1. Plants face threats from a variety of herbivorous insects and can use induced responses to defend themselves against these attackers. Induced responses are mediated by signal transduction involving phytohormones, such as jasmonic acid (JA) and salicylic acid (SA). Cross-talk between signal transduction pathways triggered by attackers with contrasting feeding styles allows plants to fine-tune defences. A central question in this emerging field is to understand how responses to single attackers interfere with responses to other attackers, especially by integratively addressing molecular and ecological aspects.
2. We examined the plant-mediated interactions between the leaf-chewing Pieris brassicae and the phloem-sucking Brevicoryne brassicae, and their respective parasitoids Cotesia glomerata and Diaeretiella rapae, when feeding simultaneously, sequentially or in isolation on the brassicaceous ecological model plant Brassica oleracea. We analysed the underlying defence mechanisms in the plant. Levels of the phytohormones JA and SA transcriptional responses of a number of selected defence-related genes and secondary plant compounds were quantified at different time points during the single and multiple infestations.
3. The caterpillars developed faster and reached a larger body mass on plants previously attacked by aphids. Aphids initially developed faster on plants with caterpillars, although the moment of moulting to adults was independent of the presence of caterpillars. Both parasitoid species performed better under multiple-infestation scenarios than in single-herbivore situations.
4. On plants attacked by aphids, the JA levels were tenfold lower than on undamaged plants or plants with caterpillars. Additionally, the low transcript levels of LOX and MYC, genes coding for a JA biosynthesis-related enzyme and a transcription factor, respectively, in aphid-infested plants, suggest that the facilitation of the caterpillar performance was mediated by interference in signal transduction. Levels of carbon and nitrogen and secondary plant compounds (glucosinolates) did not differ significantly between treatments, suggesting that these compounds did not mediate the facilitation.
5. Our data show that the leaf chewer and phloem feeder asymmetrically interact not via competition as would be expected from interspecific herbivores but instead via facilitation; the phloem feeder attenuated JA-related plant defences, thus facilitating the growth and development of the leaf chewers. In linear bitrophic systems, interactions between JA and SA signalling pathways have been proposed to allow plants to fine-tune their defences, but if facilitation frequently occurs in interspecific interguild interactions among herbivores this may represent an important constraint for plant defences. Such a constraint might be reduced if, as in our model system, parasitoids also benefit from interactions between interguild hosts and nonhosts, but parasitoids are rarely considered in model molecular systems to assess the impact of herbivore-induced plant defences.
It is well understood how induced plant defences operate under the attack of the most common insect feeding guilds, in isolation, but it remains unexplored how responses to initial attackers impact on responses to subsequent attackers (Dicke, van Loon & Soler 2009; Pieterse et al. 2009; Rodriguez-Saona et al. 2010). Cross-talk between signalling pathways is suggested to allow plants to fine-tune their defences to optimize induced resistance to the attackers (Pieterse & Dicke 2007). However, cross-talk may also allow for positive interspecific interactions between insect herbivores from contrasting feeding guilds, representing a constraint for the plant when being attacked by leaf chewers and phloem feeders (Zarate, Kempema & Walling 2007; Dicke, van Loon & Soler 2009). This is an important gap in our knowledge, because plants are usually colonized by several attackers and the chances of being attacked by herbivores from different feeding guilds are often high. Remarkably, the consequences of multiple infestation on members of higher trophic levels, such as insect parasitoids and predators, have also received limited attention so far (Dicke, van Loon & Soler 2009). Excluding such carnivores results in an underestimation of the ecological implications of the interactions between plants and multiple attackers. The primary objective of this study was to investigate whether one of the most abundant leaf-chewing and a phloem-sucking insect species of brassicaceous plants benefits when feeding on the same host plant and whether facilitation is also reflected in the plant at the chemical and at the molecular level. For this, we compared the survival, growth and development of a commonly occurring leaf-chewing and phloem-sucking species under ‘single’ and ‘multiple’ plant infestation conditions. We also investigated the effects of multiple infestation on the performance of the main insect parasitoids of the two herbivores. Chemical responses in the plant under single and multiple attack were evaluated at several time points after the insect infestation by measuring the levels of the phytohormones JA and SA, the transcriptional responses of a number of JA- and SA-induced defence genes, a well-known group of plant defence compounds in brassicaceous plants (glucosinolates) and the levels of carbon and nitrogen.
Materials and methods
The model system in our study consists of the brassicaceous plant Brassica oleracea (L.) var. gemmifera (Brassicaceae), two common herbivores that feed on Brassica spp. and that coexist in time during most of the development of the plant and their respective main koinobiont parasitoids. The herbivores were larvae of the leaf chewer Pieris brassicae (L.) (large cabbage white butterfly) (Lepidoptera: Pieridae) and nymphs of the phloem feeder Brevicoryne brassicae (L.) (cabbage aphid) (Homoptera: Aphididae), and their respective parasitoids were Cotesia glomerata (L) (Hymenoptera: Braconidae) and Diaeretiella rapae (McIntosh) (Hymenoptera: Aphidiidae). The insect species were obtained from the basic insect rearing of the Laboratory of Entomology of Wageningen University, the Netherlands. The plants were grown in a greenhouse for 6 weeks, at 21 ± 3 °C (day) and 16 ± 3 °C (night), 70 ± 10% RH and 16 : 8 h day/night. Natural daylight was supplemented by metal-halide lamps (225 μmol s−1 m−2 PAR). The plants had c. eight leaves.
Insect development assays
Unparasitized and parasitized P. brassicae caterpillars and B. brassicae nymphs were allowed to develop until pupation while feeding on the plants in isolation (single infestation) or together (multiple infestation) with their interguild counterpart, either coexisting at the same time (multiple simultaneous infestation) or temporarily separated by a period of 6 days (multiple sequential infestation) (Fig. 1). The experimental plants were placed in a greenhouse compartment at 21 ± 2 °C (day) and 16 ± 2 °C (night), 70% RH and L : D 16 : 8 h. To prevent the insects from moving among plants, the plants were placed 30 cm apart and a layer of 1–2 cm of water was maintained on the table during the experiment to ensure the isolation. A number of fitness correlates were measured along their development to subsequently compare the performance of the herbivores and their parasitoids under single and multiple infestations.
Measuring the performance of the leaf chewer and its parasitoid (Fig. 1a1)
Ten plants were each inoculated with two unparasitized and two parasitized neonate caterpillars of P. brassicae. Parasitism of the caterpillars was carried out by individually offering the neonate caterpillars to single C. glomerata females in plastic vials and removing the caterpillars that had been observed to be parasitized. The caterpillars were gently placed on the youngest fully developed leaf of the plant and were kept in a clip cage for the first 3 days to force them to feed on the selected leaf. In all treatments, the clip cages were placed gently on the leaf and were supported by sticks to avoid any pressure from their weight or damage to the leaf. To avoid food limitation, the clip cages were removed from the plant 3 days after caterpillar inoculation, and cotton wool was placed around the petiole as a physical barrier to keep the caterpillars restricted to feeding on the same leaf. After 6 days, the leaf did not provide sufficient tissue to feed the caterpillars and, therefore, the cotton wool barrier was removed allowing the caterpillars to move and feed freely on the plant until pupation.
Measuring the performance of the phloem feeder and its parasitoid (Fig. 1a2)
Ten plants were each inoculated with two unparasitized and two parasitized newly emerged aphid nymphs. Parasitism had been carried out by individually offering the nymphs to single D. rapae females in plastic vials, under a stereomicroscope, removing each nymph that had been observed to be parasitized. Nymphs were placed on the youngest fully developed leaf of the plant and kept in a clip cage during the first 6 days to force them to feed on the inoculated leaf.
This treatment was conducted to measure the performance of the caterpillar and its parasitoid on plants previously attacked by aphids (Fig. 1b1) and to measure the performance of the aphid and the aphid parasitoid on plants previously attacked by caterpillars (Fig. 1b2).
Measuring the performance of the leaf chewer and its parasitoid on plants previously attacked by the phloem feeder (Fig. 1b1)
Ten plants were each infested with 15 first-instar nymphs of the aphid B. brassicae. The nymphs were placed on the youngest fully developed young leaf of the plant and kept in a clip cage to force them to feed on the inoculated leaf. Six days later, all aphid nymphs and clip cages were removed from the plant. Immediately thereafter, two unparasitized and two parasitized neonate caterpillars of P. brassicae were placed on each plant, on the leaf just younger than the aphid pre-infested leaf. Clip cages and cotton wool barriers were utilized as previously described.
Measuring the performance of the phloem feeder and its parasitoid on plants previously attacked by the leaf chewer (Fig. 1b2)
Ten experimental plants were each inoculated with 15 first-instar larvae of the leaf chewer P. brassicae, which were placed on one of the first fully developed young leaves of the plant. To maintain the caterpillars’ feeding on the initially infested leaf, a clip cage was used, which was later replaced by a barrier of cotton wool around the petiole of the leaf as was explained in detail elsewhere. Six days after the infestation, the caterpillars and cotton wool barriers were removed from the plants. Immediately thereafter, two unparasitized and two parasitized newly emerged aphid nymphs were placed on the leaf just younger than the caterpillar-infested leaf. The nymphs were enclosed within a clip cage to maintain them feeding on the selected leaf and to ensure that they would not feed on the caterpillar pre-infested leaf. Clip cages were utilized as previously described.
This treatment was conducted to measure the performance of both the caterpillar and the aphid, and their respective parasitoids, when they feed on the plant at the same time. Two unparasitized and two parasitized neonate caterpillars and two unparasitized and two parasitized first-instar aphid nymphs were simultaneously placed on the two youngest fully developed leaves of each of ten experimental plants. Both insect species were kept feeding for 6 days on the inoculated leaves, as described elsewhere. Clip cages and cotton wool barriers were utilized as previously described.
To subsequently compare the growth and development of the leaf chewer and its parasitoid under the different scenarios/treatments, we measured the larval survival at 24 h since the inoculation, the juvenile development time until pupation and the larval weight at 3 and 6 days after inoculation, the fresh cocoon weight of the parasitoid right after pupation and the parasitoid juvenile developmental time. To compare the performance of the phloem feeder and its parasitoid, we measured nymphal survival after 24 h, total juvenile development time and the percentage of nymphs moulting to the second instar within the first 36 h since inoculation to investigate initial effects on aphid development. Aphid reproduction was evaluated by daily counting the number of nymphs during 6 days, removing them from the plant while counting. The densities that were selected for the experiments were such that the physical damage exerted by the aphids or caterpillars during the pre-infestation can be considered to be negligible.
Molecular and chemical analyses
In a separate experiment, a number of selected chemical and molecular characteristics were analysed for the plants under the different treatments. The level of the phytohormones SA and JA and the transcript levels of the following seven well-characterized plant defence-related genes, and three reference genes, were measured for the different treatments: five genes related to the JA pathway, a JA-responsive transcription factor (BoMYC: GenBank accession, EF423803), myrosinase from the glucosinate–myrosinase system (BoMYR: GenBank accession, DQ456999), lipoxygenase in the octadecanoid pathway (BoLOX: GenBank accession, EF123056), plant defensin (BoDEF: GenBank accession, EF423802) and a cysteine proteinase inhibitor (BoPIN: GenBank accession, EF423805); two SA-responsive genes, a pathogenesis-related protein (BoPR1: GenBank accession, EF423806) and phenylalanine ammonia-lyase (BoPAL: GenBank accession, EF423804); and three reference genes, glyceraldehyde-3-phosphate dehydrogenase (BoGADPH: GenBank accession, EF123055), acetyl-CoA carboxylase (BoACC1: GenBank accession, AF0445731) and the actin encoding gene (BoACT1: GenBank accession, AJ496179). We sampled plants of the different treatments over time, at the initial time points of 5, 15 and 25 h after infestation and at a later time point of 144 h. For each time point, a new set of ten plants per treatment was used, and the first fully developed young leaf comparable to the leaf where the attackers were inoculated to study their performance in the performance experiment was sampled in all treatments. The sampling of these leaves was meant to reveal the dynamics of the chemical and molecular states of the leaf that the newly emerged caterpillar/aphid faced when arriving to an experimental plant in the performance experiment. The later time point sampled, 144 h (=6 days), coincides with the time of the inoculation of the second attacker in the sequential infestation. One leaf per plant was sampled for the analysis of both gene transcription and phytohormones, from which 1·5-cm-diameter leaf discs were cut, to reduce variation within treatments as much as possible. From the same leaves, carbon and nitrogen levels were also quantified. The discs were introduced in Eppendorf tubes and immediately frozen in liquid nitrogen and stored at −80 °C for subsequent analysis. The analysis of phytohormones was carried out according to Schulze et al. (2006). Total RNA extraction, purification and cDNA synthesis, primer design and gene expression quantification through real-time qPCR were performed as described by Vandesompele et al. (2002) and Hellemans et al. (2007). Primary plant compounds were analysed as described by Troelstra et al. (2001). Secondary plant compounds (glucosinolates) were quantified as in Badenes-Perez et al. (2011) at the time point of 144 h for plants exposed to caterpillars and plants exposed to aphids and undamaged control plants. It is well documented for several brassicaceous species that glucosinolates are not induced within a few hours of herbivore attack and, therefore, initial measurements of glucosinolates were not taken (for a review, see Hopkins, van Dam & van Loon 2009; Textor & Gershenzon 2009).
The comparison of the fitness correlates of the insects was based on a mixed model. The model has one factor (treatment), with three levels [single infestation (control), pre-infestation and simultaneous infestation], and two random terms that correspond to the between-plant variation and to the within-plant variation, where both are assumed to be independent and normally distributed. Model fitting was performed by restricted maximum likelihood (REML) in Genstat 8. Levels of primary and secondary plant metabolites were statistically analysed using a general linear model (analysis of variance). The data corresponding to phytohormones were log transformed for normalization and were subsequently tested using a general linear model (unbalanced analysis of variance). The expression of the selected genes was adjusted relative to the three reference genes described in the methodology and tested using a generalized linear model, with Poisson distribution and log link function.
The performance of P. brassicae caterpillars was significantly affected by the presence of the aphid B. brassicae, although the intensity of the effects was dependent on whether the aphids had previously infested the plants or were infesting the plants simultaneously with the caterpillars (Fig. 2). The caterpillars developed faster (F = 3·4, P = 0·04) and reached a larger size (F = 10·2, P < 0·0001) when feeding on host plants previously colonized by aphids (Fig. 2a,b), while caterpillar survival did not differ among treatments (F = 0·0, P = 1·0; Fig. 2c). Similar, but less pronounced effects were recorded when the caterpillars developed on a plant that was simultaneously infested with aphids. In contrast, the performance of the aphid B. brassicae was minimally influenced by the presence of caterpillars on the plant (Fig. 2). The first-instar nymphs moulted more rapidly to the second instar when sharing their host plant with caterpillars (F = 5·3, P = 0·007; Fig. 2d), and this effect was independent of whether the caterpillars had infested the plant previously or were feeding simultaneously on the host plant. However, the enhanced aphid development initially observed was subsequently diluted, and nymphs moulted to adults at a similar time in all treatments (F = 0·4, P = 0·6; Fig. 2e). The survival of B. brassicae nymphs also did not differ among treatments (F = 0·9, P = 0·4; Fig. 2f). Aphids reproduced equally well under all treatments, producing a similar number of nymphs in all infestation scenarios throughout the duration of the experiment (F = 0·2, P = 0·7; Fig. 2g). The reciprocal effects between the two herbivores were also transferred to the next trophic level, affecting the performance of their parasitoids (Fig. 3). While the development time of C. glomerata was similar in all infestation scenarios (F = 1·5, P = 0·2; Fig. 3a), the size of the parasitoid cocoon was bigger when the parasitized caterpillars had been feeding on host plants previously infested by aphids (F = 12·9, P < 0·001; Fig. 3b). Diaeretiella rapae parasitoids developed more rapidly in aphids feeding on plants infested by caterpillars, either when feeding simultaneously or sequentially (F = 4·1, P = 0·02; Fig. 3c), attaining a similar size on plants of the different treatments (F = 0·6, P = 0·5; Fig. 3d).
The percentage of carbon and nitrogen was not influenced by herbivory (F = 2·1, P = 0·1 and F = 0·7, P = 0·5, respectively) but significantly decreased over time, independent of the presence or absence of herbivores (data not shown). Six days after inoculation (144 h time point), the percentage of carbon slightly but significantly decreased (F = 14·3, P < 0·001) and the percentage of nitrogen decreased (F = 169·3, P < 0·001), although nitrogen levels were always relatively high, above 2% (data not shown). Neither total nor individual levels of glucosinolates differed significantly among treatments (F = 1·57, P = 0·22) (Fig. 4a). Aphid infestation significantly affected the levels of JA; after 144 h, JA was reduced c. 10 times in the higher systemic leaf of the plants with aphids or caterpillars plus aphids (F = 36·9, P ≤ 0·001). The level of the phytohormone SA was similar among treatments for all time points (F = 0·96, P = 0·41) (Fig. 4b,c).
The systemic responses of LOX, a class 2 lipoxygenase that shares characteristics with many LOX genes in class 2 that code for enzymes involved in an early step in JA biosynthesis (Zheng et al. 2007), and MYC, a transcription factor involved in the JA-signalling cascade, significantly differed after aphid and caterpillar attack (F = 14·3, P < 0·001 and F = 12·1, P < 0·001, respectively; Fig. 5). Both LOX and MYC had similar transcript levels in undamaged plants and in plants attacked only by aphids, while significantly higher transcript levels were found in plants attacked by caterpillars, independently of the presence of aphids, 144 h after attack (Fig. 5a,b). A significantly higher transcript level of MYC was observed earlier, 25 h after attack, but only when the plant was attacked by aphids and caterpillars simultaneously. The expression level of PIN, a JA-responsive gene, was first up-regulated in aphid-infested plants 15 h after attack, subsequently up-regulated in plants attacked simultaneously by aphids and caterpillars and finally up-regulated in aphid-infested plants, caterpillar-infested plants and plants infested simultaneously by aphids and caterpillars 144 h after attack (F = 3·9, P < 0·001; Table 1). The transcript levels of the other measured JA-responsive genes, DEF (F = 0·6, P = 0·7) and MYR (F = 17·5, P = 0·2), showed no differences among treatments (Table 1 and Fig. 5c, respectively).
Table 1. Transcript levels of PIN, DEF and PR1, relative to the transcript levels of three reference genes GADPH, ACC1 and ACT1, at 5, 15, 25 and 144 h in plants undamaged (C), after infestation by the phloem feeder Brevicoryne brassicae (B), after the infestation by the leaf chewer Pieris brassicae (P) and after the simultaneous infestation of the leaf chewer and the phloem feeder (B + P)
B + P
Different letters indicate significant differences (P < 0·05) between treatments within a time point.
The transcript levels of PAL (F = 1·8, P = 0·08), a gene involved in SA biosynthesis, showed no significant differences among treatments (Fig. 5d). On the other hand, the transcript level of the SA-responsive gene PR1 was lower in plants attacked by aphids (F = 2·6, P = 0·01) and intermediate in plants attacked by caterpillars or by both aphids and caterpillars 25 h after infestation compared to control plants. After 144 h, the PR1 transcript level in plants attacked by both aphids and caterpillars was significantly higher than in control plants (Table 1).
Our study shows that a previous infestation of B. oleracea plants by B. brassicae aphids facilitates the growth and development of P. brassicae caterpillars. This positive effect was less pronounced when caterpillars and aphids simultaneously infested the plant. This suggests that the response of the plant to the aphids requires time to build up and, therefore, the caterpillars only benefit when they arrive later on the plant. Alternatively, the response triggered by the aphids may be stronger when the plant is exposed solely to aphids, while milder responses are induced when the plant is exposed to both attackers – from contrasting feeding guilds – simultaneously. The aphid B. brassicae initially performed better on plants shared with caterpillars: The nymphs reached the second stage earlier than when feeding on a control, previously undamaged, plant. However, the overall juvenile development was equally fast on plants with or without caterpillars. The slight benefit that aphids achieved initially from the interaction between the plant and caterpillars was similar on plants that had experienced caterpillar infestation previously or simultaneously. The facilitation between the two herbivores, therefore, is asymmetric. Considering that the first attacker was removed from the plant before the arrival of the second attacker and that the second attacker was not allowed to feed on the leaf colonized by the first attacker, facilitation occurred across both time and space. Unlike competition, facilitation between insect herbivores seems to be a relatively uncommon outcome of interspecific plant-mediated interactions (Kaplan & Denno 2007). It is possible that the proposal of competition as the most common outcome of interspecific interactions emerges as a result of an overrepresentation of studies addressing insect herbivores from similar feeding guilds. Therefore, including more studies of interguild interactions may substantially change the view that competition is the major plant-mediated effect of one herbivore on another.
Levels of carbon and nitrogen were similar among treatments, suggesting that the insect densities used in the experiments did not weaken the nutritional status of the plants. Glucosinolates did not significantly differ among treatments, indicating that these secondary plant compounds did not contribute to the observed facilitation. A significant tenfold reduction in JA concentration in aphid-infested plants, together with the low expression of well-known JA-responsive genes in plants with aphids, suggests that facilitation was mediated by interference with signal transduction. Leaf chewers have been consistently reported to perform better on JA-deficient mutant plants (Kessler & Baldwin 2002), and therefore, it is likely that the significantly lower levels of JA in plants infested by aphids play a role in the plant-mediated facilitation for subsequent attacking caterpillars. Aphids and whiteflies have been reported to induce SA levels that subsequently weaken JA-related responses mediated by cross-talk between signalling pathways (for reviews see Walling 2008; Giordanengo et al. 2010). However, such SA induction can be a local phenomenon around the site of aphid infestation (De Vos et al. 2005). We recorded overall similar levels of SA in our experimental plants with and without aphids, and therefore, we did not find evidence of indirect suppression of JA via increased SA levels. Overall, our findings are in line with results reported in the previous molecular studies that suggest that the saliva of phloem feeders might contain effectors that actively suppress JA (De Vos et al. 2005; De Vos, Kim & Jander 2007; Zhang et al. 2009). The JA concentration in plants simultaneously exposed to caterpillars and aphids similarly showed a tenfold reduction in JA concentration, but caterpillars only performed better on plants previously infested by aphids. However, it is important to remember that the low JA concentration was observed only 144 h after infestation and this relates to different ‘developmental points’ for the caterpillars on the sequential and simultaneous infestation treatments. In the sequential infestation, this time point referred to the time when the just-emerged caterpillars were inoculated, while for the simultaneous infestation it referred to 6 days after caterpillar inoculation when the caterpillars have further developed and could be less vulnerable to plant defences. In fact, the JA concentration in plants simultaneously attacked by aphids and caterpillars did not differ from control plants during the first 25 h, showing that neonate caterpillars in the simultaneous infestation were not confronted with a reduced JA level.
Overall, the selected gene involved in JA biosynthesis (LOX) and the transcription factor involved in the JA-signalling cascade (MYC) were not up-regulated in plants attacked by aphids, illustrating the suppression of JA induction in aphid-infested plants also at the gene level 144 h after aphid attack. Importantly, 144 h is the time point when the first attacker was removed from the plant in the sequential infestation treatment and the second attacker was inoculated. Therefore, our results show that caterpillars that were placed on plants previously infested with aphids were initially feeding on leaves that had tenfold times less JA – consistent with the low expression of LOX and MYC. It is important to underline that phytohormones and the expression of the selected genes were measured not on the infested leaves (local effects) but in younger (systemic) leaves. Most data linking phytohormone levels and changes in the expression of JA- and SA-related genes refer to local measurements at the site of attack. The speed of signalling between leaves remains poorly understood; even within the same leaf, the transcript levels of certain genes can quickly increase at the precise site of attack but not a few centimetres away (Stork et al. 2009). Systemic effects are less well documented, particularly for plant–aphid model systems because phloem feeders are believed to manipulate plant defences at the site of attack (Moran & Thompson 2001; Thompson & Goggin 2006; Walling 2008), making direct comparisons with our results difficult (but see Divol et al. 2005; Heil & Ton 2008). Unexpectedly, the cysteine proteinase inhibitor gene PIN that is JA-regulated was up-regulated by aphids. The expression of MYR, a myrosinase from the glucosinate–myrosinase system, was similar among treatments and consistent with the similar levels of glucosinolates recorded in all treatments. The expression of PAL, a gene involved in SA biosynthesis, did not differ among treatments, which is consistent with the similar levels of SA among treatments. The SA-responsive gene PR1, however, showed erratic responses, significantly differing among treatments and over time. Although the differences in the expression of PR1 were statistically significant, the levels were always within one and a half fold change. Yet, it is remarkable that aphid infestation at such low densities clearly induced plant-mediated effects on the growth and development of the caterpillars and that such facilitation was overall reflected in the plant.
The facilitation between the herbivores also had consequences for their parasitoids. The increase in the larval weight observed in P. brassicae when feeding on plants previously colonized by aphids was carried over to its parasitoid C. glomerata. This is not surprising because a better-quality host is known to provide a better food source for its parasitoid (Harvey 2005; Rodriguez-Saona et al. 2005). Body size in parasitoids is correlated with fitness-related characteristics such as searching and mating efficiency, longevity and fecundity (Harvey, Harvey & Thompson 1994; Bezemer & Mills 2003; Harvey 2005). In case of the aphid parasitoid D. rapae, it is interesting that even when the development of the aphids appeared to be only slightly faster initially on plants shared with caterpillars, the koinobiont parasitoid developed significantly faster in hosts feeding on plants previously attacked by caterpillars. Development time is another key target of selection for insects, especially in environments with a high selection pressure of natural enemies; attaining the maximum size in the shortest time will reduce the risk of predation and parasitism, thus maximizing survival and therefore fitness (Loader & Damman 1991). Therefore, in the aphid trophic chain positive effects that were not evident at the herbivore level became apparent for the next trophic level. Overall, both the caterpillar– and aphid–parasitoid benefited from the facilitation between their hosts and nonhosts from different guilds. Parasitoids depend exclusively on the quality of the single host that was selected by their mother, which implies a strong selection pressure for females to find the most profitable hosts for their offspring. Variation in host quality among patches, resulting from the presence of aphids on some of the plants, is expected to influence the preference of female parasitoids searching for hosts. If parasitoids preferentially forage for hosts feeding on aphid-infested plants, in environments with a high pressure of natural enemies, the facilitation that caterpillars can obtain on aphid-infested plants might be compromised.
Overall, our results show a positive asymmetric effect in indirect plant-mediated interactions between herbivores of different feeding guilds. The leaf chewer clearly benefited when feeding from plants previously attacked by the aphid while aphids performed slightly and only initially better on plants with caterpillars. Reduced levels of JA and low expression of well-known JA biosynthesis and JA-regulated genes in systemic leaves of plants attacked by aphids are consistent with the previous molecular findings and suggest that the facilitation in terms of caterpillar development was mediated by interference with signal transduction as a result of aphid feeding. Cross-talk between signalling pathways has been suggested to be a mechanism that allows plants to fine-tune their defences (Pieterse et al. 2009), but if facilitation frequently occurs in interguild interactions among herbivores, it may represent a constraint for plant defences when insects from different guilds feed on the same plant. However, such a constraint might be reduced if, as in our model system, parasitoids are found to also benefit from plant-mediated interactions between hosts and nonhosts from different feeding guilds. Parasitoids are known to affect the outcome of behavioural interactions between plants and herbivores under single infestation (Price et al. 1980; Schmitz, Hamback & Beckerman 2000), but surprisingly the impact of cross-talk on higher trophic levels than herbivores and plant pathogens remains unexplored so far. Exploring interguild interactions in different ecological model systems, considering the effects and feedback on natural enemies on the plant-attacker system, linking the performance/behavioural outcome of the interactions and the underlying molecular mechanisms is of crucial importance to further understand the complex functioning of plant–insect communities. A central research question in this emerging field is how responses to single attackers interfere, at the molecular, phytochemical and behavioural levels, with responses to secondary attackers with distinct feeding modes. This study contributes to a first step towards a better understanding on how double infestation affects tritrophic interactions.
We thank André Gidding, Frans van Aggelen and Leon Westerd for culturing the insects and Andrea Lehr for helping with the phytohormone extraction. We thank Nawaporn Onkokesung and Erik Poelman for their comments on a previous version of the manuscript. RS was financially supported by a VENI grant (nr 863.08.028) from the Earth and Life Sciences Foundation (ALW), which is subsidized by the Netherlands Organization for Scientific Research (NWO).