Can plant–natural enemy communication withstand disruption by biotic and abiotic factors?

Abstract The attraction of natural enemies towards herbivore‐induced plant volatiles is a well‐documented phenomenon. However, the majority of published studies are carried under optimal water and nutrient regimes and with just one herbivore. But what happens when additional levels of ecological complexity are added? Does the presence of a second herbivore, microorganisms, and abiotic stress interfere with plant–natural enemy communication? or is communication stable enough to withstand disruption by additional biotic and abiotic factors? Investigating the effects of these additional levels of ecological complexity is key to understanding the stability of tritrophic interactions in natural ecosystems and may aid to forecast the impact of environmental disturbances on these, especially in climate change scenarios, which are often associated with modifications in plant and arthropod species distribution and increased levels of abiotic stress. This review explores the literature on natural enemy attraction to herbivore‐induced volatiles when, besides herbivory, plants are challenged by additional biotic and abiotic factors. The aim of this review was to establish the impact of different biotic and abiotic factors on plant–natural enemy communication and to highlight critical aspects to guide future research efforts.

The majority of studies on tritrophic interactions have been performed using monoclonal, herbaceous cultivated species under controlled conditions, which, while useful from a logistical standpoint, poorly reflect natural ecosystems where plants exist as mixedgenotype populations in heterogeneous landscapes, and usually interact with multiple biotic players under variable abiotic conditions (Bezemer & van Dam, 2005;Dicke & van Loon, 2000;Hunter, 2002;Takabayashi, Dicke, & Posthumus, 1994). An increasing number of field studies demonstrate that attraction of natural enemies to HIPVs is widespread under natural conditions, suggesting that volatile cues are sufficiently robust to withstand certain levels of environmental variation (Birkett et al., 2000;De Moraes, Lewis, Pare, Alborn, & Tumlinson, 1998;Kessler & Baldwin, 2001;Clavijo McCormick, Irmisch, et al. 2014;Thaler, 1999). However, the extent of the impact of interacting biotic and abiotic factors remains poorly documented.
During the last decade, attention has been paid on the potential effects of climate change on multitrophic interactions. However, as a recent meta-analysis reveals, of over 2000 selected publications on climate change and trophic interactions, the majority dealt with only two trophic levels, and only 15% evaluated the effects of one or more abiotic factors on the outcome of multitrophic interactions (Rosenblatt & Schmitz, 2014). This meta-analysis suggests that many climate change studies are overlooking ecological complexity, and a question emerges about how can we truly understand the consequences of climate change on these interactions if we do not yet grasp the range of variation occurring under "normal" natural conditions. Hence, one of the main challenges in the study of multitrophic interactions is progressing from evaluating linear systems under controlled settings, into more complex scenarios incorporating additional biotic and abiotic conditions (Dicke, van Loon, & Soler, 2009;Mumm & Dicke, 2010). As volatile compounds are a primary currency mediating plant communication, their study under complex scenarios is vital to understand the community dynamics and how biotic and abiotic factors shape these.

| MULTIPLE VARIABLES AFFECT PLANT VOLATILE EMISSIONS AND NATURAL ENEMY RESPONSES
The first attempts to understand and predict the outcome of tritrophic interactions under complex ecological settings come from the knowledge that different types of herbivore damage can elicit different defense signaling pathways. In general, phloem feeders (whiteflies and aphids) activate the salicylic acid (SA)-dependent shikimic acid pathway, while chewing insects (beetles and caterpillars) and cell-content feeders (mites and thrips) induce the jasmonic acid (JA)-dependent octadecanoic pathway. Each of these pathways regulates the expression of different sets of downstream genes associated with indirect plant defenses (i.e., those defenses promoting the efficiency of natural enemies to control herbivores (Gols, 2014), leading to the emission of distinct volatile blends (Erb, Meldau, & Howe, 2012;Heil & Ton, 2008;Walling, 2000).
Initial evidence that the JA and SA pathways act antagonistically led to the hypothesis that induced plant volatile phenotypes and the outcomes of volatile-mediated interactions may be predictable based on the knowledge of the attacker (Erb et al., 2012;Heil & Ton, 2008;Walling, 2000). For instance, a JA-inducing herbivore would be expected to disrupt the attraction of natural enemies of a SA-inducing herbivore under simultaneous attack and vice versa. Although this outcome is possible (Zarate, Kempema, & Walling, 2007), it is now apparent that knowledge of herbivore damage type is insufficient to predict plant volatile phenotypes. For example, recent studies suggest that interactions between the JA and SA pathways do not always result in one pathway disrupting the other, but may involve more back-andforth communication or "cross talk." Besides, other phytohormones, such as ethylene and abscisic acid, play a significant role in defense signaling cascades acting synergistically or antagonistically with both JA and SA (Bostock, 2005;Dicke et al., 2009;Koornneef & Pieterse, 2008;Pieterse, Leon-Reyes, Van der Ent, & Van Wees, 2009;Stam et al., 2014).
These responses may be further modified by exposure to the HIPVs of damaged plant parts or nearby attacked neighbors, which "prime" undamaged plants or plant parts to respond more efficiently, and to a higher degree, to subsequent herbivore damage (Engelberth, Alborn, Schmelz, & Tumlinson, 2004;Heil & Kost, 2006;Heil & Silva Bueno, 2007;Ruther & Furstenau, 2005) (Figure 1). As an example of this phenomenon, corn seedlings exposed to green leaf volatiles (GLVs) from neighboring plants produced significantly more JA and volatile sesquiterpenes after mechanical damage in combination with caterpillar regurgitant than seedlings not exposed to GLVs, leading authors to hypothesize that priming may affect plant-plant and plant-insect interactions (Engelberth et al., 2004). Last but not least, trade-offs between direct and indirect defenses in combination with specific ecological settings can also result in unique "plant defense syndromes" involving differences in HIPV emission (Agrawal & Fishbein, 2006). Nevertheless, a critical factor determining the relative importance of HIPVs, and hence the tolerance to cue disruption, is the foraging behavior of the natural enemy. The foraging behavior is a complex process product of the co-evolution of prey and predator and is largely determined by the prey's behavior and defense mechanisms, as well as by the community characteristics such as diversity and complexity (Malcom, 2009;de Rijk, Dicke, & Poelman, 2013). In the case of herbivore's natural enemies, the foraging behavior will determine to which extent parasitoids and predators rely on other nonchemical cues (e.g., visual, acoustic, and vibrational signals) and on other sorts of chemical cues rather than HIPVs (e.g., habitat related cues, host-derived odors, and odors of conspecifics) to find their prey (Steidle & van Loon, 2003;Wäschke, Meiners & Rostas, 2013).
A recent theoretical study (Yoneya & Miki, 2015) suggests that co-evolution of foraging behavior in herbivores and natural enemies allows both groups of organisms to use HIPVs as multifunctional signals depending on the intensity of the attack. For example, a recent

| Multiple herbivory
In nature, most plants are exposed to numerous attackers, acting simultaneously or sequentially . Early studies on the effect of multiple herbivory on indirect defense focused on aboveground interactions, but recent work has brought to our attention that simultaneous above-and belowground attack can also have profound impacts on natural enemy recruitment (Bezemer &  In the case of specialists, the only available study reports disruption due to multiple attackers, yet how this relates to changes in HIPV emissions and whether disruption is common for other specialists remain unclear. An exhaustive study of 140 research papers on natural enemy attraction to infochemicals showed that there is no significant difference between specialist and generalist natural enemies in the proportion species that use volatiles during foraging; however, the ability to learn and display plastic responses to these compounds seems to be more common in generalist species (Steidle & van Loon, 2003). Additional studies suggest that generalists and specialists may differ in their use of volatile cues, with generalists relying on widespread damage-related compounds such as GLVs, while specialists utilize more precise volatile signatures associated with their preferred prey (Cortesero et al., 1997;Ngumbi, Chen, & Fadamiro, 2009.
However, whether differences in feeding specialization render one of these two groups more susceptible to signal disruption than the other remains to be investigated.
In simultaneous above-and belowground herbivory scenarios, the most common outcome is decreased natural enemy attraction (both above-and belowground), independently of the feeding guild of the natural enemy or the changes in total volatile emissions ( of particular compounds used as cues by natural enemies (Bezemer & van Dam, 2005;Soler et al., 2007).
Root herbivory is likely to be a major factor disrupting plant-natural enemy communication in nature, due to its significant negative impact on plant and herbivore communities (Blossey & Hunt-Joshi, 2003

| Presence of microorganisms
Plants are not only challenged by multiple herbivores but by beneficial microorganisms and pathogens, which can also elicit distinct signaling pathways. For example, biotrophic pathogens (those growing and feeding within the living cells of their hosts) typically elicit SA-mediated induced defenses. Necrotrophic pathogens (those killing its host cells and then feeding on the dead matter) often induce JA/ ethylene-mediated defenses (Glazebrook, 2005;Thomma, Penninckx, Broekaert, & Cammue, 2001), and interactions with beneficial microorganisms are generally mediated by the JA signaling pathway (Glazebrook, 2005). Although much is known about the molecular basis of plantpathogen interactions, few studies have explored the effect of herbivore attack in combination with microorganisms on plant volatile emission and its effects on natural enemy recruitment (Ponzio, Gols, Pieterse, & Dicke, 2013). Available studies involving beneficial and nonpathogenic microorganisms report multiple outcomes ( Table 2) Contrastingly, the few studies on pathogenic microorganisms show an increased attraction of natural enemies toward pathogeninfested plants (
Despite the expected negative effects, the available reports ( of drought and changes in CO 2 concentration. Disruption due to alterations in CO 2 levels and drought is comprehensible as carbon dioxide and water are crucial for primary metabolism, which in turn is the main energy provider for plant growth and development, as well as for the production of secondary metabolites involved in plant defense (Bolton, 2009;Lawlor & Cornic, 2002). However, as shown in the case of CO 2 , different plant genotypes (Sun, Feng, Gao, & Ge, 2011) Winter & Rostás, 2010). However, this is not always the case (Bezemer, Jones, & Knight, 1998;Stacey & Fellowes, 2002;Sun et al., 2011). For example, a study on the long-term effects of temperature on populations of the aphid Myzus persicae and its parasitoid Aphidius matricariae reported that elevated temperature decreased plant biomass while increasing leaf nitrogen concentrations, which in turn enhanced herbivore abundance and increased parasitism rates (Bezemer et al., 1998). Such studies evidence that bottom-up effects of abiotic stress are not always negative.
Another interesting aspect is that under controlled settings, plantnatural enemy communication can withstand disruption due to abiotic stress, yet when offered a choice, natural enemies would prefer "healthy" herbivore-induced plants to those under stress conditions (Olson, Cortesero, Rains, Potter, & Lewis, 2009 It is possible that effects of abiotic factors on natural enemy recruitment vary depending on the magnitude of the stress and its impacts on the plant metabolism, with severe stress having stronger effects due to constraints in resource availability and allocation affecting HIPV production and release. For example, existing studies show that mild drought increases HIPV emissions or has no effect, whereas severe drought decreases emissions (Becker et al., 2015;Lavoir et al., 2009;Peñuelas & Staudt, 2010). Moreover, responses may vary for individual plant species, as some plants have evolved unique adaptations to stress, and the presence or absence of stress-tolerance traits will determine the threshold levels for a particular species (Bray, 1997;Pareek, Sopory, Bohnert, & Govindjee, 2010;Wang, Vinocur, & Altman, 2003).
It is evident that individual abiotic factors affect HIPV emission, but there is much potential for interaction among them, leading to different outcomes from those caused by a single stress or those expected by additive effects (Becker et al., 2015;Bezemer et al., 1998;Peñuelas & Staudt, 2010). Studying these interactions among abiotic factors is necessary, especially in scenarios of global warming where multiple abiotic stress factors are likely to occur simultaneously.
The predicted impacts of climate change on natural enemies are severe and include, but are no restricted to: loss of fitness due to poor prey quality, lower susceptibility of herbivores to parasitism or predation due to changes in plant phenology and altered timing of herbivore life cycles, permanent loss of prey due to prey extinction or changes in plant and herbivore distribution, and increased competition with new natural enemies, due to changes in distribution ranges (Boullis et al., 2015;Hance, Van Baaren, Vernon, & Boivin, 2006;Thomson, Macfadyen, & Hoffmann, 2010). In agricultural systems, a number of additional effects may appear as a result of adaptive management strategies adopted by farmers to cope with climate change (Thomson et al., 2010). Whether disruption in plant-natural enemy communication needs to be incorporated to the list remains to be investigated.

| Combining biotic and abiotic factors: a new approach
Recently, two pioneer studies have brilliantly incorporated the effects of abiotic factors with above-and belowground organisms and their effects on the attraction of natural enemies (Johnson, Staley, McLeod, & Hartley, 2011;Tariq, Wright, Bruce, & Staley, 2013). The first study evaluated the effects of summer drought on plant community containing Hordeum vulgare (barley), Capsella bursa-pastoris (shepherd's purse), and Senecio vulgaris (common groundsel), in the presence of the earthworm Aporrectodea caliginosa, the aphid Rhopalosiphum padi and its parasitoid, Aphidius ervi (Johnson et al., 2011). Johnson and co-authors found that summer drought alone had a negative impact on plant shoot and root biomass, but the addition of earthworms significantly reduced root biomass loss. Drought also led to a significant decrease in aphid abundance, which was moderated by the presence of earthworms, and these effects reflected on parasitism rates. Interestingly, the effect of earthworms was much higher in one-plant species plots than in multiple species plots, suggesting that other community members can also have an impact on the outcome of tritrophic interactions.
The second study evaluated the effect of drought in a system comprising Brassica oleracea, the root herbivore Delia radicum, the aphids Myzus persicae and Brevicoryne brassicae, and the parasitoids Aphidius colemani and Diaeretiella rapae (Tariq et al., 2013). Their results showed that drought conditions and root herbivory separately had negative effects on parasitism rates. However, there was a significant interaction between drought and root herbivory, in which drought stress partially reversed the negative effect of root herbivory on parasitism rates.
These rare examples demonstrate that multiple biotic and abiotic factors interact, having a strong impact on plant-natural enemy communication. It is hoped that we will be seeing more such studies in the future, which are closer to the natural situation of plants under both cultivated and natural conditions. Similar studies could be useful to investigate plant-natural enemy communication in climate change scenarios.

| CONCLUSIONS AND OUTLOOK
To wrap up this review, I will answer the questions proposed in the introduction in light of the available literature. Due to the limited amount of available of literature, it is difficult to predict accurately which factors disrupt plant-natural enemy communication. Each system is unique and needs to be explored in the ecological context in which it occurs, including the interactions between multiple biotic and abiotic factors. However, the literature reviewed here suggests that belowground herbivory consistently disrupts natural enemy attraction, presumably due to the strong effects of root herbivory on nutrient uptake and plant metabolism that impact plant signaling and herbivore quality as a prey. More studies are required to support or reject this hypothesis.

3.
Are there common patterns allowing us to make predictions about the outcome of these tritrophic interactions under biotic and abiotic stress scenarios?
Although it may be tempting trying to predict the outcome of plantnatural enemy interactions by investigating only one the actors involved, this is often insufficient and pays no heed to ecological complexity. A more systemic approach is needed to understand the stability and direction of these interactions in nature, and under biotic and abiotic stress.
There is a common thread in the existing reports, suggesting that natural enemies can infer host quality based on volatile cues. Hence, the bottom-up effects (both positive and negative) of biotic and abiotic factors on plant quality for the herbivore, and of this as host for the natural enemies, are likely to play an important role determining the outcome of the interaction. Therefore, investigating these bottom-up effects is crucial for further studies aiming to understand the impact of biotic and abiotic factors on plant-natural enemy interactions.
Research on multitrophic interactions has slowly progressed from evaluating linear plant-herbivore-natural enemy systems under controlled conditions into more complex models incorporating multiple attackers and abiotic conditions. However, even at this level, there is a high risk of oversimplification, as both biotic and abiotic factors are likely to interact in complex ways, rather than just having additive effects.
Critical aspects for future research to understand the stability of plant-natural enemy interactions in nature include the effects of biotic and abiotic stress on natural enemy foraging behavior, the impact of the stress intensity on volatile emission and natural enemy recruitment, and the complex role of microorganisms on plant-natural enemy interactions. The ultimate goal is to establish the impact of multiple co-occurring biotic and abiotic factors that recreate natural and climate change scenarios, and the identification and exploration of newly emerged and threatened interactions as a result of climate change.

ACKNOWLEDGMENTS
I would like to thank the members of the group Biocommunication and