Dependent

Diabrotica virgifera was slightly attracted to 0·5 μL (E)-β-caryophyllene dispensers (n = 11; glm: P = 0·064; Fig. 4), and strongly preferred 1 μL (E)-β-caryophyllene dispensers over controls (n = 10; glm: P < 0·001; Fig. 4). The larvae were not attracted to dispensers with a bigger capillary volume (n 2 μL = 11, n3, 6 and 25 μL = 10; glm: P > 0·05; Fig. 4).

Figure 4.

Diabrotica virgifera attraction to (E)-β-caryophyllene is dose-dependent. Number of larvae (mean ± SE) that oriented towards a healthy plant or a healthy plant whose rhizosphere was complemented with different amounts of (E)-β-caryophyllene using slow-release capillary dispensers. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).

D. virgifera Induces Systemic Susceptibility in Maize Roots

Larvae feeding on the systemic side of infested roots grew more than five times better than larvae feeding on systemic roots of healthy plants (n = 7; t-test: P = 0·008; Fig. 5). The effect was still present when the sand moisture level was increased to 20% to avoid the induction of water-stress (see Fig. S4, Supporting information). Although the general exposure to plant volatiles stimulated D. virgifera larvae to feed (n diet = 6, n healthy = 9, n infested = 12; Kruskal-Wallis on ranks: P = 0·039; n diet = 6, n plants = 21; t contrast on ranks: P = 0·050; Fig. 6), no difference was found between the performance of larvae exposed to the volatile bouquet of healthy or infested plants.

Figure 5.

Diabrotica virgifera larvae benefit from plant-mediated interactions with spatially separated conspecifics. Diabrotica virgifera growth (mean ± SE) over a 6-h feeding period on healthy plants or on the systemic undamaged roots of a plant that had been infested for 4 days with conspecifics. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).

Figure 6.

Exposure to Diabrotica virgifera-induced plant volatile does not stimulate larval performance. Diabrotica virgifera larvae relative weight gain (mean ± SE) when feeding for 24 h on artificial diet (Pleau et al. 2002) and exposed to volatiles from conspecifics feeding on diet, healthy plant or D. virgifera infested plant volatiles. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001) within each larval density using a post hoc Dunn's test.

Root Resource Allocation is Altered Both Locally and Systemically Upon Root Herbivory

Diabrotica virgifera attack led to an increase in free amino acid concentrations, both locally and systemically. Locally, asparagine, aspartic acid, glutamine, histidine, phenylalanine and tryptophane significantly increased (n = 6, t-tests or Mann–Whitney rank sum tests: P < 0·05; Fig. 7a), and leucine, serine and tyrosine showed similar trends (n = 6; t-tests: 0·05 > P > 0·10; Fig. 7a). Systemically, the concentrations of histidine, phenylalanine, tryptophan and tyrosine increased upon infestation (n = 6; t-tests: P < 0·05; Fig. 7b). Asparagine, aspartic acid and glutamic acid concentrations also showed similar trends upon infestation (n = 6; t-tests: 0·05 > P > 0·10; Fig. 7b). Carbohydrate partitioning in D. virgifera attacked roots was affected only in the local roots: glucose contents were reduced by 50% and more sucrose accumulated (n = 6; t-tests or Mann–Whitney rank sum tests, sucrose and glucose: P < 0·05; fructose: P > 0·05; Fig. 7c). On the other hand, the partitioning of sucrose and hexoses was not affected in the undamaged systemic roots of infested plant (n = 6; t-tests or Mann–Whitney rank sum tests: P > 0·05; Fig. 7d). In accordance with the sugar measurements, both vacuolar (ivr) and cell wall (incw) invertase genes were downregulated locally upon herbivory (n = 7, t-tests: P < 0·05) with the exception of ivr1, whose expression was enhanced (n = 7, t-tests, P < 0·05; Fig. 8a).

Figure 7.

Root herbivory leads to reconfiguration of the primary metabolism. (a) Local amino acid contents in healthy or infested roots. (b) Systemic amino acid contents in roots of healthy or infested plants. (c) Local carbohydrate contents in healthy or infested roots. (d) Systemic carbohydrate contents in roots of healthy or infested plants. Mean ± SE are presented (ng g−1 of fresh weight). Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).

Figure 8.

Local carbohydrate metabolism changes upon root infestation. (a) Local ln fold changes in expression of vacuolar (ivr) and cell wall (incw) invertases. (b) Local ln fold changes in expression of carbohydrate transporters. (c) Systemic ln fold changes in expression of vacuolar (ivr) and cell wall (incw) invertases. (d) Systemic ln fold changes in expression of carbohydrate transporters. Mean ± SE are presented. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).

The expression of carbohydrate transporter homologues was induced locally upon infestation (n = 7; t-tests: P < 0·05), except for c4, which showed lower expression, and zifl2, which showed no significant change (n = 7; t-tests: P < 0·05 and 0·05 < P < 0·10 respectively; Fig. 8b). No change in invertases and carbohydrate transporters was detected in the systemic roots of D. virgifera infested plants (n = 7; t-tests: P > 0·05; Fig. 8c,d).

D. virgifera does not Suppress Defence Markers

Roots responded strongly to the infestation by D. virgifera larvae. Marker genes of volatiles [(E)-β-caryophyllene (tps23) and indole (igl), lipoxygenases (lox3, lox5 and lox8)], direct defences (proteinase inhibitors: cysII, serpin, mpi and benzoxazinones: bx1) and pathogenesis-related proteins (pr1 and pr5) were upregulated upon herbivory (n = 7; t-tests and Mann–Whitney rank tests: P < 0·05; Fig. 9a). On the other hand, the expression of marker genes of hormones like ethylene (acs6), auxin (saur2), abscisic acid (nced) and jasmonic acid (opr7) remained unaffected upon infestation (n = 7; t-tests: P > 0·05; Fig. 9a). Systemically, only a few of the markers responded: igl, acs6 and lox8 were upregulated (n = 7; t-tests and Mann–Whitney rank tests: P < 0·05), while all the other tested marker genes expression remained unchanged (n = 7; t-tests and Mann–Whitney rank tests: P > 0·05; Fig. 9b).

Figure 9.

Plant defences are strongly induced locally. (a) Local ln fold changes in expression of volatile, hormone signalling and direct defence marker genes. (b) Systemic ln fold changes in expression of volatile, hormone signalling and direct defence marker genes. Mean ± SE are presented. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).

Infestation Attenuates the Plant's Responsiveness to Future Attacks

Following the infestation of one side of the root system by D. virgifera larvae, the undamaged systemic root side responded less to jasmonic acid application than the systemic roots of uninfested plants (n = 8; t-tests: cysII; cyst and bx1: P < 0·05; Fig. 10), indicating a relaxation of defensive inducibility.

Figure 10.

A first infestation by Diabrotica virgifera attenuates subsequent defence responses. Ln-transformed fold-induction values (mean ± SE) in the expression of marker genes involved in plant direct defence in systemic roots of infested plants relative to healthy plants. Systemic roots of infested and healthy plants were induced with 100 μm of jasmonic acid for 12 h. Stars indicate significant differences (*P ≤ 0·05, **P ≤ 0·01; ***P ≤ 0·001).


This study demonstrates that plant-mediated facilitation occurs between D. virgifera conspecifics, a phenomenon that is likely the result of a combination of volatile-mediated host location, plant resource reallocation and weakened plant defences. Diabrotica virgifera larvae were found to benefit from feeding in groups, as they performed better on plants infested with other larvae in laboratory and field conditions. Yet the observed feeding facilitation was reversed when D. virgifera fed in large groups: in our assays, both larval performance and head capsule width of emerging adults decreased at high densities. Negative density-dependent effects are likely due to competition for limiting plant-resources. For instance, in our assays, high densities of larvae considerably decreased the root biomass available for conspecifics, and the strong wilting of the leaves may have reduced overall plant quality. Similar effects were reported in previous field studies, where adult D. virgifera emergence decreases at high egg density (Onstad et al. 2006; Hibbard et al. 2010). Interestingly, our laboratory assays show that the specialist D. virgifera was able to select host plant with a suitable density of root herbivores: When given a choice between a healthy plant and a plant infested with different numbers of conspecifics, D. virgifera larvae preferentially oriented towards plants infested with an intermediate number (six larvae), but were not attracted to plants infested with low or high densities. This behaviour can be beneficial for the root herbivore, because it enables it to locate the best host plants. Together with our previous study showing that D. virgifera can distinguish host plants of different quality (Robert et al. 2012a), these results demonstrate the remarkable ability of this root herbivore for host selection.

We propose here that D. virgifera uses plant volatiles to find plants with a suitable infestation density. Although CO2 is known to be highly attractive to the herbivore (Bernklau & Bjostad 1998), our results suggest that other plant volatiles also play a key role in host selection by D. virgifera, as in our experiments CO2 emissions remained constant upon infestation by different densities, but several induced volatiles were produced in a density-dependent manner. (E)-β-caryophyllene, α-humulene, α-copaene, tetradecane and tetradecene in particular showed a parabolic pattern, with peak emission occurring at medium densities of infestation. It should be taken into account that as root biomass decreases upon infestation in higher densities, our analyses likely over-estimated the total amounts emitted from plants infested by nine or twelve larvae. In vivo analyses of root volatile emission of D. virgifera infested plants show that among the compounds detected in the present study, only (E)-β-caryophyllene, α-humulene and α-copaene are actually released into the rhizosphere (data not shown), which nicely matches the results from a diffusion study by Hiltpold & Turlings (2008). Analyses of covariance (ancova) using these three compounds as covariates showed that (E)-β-caryophyllene and α-humulene, but not α-copaene, can improve the fit of the model of larval choice, suggesting that these two compounds may be used by D. virgifera to distinguish plants infested by different densities of conspecifics. (E)-β-caryophyllene and α-humulene are both products of a single terpene synthase, tps23, and well known to be induced upon infestation by the root herbivore (Kollner et al. 2008). As (E)-β-caryophyllene is emitted in much higher amounts than α-humulene (Erb et al. 2011a) and was shown to be an attractant for D. virgifera larvae (Robert et al. 2012a), we focused on that compound to investigate its dose-dependent effect on the root herbivore. (E)-β-caryophyllene was attractive to D. virgifera only when released at a rate of 40 ng h−1 (1 μL capillary dispensers), which corresponds to the release rate of plants infested with six conspecifics (Robert et al. 2012a). Although the attractiveness of other chemical remains to be tested, it is highly probable that D. virgifera uses (E)-β-caryophyllene in a dose-dependent manner to locate good host plants.

The feeding facilitation of D. virgifera when feeding in medium-sized group may be attributed to either plant-mediated effects or the direct influence of conspecifics. The split root experiment shows that spatially separated larvae grew, over a period of only 6 h, five times bigger on previously infested plants compared to healthy plants, showing that plant-mediated effects are sufficient to explain the positive density dependence observed in the field. As upon below-ground attack, (E)-β-caryophyllene is produced both locally and systemically (Hiltpold et al. 2011), we first tested the hypothesis that this sesquiterpene may directly stimulate feeding. Many lepidopteran leaf-herbivores for example are stimulated by green leaf volatiles released from fresh wounds (Meldau, Wu & Baldwin 2009) or by volatile breakdown products of induced secondary metabolites like glucosinolates (Agrawal & Sherriffs 2001; Nielsen et al. 2001). We found that larvae exposed to D. virgifera-induced volatiles grew similarly than larvae exposed to the volatile bouquet of healthy plants, which shows that induced plant volatiles do not stimulate larvae to feed. Interestingly, exposure to plant volatiles in general increased larval weight gain, suggesting that constitutive volatile compounds do have a stimulatory effect on D. virgifera.

Diabrotica virgifera attack led to changes in the primary metabolism of maize roots. Larval feeding induced the accumulation of free amino acids both locally and systemically. Free amino acids can be involved in (i) osmotic adjustment (Navari-Izzo, Quartacci & Izzo 1990; Marur, Sodek & Magalhes 1994), for example in response to the water-stress imposed by the root herbivore (Dunn & Frommelt 1998; Erb et al. 2011b), (ii) defence (D'Auria & Gershenzon 2005; Tzin & Galili 2010; Vogt 2010) or (iii) nitrogen transport away from the roots (Paine, Redak & Trumble 1993; Trumble, Kolodny-Hirsch & Ting 1993). At the same time, amino acids are known to be the growth-limiting factor of herbivorous insects (Behmer 2006) and their accumulation in the roots of infested plants may, therefore, explain the better performance of D. virgifera. Apart from nitrogen metabolism, D. virgifera also affected carbon distribution: Upon infestation, attacked roots accumulated more sucrose, but less glucose than roots of healthy plants. Three putative carbohydrate transporter genes were more strongly expressed in attacked roots, while invertases were generally downregulated. Ivr1 and Ivr2 showed divergent expression patterns, supporting previous work showing that the depletion of carbohydrates upregulates the expression of Ivr1 but downregulates the expression of Ivr2 (Xu et al. 1996). Invertases play a key role in regulating root sink strength (Weil & Rausch 1990; Miller & Chourey 1992; Kim et al. 2000; Roitsch et al. 2003; von Schweinichen & Buttner 2005). Upon water-stress, vacuolar invertases were reported to be strongly induced in maize roots, resulting in higher ratios between hexoses and sucrose that contributes to the osmotic adjustment (Kim et al. 2000). The general downregulation of the invertase marker genes observed in our assays suggests that D. virgifera attack may reduce the plants' capacity to react to the accompanying water-stress conditions, as has been observed before (Erb et al. 2009). However, as the changes in carbohydrate concentrations were limited to the local tissue, they are vunlikely to explain the differential performance of D. virgifera on systemic roots.

Diabrotica virgifera attack induced a pronounced local defence response, as indicated by the increased expression of marker genes involved in hormonal signalling, such as lipoxygenases (lox genes), direct defences, such as proteinase inhibitors (mpi, serpin, cysII), benzoxazinoids (bx1), pathogenesis-related proteins (pr1 and pr5), and indirect defences such as volatile production (igl and tps23). These results show that D. virgifera does not strongly, if at all, manipulate host defences (Zarate, Kempema & Walling 2007; Sarmento et al. 2011). Although the undamaged part of the root system was barely induced, with the exception igl, acs6 and lox8 whose expression was slightly upregulated, a second inductiovn of the undamaged part of infested root systems by jasmonic acid resulted in a lower induction of direct defence marker genes compared to plants that had not been previously infested. This indicates that D. virgifera larvae that attack an already infested plant may encounter a plant immune system that is less inducible, and consequently, less resistant. It remains to be determined whether this relaxation of inducibility can explain the higher performance of D. virgifera larvae on attacked plants. Testing this hypothesis would need a more detailed understanding of the mechanisms of induced resistance in maize roots.


Overall, our study shows that D. virgifera attack changes the root metabolism of maize plants, leading to systemically induced susceptibility. Diabrotica virgifera was found to use (E)-β-caryophyllene as a signal to locate plants with a suitable density of conspecifics. The presented experiments allow us to rule out volatile-mediated stimulation of feeding and direct effects of larval behaviour as explanations for the increase in larval growth. The hypotheses that either the higher amino acid levels or the relaxation of inducibility may be responsible for the enhanced D. virgifera performance remain to be tested. Understanding the mechanisms behind induced susceptibility is likely to improve our understanding of the extraordinary success of D. virgifera as a maize pest.


We thank Matt Higdon, Rebecca Higdon, Sarah Zukoff, Julie Barry and the whole student crew from the USDA-ARS Plant Genetics Research Unit (Columbia, Missouri) for their kind contribution to field experiments. This project was partially funded by the National Centre of Competence in Research (NCCR) ‘Plant Survival’ and supported by a Swiss National Science Foundation Fellowship to M.E. (PBNEP3-134930).