In nature, plants interact with many organisms and need to integrate their responses to these diverse community members. Knowledge on plant–insect relationships has accumulated rapidly during the last decades. Yet most studies on direct or indirect defences of plants against herbivory have treated herbivores as individual stressors. However, herbivores often consist of communities themselves, comprising organisms such as parasites and symbionts, which may have important effects on the herbivore phenotype, and consequently on interactions of the herbivore with its food plant. Here, we review how herbivore-associated organisms affect plant–herbivore interactions. Organisms associated with herbivores can directly affect how a plant interacts with their herbivorous hosts, by interfering with plant signal-transduction pathways, repressing the expression of plant defence-related genes, or altering plant secondary metabolism. In addition, herbivore-associated organisms can also affect plant responses indirectly by their effect on the behaviour and physiology of their herbivore host. The changes in plant phenotype that arise from herbivore-associated organisms may subsequently affect interactions with other community members, thereby impacting community dynamics. Furthermore, herbivore-associated organisms may act as a hidden driving force of plant–herbivore coevolution. Therefore, to understand plant–herbivore interactions it is important to realize that every single herbivorous insect constitutes a community in itself.
As members of diverse ecological communities, plants and insect herbivores have coevolved for c. 350 million yr. Insects are the most speciose group of organisms on the planet, and c. 50% of them feed on plants (Schoonhoven et al., 2005).
In natural ecosystems, plants interact with many organisms simultaneously, which may influence the pairwise interactions between plants and insects profoundly (Fig. 1a; Stout et al., 2006; Stam et al., 2014). When a plant is attacked by multiple attackers, the responses of the plant to the individual attackers may interact and consequently result in unique plant responses based on the order of colonization, type of feeding behaviour and time lag between arrival of the attackers (Voelckel & Baldwin, 2004; Stam et al., 2014). In fact, plants are not alone when interacting with herbivores. Organisms associated with plants may affect the interactions between plants and herbivores either positively or negatively (Philippot et al., 2013). The presence of a microbial community on plant roots may affect the growth and defence phenotype of a plant and thereby influence multitrophic interactions in the rhizosphere and plant-mediated below-ground–above-ground species interactions (Oldroyd, 2013). Some plant-associated organisms, such as endophytes, are even integrated in plants and provide plants with additional defence properties against insect herbivory (Kogel et al., 2006). Therefore, plants and their associated organisms constitute a community that is faced with the challenges imposed by herbivores.
It has often been ignored in studies of plant–herbivore interactions that each individual herbivore also represents a community in itself, consisting of different herbivore-associated organisms (HAOs; Fig. 1b). Yet it is well known that all higher organisms are complexes of many species that live in symbiosis and which may determine the phenotype of the individual with which they are associated (Gilbert et al., 2012). For insect herbivores, for example, it is well known that aphids harbour important endosymbiotic bacteria that provide them with nutrients, protect them against parasitism or aid them in dealing with plant defences (Douglas, 2009; Frago et al., 2012). The composition of the herbivore-associated community may be affected by the secondary compounds of the herbivore's food plant (Kohl & Dearing, 2012). However, in addition, HAOs may also influence plant responses to insect herbivory. One of the major groups of HAOs consists of insect parasites that live in or on their host and extract resources from it, leading to a loss of host fitness (Hughes et al., 2012). The insect parasites can be microorganisms (such as fungi and bacteria), viruses or macroorganisms (such as parasitic worms and parasitic wasps; Fig. 1b; Hughes et al., 2012). Insect parasites have evolved remarkable strategies to manipulate their host's development, physiology, morphology, evolution and ecology (van Houte et al., 2013). Yet other HAOs may be beneficial to herbivores, such as the endosymbiotic microbes of aphids (Douglas, 2009). The presence of HAOs often results in an extended phenotype of the insect host, and this may affect induced responses of plants to feeding damage by their insect host. There is growing evidence of the importance of HAO in plant–insect interactions, which suggests that we should consider the herbivore and its hidden associated community of HAOs as an integrated stressor that interacts with the plant (Fig. 1c; Frago et al., 2012; Hughes et al., 2012).
In this review, we discuss recent progress in the study of plant–insect interactions, with a focus on how plants deal with integrated stressors of herbivores and their associated organisms. We review how HAOs directly affect plant responses to herbivory; how HAOs indirectly affect plant responses to herbivory by affecting herbivore behaviour and physiology; how plant responses to integrated stressors result in altered interactions of plants with other community members; and whether HAOs are involved in plant–insect coevolution. Finally, we provide future directions for studying these interactions using genomics tools.
HAOs directly affect plant responses to herbivory
To cope with attack from herbivores, plants have evolved sophisticated direct and indirect defences (Karban & Baldwin, 1997; Schoonhoven et al., 2005; Heil, 2008; Dicke & Baldwin, 2010; Kessler & Heil, 2011). To activate defence responses, plants recognize insect attack by their damage pattern and by perceiving herbivore-derived chemical cues, such as herbivore-associated elicitors or herbivore-associated molecular patterns (HAMPs; Bonaventure, 2012). The elicitors induce signal-transduction pathways regulated by phytohormones and gene transcripts that modulate herbivory-induced responses in plants (Erb et al., 2012; Pieterse et al., 2012). HAOs may come into contact with plants and affect the induction of plant defence responses, secondary metabolism, physiological status and, consequently, plant–herbivore interactions (Kaiser et al., 2010; Chung et al., 2013; Luan et al., 2013).
Plants are able to induce specific responses to herbivory, affected by the identity of the attacker. These finely tuned induced plant responses can depend on the specialization and feeding guild of the insect herbivores (Voelckel & Baldwin, 2004; Ali & Agrawal, 2012; Zhang et al., 2013). Several HAMPs that plants use in herbivore recognition have been identified in the regurgitant (Bonaventure, 2012) of caterpillars that come into contact with plants during herbivore feeding (Vadassery et al., 2012). However, plant wounds are open not only to herbivore elicitors, but also to the community of microorganisms that inhabits the foregut of these caterpillars (Fig. 2a). Moreover, these microorganisms may be expected to influence HAMP-dependent herbivore recognition, just as parasitic wasps developing in a herbivore may influence elicitor-mediated plant responses (Poelman et al., 2011b). In addition, an increasing number of studies indicate that numerous insect species vector plant viruses or pathogenic bacteria and fungi that may influence plant responses to herbivory (Stout et al., 2006; Luan et al., 2013). Insects may benefit from vectoring plant pathogens, because the induced defence of plants against pathogens often interferes with the induced defence against insects. This is a result of the antagonistic cross-talk between the signal-transduction pathways activated in response to herbivore and pathogen attack (Stout et al., 2006; Thaler et al., 2012). For instance, Bemisia tabaci whiteflies perform better on tobacco plants infected with begomovirus that is vectored by the insects. The enhanced performance of insects is the result of suppression of the biosynthesis of major defence compounds, particularly terpenoids (Luan et al., 2013). Interestingly, antibiotic-treated Colorado potato beetle (Leptinotarsa decemlineata) larvae lose the ability to suppress antiherbivore defences in tomato; extensive analyses show that microbial symbionts residing in the beetle's oral secretions are involved in defence suppression (Chung et al., 2013). When honeydew excreted by aphids drops onto the plant, it may suppress defence-related jasmonic acid accumulation by inducing salicylic acid, suggesting that bacteria within the honeydew may make plants less resistant to the aphids (Fig. 2a; Schwartzberg & Tumlinson, 2014). Moreover, a transcriptomics analysis of maize plants revealed that defence-related genes were down-regulated by feeding of beetles carrying endosymbiotic Wolbachia bacteria (Barr et al., 2010). The insect vectors can also benefit from virus-infected plants with increased growth rates, and consequently a reduced period of vulnerability to predation (Belliure et al., 2008). Although the virus may benefit its insect vector by suppressing plant defensive responses, it may cause negative effects on nonvector insects that feed from the same plant (Donaldson & Gratton, 2007).
Herbivore-associated organisms may also influence plant–insect interactions by altering the emission of plant volatiles that make plants apparent to other community members and consequently play an important role in interactions within ecological communities. This has mainly been studied in tripartite pathogen–insect vector–plant interactions. Increasing evidence indicates that bacterial pathogens and viruses are able to alter the foliar and floral volatile emissions of their host plants, consequently enhancing both vector recruitment to infected plants and subsequent dispersal to healthy plants, thus revealing pathogen–insect mutualisms (Mauck et al., 2010; Shapiro et al., 2012). Compared with effects of pathogen-induced plant volatiles on vector attraction, little is known about how nonvectors and community members from other trophic levels respond to these induced changes in plant volatiles, but it is likely that other community members respond to these induced changes in the plant phenotype (Dicke & Baldwin, 2010).
HAOs indirectly affect plant responses to herbivory
Herbivore-associated organisms may also interact indirectly with host plants via their herbivore host, without physical contact of the HAO with the plants. HAOs are well known for their host manipulation abilities, both behaviourally and physiologically (Fig. 2b; Godfray, 1994; Hughes et al., 2012; van Houte et al., 2013), resulting in extended phenotypes of their herbivore hosts. Thereby, the presence of HAOs could lead to altered herbivore traits that might affect plant responses to herbivory (Fig. 2b).
HAOs influence host behaviour
The presence of HAOs often leads to changes in the behaviour of the host insect, including reproduction, feeding behaviour and locomotion (Hughes et al., 2012). Parasite-induced changes in host behaviour are often thought to increase the fitness of the parasite and may be actively driven by the parasite (Lefevre et al., 2009). Changes in movement of hosts as a result of parasitism have been well investigated in different parasite–host systems (van Houte et al., 2013). For instance, fungal or viral infections may manipulate the behaviour of their insect host such that the host now moves to the top of the canopy, which is beneficial for reproduction and spread of the parasites (Hoover et al., 2011). Also, parasitic worms or wasps manipulate host movement for their own benefit (Godfray, 1994; Biron et al., 2006; Libersat et al., 2009; van Houte et al., 2013). Because leaves at different positions in a plant may differ in their responses to herbivory (Rostas & Eggert, 2008), differential distribution of feeding by infected vs uninfected herbivores may result in differential spatial arrangements of the induced plant phenotype, which may consequently affect other attackers.
In some extreme cases, parasitoids manipulate their host to the extent that it becomes a ‘bodyguard’ that physically protects the parasitoids against subsequently approaching predators (Harvey et al., 2008). Parasites can also induce changes in the feeding behaviour of their host, and such effects are often specific to the species of parasite developing in the herbivore (Godfray, 1994; Poelman et al., 2011a). Because feeding behaviour characteristics influence plant responses (Mithofer et al., 2005), parasites may indirectly influence plant responses through their effects on feeding behaviour.
Little is known about the molecular mechanisms behind manipulations of host behaviour by parasites; the available knowledge has been gained primarily from model systems using viral parasites (van Houte et al., 2013). For more complex organisms, genes and/or proteins of other parasites (such as bacteria, fungi, parasitic wasps) involved in behavioural manipulation of the host have been less well studied. However, the observed changes in herbivore movement patterns show high similarity across different groups of HAOs. These similarities indicate that mechanisms behind host manipulation may be highly conserved among parasites to maintain their parasitic life history (Ponton et al., 2006). On the other hand, similar patterns in behavioural changes in host herbivores may also indicate conserved strategies of the hosts in response to parasitism. Future studies are required to unravel why and how the parasites alter their host's behaviours.
HAOs influence host physiology
In addition to host behaviour manipulation, HAOs also alter host physiology. The presence of HAOs can affect host development. When developing in their herbivore host, parasitoids influence host growth by interfering with the production of juvenile hormone and ecdysone of their host, which are responsible for maintaining the juvenile characters of the host and initiate moulting to the next larval instar, respectively (Godfray, 1994). Parasitoids may induce their hosts to stay longer in the larval feeding stage, which has been shown for the gregarious parasitoid Cotesia congregata, which parasitizes tobacco hornworm (Manduca sexta) caterpillars (Godfray, 1994). The parasitoid prevents metamorphosis of its host larva by suppressing the drop in juvenile hormone production before pupation, leading to a sixth supernumerary larval stage. This prolonged feeding stage of the host is beneficial to the parasitoid larvae, allowing them to acquire more nutrients. By contrast, the solitary parasitoid Cotesia rubecula arrests the growth of its host, caterpillars of Pieris rapae, in the third or fourth larval instar (Harvey et al., 1999). These changes in host physiology affect feeding rate and might thus affect plant responses to herbivory. Parasitoid species could even have a further unique effect on their herbivore host's physiology, by altering the herbivore's oral secretion, which plays a vital role in eliciting plant responses (Poelman et al., 2011b).
Symbiotic microbes of insect herbivores could also contribute significantly to modulation of host physiology. Microbial symbionts can provide essential nutrients to the host, such as amino acids, vitamins and sterols (Douglas, 2009). Symbionts of herbivorous insects could greatly improve nutrient uptake and open niches to their insect host, allowing colonization of a broad range of host plants (Douglas, 2009). The identity of microbial symbionts of phloem-feeding herbivores affects the capacity of the herbivores to switch between food plant species (Tsuchida et al., 2004; Oliver et al., 2010). Microbial symbionts may also contribute to herbivore resistance to insecticides; for example, susceptibility to insecticides in the silverleaf whitefly Bemisia tabaci depends on the density of endosymbionts (Ghanim & Kontsedalov, 2009). Whether these effects of microbial symbionts on herbivore physiology affect plant responses to herbivory remains to be investigated.
Similar to physiological modulations, the immune system of an insect herbivore is not only regulated by the herbivore itself but also by HAOs. Insects largely depend on their immune system to combat invasions by other organisms. HAOs could provide their host with protection against a wide range of natural enemies (Oliver et al., 2014). Some symbionts can directly protect the host from attack by natural enemies by producing toxins or deterrents (Hansen et al., 2012). Some others provide host protection indirectly by modulating the host immune system, such as Wolbachia bacteria that promote host resistance to viral infection in Drosophila fruit flies, resulting in protection of the fruit flies against a wide range of RNA viruses (Hedges et al., 2008). Although the mechanisms underlying host immune system modulation by HAOs remain to be further investigated, these direct and indirect protections provided by HAOs are likely to contribute to the ability of insect herbivores to overcome the challenges imposed by their food plants and their natural enemies (Oliver et al., 2010; Frago et al., 2012).
The extended phenotype of the herbivore that results from the HAO-induced behavioural and physiological manipulations affects the interaction of the herbivore with its food plant (Fig. 2b). Manipulations by HAOs of host-feeding behaviour, including amount of food consumed, feeding pattern and shifts in feeding sites (between old and young tissues or vegetative and reproductive tissues), could have important consequences for plant growth and defence responses. Moreover, the physiological changes that are expressed in the host's oral secretions affect recognition of the attacker and induced plant responses (Poelman et al., 2011b). The behavioural and physiological manipulations by HAOs could further indirectly affect plant responses to herbivory.
Community-wide consequences of HAO-mediated changes in plant–herbivore interactions
The fact that HAOs can manipulate the herbivore's phenotype, and consequently the herbivore's interaction with its food plant and the plant's responses, means that HAOs affect the plant phenotype (Fig. 2c). That these HAO-induced plant phenotypes result in altered interaction networks of the herbivore hosting the HAOs and other plant-associated organisms has been well established for interactions between parasitoids and their herbivore host. The larvae of parasitic wasps that feed within their herbivorous host do not have physical contact with host plants. Yet, in a parasitoid species-specific manner, they affect the growth of the herbivore as well as the composition of its oral secretion and thereby interact with the host plant through their herbivorous hosts. The HAO-mediated altered composition of the oral secretions induced defence-related genes and volatile emissions differentially for the presence or absence of parasitoid larvae. Moreover, the parasitoid species can have a more pronounced effect on plant gene transcription than the herbivore species in which the parasitoid resided (Poelman et al., 2011b). The changes in plant phenotype subsequently affected foraging behaviour and performance of insects at the second up to the fourth trophic level (Poelman et al., 2011a,b, 2012). For instance, the herbivore Plutella xylostella (the diamondback moth) exhibits an altered oviposition preference: the moths preferred to oviposit on plants infested with unparasitized caterpillars than on plants infested with caterpillars parasitized by a parasitic wasp that cannot attack P. xylostella (Poelman et al., 2011b). In addition, plant responses induced by parasitoid larvae that develop within their host herbivore can also be perceived by top consumers at the fourth trophic level (Poelman et al., 2012). The hyperparasitoid wasp Lysibia nana differentiates between the blends of plant volatiles induced by unparasitized herbivores and herbivores carrying parasitoid wasp larvae, and uses this to successfully locate their hosts.
These interaction networks that are driven by direct and indirect effects of HAOs on plant traits are likely to be found for other tritpartite systems, such as virus–herbivore–plant or symbiont–herbivore–plant associations as well. Because plants are the basis of food chains in terrestrial ecosystems, phenotypic changes in plants may significantly influence the community structure and dynamics through bottom-up effects (Bukovinszky et al., 2008), and thus HAO effects on plant phenotypes have a strong potential to shape community processes.
HAOs as a potential driving force of plant–insect coevolution
Although it was previously known that some HAOs are able to manipulate host behaviour and physiology, we are now beginning to realize that these HAOs also play a role in the interactions between the their host and the food plants of their host. Unexpected interactions are being recorded between herbivores and their host plants when HAOs are considered as components of the plant–herbivore interaction (Poelman et al., 2011b; Frago et al., 2012; Chung et al., 2013; Luan et al., 2013). The emerging view is that plant–insect interactions across different trophic levels in food webs are more complex than commonly considered. The presence of HAOs may interfere with the plant to recognize its herbivore, for example, through interference with signal transduction in the plant and with defence responses. Thus, HAO-mediated effects result in extended phenotypes in plants. It is likely that HAOs are even involved in the coevolutionary arms race between herbivores and plants. Therefore, a major question concerns the driving force in herbivore–plant coevolution: is it the herbivore itself, the HAO, or the combination of herbivore plus its HAO as an integrated stressor? This is likely to have important consequences for our view on the evolution of plant–herbivore interactions. For instance, instead of evolving an adaptation to a plant defence through, say, enzymatic breakdown of a plant toxin, a herbivore could also evolve to interact with a new HAO that eliminates the effects of the plant defence. For instance, the lack of endosymbionts in the weevil Sitophilus linearis can be associated with a switch from feeding on nutrient-poor host plants to feeding on plants that provide a higher nutritional value (Clark et al., 2010). Moreover, genetic changes in the endosymbiont may affect selective pressures on the insect host (Clark et al., 2010). The herbivore may benefit from microbial evolution that results in microbial genotypes that evolve to deal with plant secondary metabolites. After all, a selective advantage for the herbivore also favours its endosymbionts. Because generation times of microbes are much shorter than generation times of insects, this may mean that adaptation can be even faster.
Most of the studies addressing herbivore-induced plant responses have been based on the assumption that herbivores interact with their host plants as ‘individuals’ (Gilbert et al., 2012). However, the phenotype of the herbivore that interacts with the plant is a complex community in itself and each of the community members can influence the herbivore's phenotype (Fig. 1b). The emerging view that HAOs have important effects on herbivore behaviour, development and a herbivore's interactions with host plants, gives rise to new research directions in the field of the evolutionary biology of plant–insect interactions.
The knowledge emerging from studies of tripartite microbe–plant–insect interactions, insect–microbe symbiosis and Herbivore-Induced Plant Volatiles (HIPV)-mediated plant–hyperparasitoid interactions urges us to consider HAOs as important hidden players in plant–insect interactions and to study the effects of HAOs on plant responses to herbivory (Frago et al., 2012). So far, studies have focused, in particular, on the effects of individual HAOs on host manipulations and plant–herbivore interactions. Because the herbivore constitutes a community of HAOs in itself, one of the challenges is to assess HAO composition and identity and their effects on the phenotype of their insect hosts. Metagenomic approaches provide excellent opportunities to characterize the entire microbiota that reside in or on herbivorous insects (Philippot et al., 2013). Further analysis of these communities will yield insight into HAO diversity and dynamics, as well as the interactions among HAOs, which could profoundly influence not only the phenotype of the herbivorous host but also that of the food plant of the herbivore.
To understand the role of HAOs in host manipulation, recent advances in genomics and proteomics provide reliable tools to study host–parasite interactions from the level of the individual to unravelling the underlying molecular mechanisms (Biron et al., 2006; Lefevre et al., 2007). Through this approach, host manipulation by specific parasites can be studied, and mechanisms can be compared between insect–HAO combinations. Apart from direct effects of HAOs on the herbivore, direct and indirect effects of HAOs on plant responses to herbivory can also be addressed. By comparing plant metabolome and transcriptome profiles in response to herbivores with or without HAOs, for example, the effects of HAOs on herbivore-induced plant responses can be investigated. With rapidly advancing sequencing techniques, we are no longer restricted to model species; genomic information for many other nonmodel but ecologically relevant organisms will become available and will aid studies in this field.
Although it is now recognized that plants are able to respond specifically to different attackers, we will never fully understand how plants cope with herbivores as integrated stressors when the effects of HAO are ignored. Each member of the plant–insect community constitutes a community in itself; therefore, studies of plant–insect interactions in fact address the interactions among different communities rather than interactions between individual organisms (Fig. 1c). Although there is a lack of information on associated organisms in the community members at the third or higher trophic level (Dicke, 1996), at least some viruses associated with parasitic wasps are known to be involved in parasitoid–host interactions (Harvey et al., 2013), and ant-associated bacteria are known to contribute to ant–plant defensive mutualisms (Gonzalez-Teuber et al., 2014). The changes in plant phenotype that are induced by a herbivore holobiome (sensu Gilbert et al., 2012) will affect other community members at different trophic levels and exert ‘bottom-up’ effects on the structure of the plant–insect metacommunity.
Combining information from different disciplines and at different degrees of biological complexity will deepen our understanding of how HAOs affect plant phenotypes through the manipulation of their insect host, resulting in community-wide consequences for HAO–plant interactions. It is not only herbivores that constitute communities; in fact, every macroorganism constitutes a community that includes microorganisms (Gilbert et al., 2012). This means that the units within food webs and communities of macroorganisms are actually all communities rather than individuals. Expanding studies of plant–insect interactions from interactions between individuals to interactions between communities raises fundamental questions on the key species that drive the system. This makes the study of plant–insect interactions more complex but definitely also more intriguing. Realizing that organisms often do not act as individuals will be the start of new, exciting developments in this research field.
Our research was financially supported by the Earth and Life Sciences Council of the Netherlands Organisation for Scientific Research (NWO-ALW) through Ecogenomics Grant 844.10.005.