Interplant signalling through hyphal networks




Mycorrhizal fungi can form common mycelial networks (CMNs) that interconnect plants. Here, we provide an insight into recent findings demonstrating that CMNs can be conduits for interplant signalling, influencing defence against insect herbivores and foliar necrotrophic fungi. A likely mechanism is direct transfer of signalling molecules within hyphae. However, electrical signals, which can be induced by wounding, may also enable signalling over relatively long distances, because the biophysical constraints imposed by liquid transport in hyphae and interaction with soil are relieved. We do not yet understand the ecological, evolutionary and agronomic implications of interplant signalling via CMNs. Identifying the mechanism of interplant signalling will help to address these gaps.

I. Introduction

Most soils contain vast lengths of fungal hyphae (Leake et al., 2004), much of which is assumed to comprise extraradical mycorrhizal mycelium. The length of hyphae in soil and the ability of mycorrhizal fungi to form multiple points of entry into roots can lead to the formation of a common mycelial (or mycorrhizal) network (CMN) that interconnects two or more plants. Whilst some aspects of CMN function remain equivocal and difficult to test, the consensus is that CMNs are crucial components of ecosystems. CMNs can lead to ‘network-enhanced bioactive zones’ (Barto et al., 2012), and have key roles in facilitating nutrient transport and redistribution (Selosse et al., 2006). Recently, the potential ecological importance of CMNs formed by arbuscular mycorrhizal (AM) fungi has been extended by a series of experiments demonstrating that CMNs facilitate defence against insect herbivores (Babikova et al., 2013a,c; Song et al., 2014) and foliar necrotrophic fungi (Song et al., 2010) by acting as the conduits for interplant signalling (Fig. 1). Here, we provide insight into the mechanisms, evolutionary consequences and circumstances under which interplant signalling via AM fungi could have evolved, and highlight key gaps in our understanding. We use the term ‘signalling’ throughout this paper, but draw attention to current discussions concerning the use of similar terms, including communication (e.g. Scott-Phillips, 2008).

Figure 1.

Interplant signalling. Donor plants infested with phloem-sucking insects such as aphids, chewing insects such as caterpillars and necrotrophic fungi induce production and release of volatile organic compounds (VOCs) from plant tissues. Neighbouring uninfested receiver plants may respond to chemical cues transmitted aerially or via a common mycelial network (CMN). Interplant signalling via CMNs may involve active chemical transport, passive transport on surface liquid films, and induction of ‘action potentials’ (electrical signals) as a result of membrane depolarization. Receiver plants induce jasmonic acid and salicylic acid biosynthesis pathways and release VOCs to repel aphids and attract their natural enemies, such as parasitoids.

II. Evidence that CMNs facilitate interplant signalling

The idea that plants might communicate with each other through production of molecules released into the atmosphere is not new (e.g. Baldwin & Schultz, 1983; Rhoades, 1983). Subsequent experiments demonstrated unequivocally aerial inter- and intraplant signalling of volatile organic compounds (VOCs) induced by herbivory (often known as herbivore-induced plant volatiles; HIPVs), which has led to discussion regarding the potential evolutionary consequences of this process (Heil & Karban, 2009). VOCs produced by plants infested with insects such as aphids are often produced systemically and can be released into the rhizosphere (Chamberlain et al., 2001), which led to the hypothesis that mycorrhizal fungi may facilitate transfer to neighbours (Dicke & Dijkman, 2001). Babikova et al. (2013a) were the first to demonstrate that CMNs are instrumental in up-regulation of HIPVs in undamaged neighbours. Aphids were added to ‘donor’ plants, which led to the release of methyl salicylate from their leaves that were repellent to aphids but attractive to parasitoid wasps, a natural enemy of aphids. If uninfested neighbouring ‘receiver’ plants were connected to the donor by a CMN, they behaved exactly like the donor plant by producing VOCs that had identical effects on aphid and parasitoid behaviour. By contrast, if uninfested neighbouring ‘receiver’ plants were not connected to the donor by a CMN, they did not produce aphid-repelling and parasitoid-attracting VOCs.

Subsequent work (Song et al., 2014) has similarly demonstrated that CMNs can be signalling pathways between plants under herbivore attack, using tomato plants infested with the caterpillar Spodoptera litura. In addition, they determined that the jasmonic signalling pathway is implicated in the response to the interplant signalling by using mutant receiver plants (effectively blocking jasmonic acid production and aspects of the jasmonic acid signalling pathway) compared with wildtype receivers. S. litura gained significantly less weight when feeding on the wild-type receiver plants compared with two types of mutant receiver which also, importantly, showed that the interplant signalling has negative fitness consequences on the herbivore.

It is not only insect herbivores that induce defences in neighbouring, uninfested plants. The first demonstration of interplant signalling via CMNs showed that the foliar necrotrophic fungus Alternaria solani colonising tomato plants led to up-regulation of six defence genes in uninfested donors connected via a CMN (Song et al., 2010). The expression of these genes provides a link to the salicylic acid and jasmonic acid pathways, suggesting they are also implicated in defence against necrotrophic fungi as well as insects.

It has been suggested that the process of forming mycorrhizas itself primes plant defences against multiple biotic challenges (so-called ‘mycorrhiza-induced resistance’; Cameron et al., 2013); however, the effects seen in recent work on interplant signalling are in addition to this process because the key treatment comparisons in these experiments all used plants in the mycorrhizal condition, which is a crucial requirement of their experimental designs. Nevertheless, there are potentially important methodological issues that should be highlighted. For example, Song et al. (2010) did not provide complete control for molecules released into the soil from hyphae of donors that are taken up again by hyphae of receivers (hyphae–soil–hyphae pathway); they used either waterproof membranes between mycorrhizal donors and receivers that completely prevent diffusion, or nonmycorrhizal donors and receivers. The development of extraradical mycelium can increase the surface area of absorptive tissue compared with nonmycorrhizal roots, so nonmycorrhizal plants are poor controls for this process. Indeed, the lack of consideration of the hyphae–soil–hyphae pathway is established (Robinson & Fitter, 1999) as confounding the interpretation of interplant transfers of molecules via CMNs. These issues can be easily overcome by using rotated and static mesh core systems that control development of CMNs but allow for diffusion (Johnson et al., 2001; Babikova et al., 2013a).

III. Potential mechanisms of interplant signalling

There are three possible mechanisms by which CMNs enable interplant signalling: transport of molecules in liquid films on the external surface of hyphae via capillary action or microbes; delivery of signal molecules via cytoplasmic streaming within hyphae; and conduits for wound-induced electrical signals (Fig. 1).

Liquid film transport on surfaces is unlikely over relatively large distances, where close interaction of fungal hyphae with soil particles will restrict its efficacy, and so the second mechanism is more likely to explain the recent observations of signalling via CMNs. Regardless of the transport mechanism, the obvious question arises: which molecules are involved? Answering this question may be addressed by focusing on molecules already known to be transported in AM fungal hyphae (e.g. lipids such as triacylglycerols (Bago et al., 2002), phosphate transporters and amino acids (Jin et al., 2005)) or on compounds known to elicit, or be produced by, established plant–mycorrhizal fungus signalling pathways (Nadal & Paszkowski, 2013). Also, we do not yet know whether signalling compounds are specific to particular species of herbivore or foliar fungi or even to types of damage (e.g. sucking, chewing).

Electrical signals, a result of membrane depolarization, can be produced by plants in response to both artificial mechanical damage (Mousavi et al., 2013) and chewing insects (Salvador-Recatala et al., 2014). These events are often associated with production of cytosolic Ca2+, are propagated by glutamate receptor-like genes, and can induce proteinase inhibitors, a well-known defence response in plants (Wildon et al., 1992). A key advantage of electrical-induced defence over a mobile chemical is the speed of delivery and relief of biophysical constraints involved with phloem transport (such as resistivity of the vessel to flow). Could systemically produced electrical potentials be implicated in interplant signalling via CMNs? There is certainly evidence that membrane depolarization events lead to action potentials that affect fungal physiology and activity, such as orientation (Brand & Gow, 2009), spatial control of nutrient uptake and intrahyphal signalling, and changes in the electrical potential of leek Allium ampeloprasum roots have been measured in response to AM fungi (Ayling et al., 1997).

IV. Evolutionary drivers of interplant signalling

There are many fascinating evolutionary implications of plant-to-plant signalling via aerial pathways (Heil & Karban, 2009). The recent discoveries of interplant signalling via CMNs raises additional questions, in terms of both the drivers and circumstances that led to the phenomenon and the fitness consequences of it to both plants and fungi. Greater understanding is required concerning the costs and benefits of emitting, receiving and transporting the signal/s. Quantifying the costs and benefits are challenging and, to date, most attempts have been unsuccessful; identification of the mechanisms involved in interplant signalling is vital to achieve this.

Babikova et al. (2013d) took a mycocentric view and hypothesized that, assuming the fungus has some control over the strength and direction of the signal, it will act to protect the net carbon source it derives from the plants to which it is connected. Thus, they speculated that the fungus would benefit most from preferentially allocating signal to plants that provide the most carbon, produce the most VOCs, or to plants that are not connected primarily to other competing fungi. To test any of these ideas, we need to identify and quantify the signal, as well as determine whether the fungus can control its strength and delivery point.

It is interesting to speculate further about particular circumstances that could affect the evolution of interplant signalling. For example, in very stable environments where CMNs can persist (such as extensively managed grassland), there may be a bigger cost to the fungus of losing a particular host to herbivory or disease, and therefore a greater benefit to transferring warning signals compared with unstable environments that have fast turnover of CMNs.

One further question is whether the degree of specialization of the herbivore–plant–fungal interaction would affect the evolution of the CMN-based signal transfer between plants. If a plant is under attack from a particular aphid species and a generic herbivore signal is sent through the CMN, receivers cannot detect if the attacking aphid is relevant to them or not. If the receiver is the same species of plant, it would benefit from responding (assuming it could also be under attack from the same aphid species), but if the receiver is a different species (not a host to the attacking aphid), its response may have a large cost and a small benefit to the receiver, fungus and emitter. Therefore, it can be hypothesized that in a mixed-species environment, and where a large proportion of the herbivore–plant–fungal interactions are species-specific, there should be selective pressure for specific signals to evolve. However, in single-species environments or those where most insect herbivores are generalists, all parties should benefit from a generic signal (Fig. 2). Yet nothing is known of the specificity of signalling pathways through CMNs. Plant responses to herbivores tend not to be related to the degree of specificity of herbivorous insects, but do vary with the broad taxa of the herbivore (e.g. Bidart-Bouzat & Kliebenstein, 2011), and this may act as a constraint to the evolution of very specific CMN-based signals (or a reflection that there is no strong selection on specificity). For example, there may be no additional benefit of transporting a signal specific to the species of attacker if the receiving plant responds only to broad taxa.

Figure 2.

Hypothetical relationships between plant species richness and the probability of any one receiver plant benefiting from responding to a signal (e.g. emitting herbivore induced plant volatiles) via mycorrhizal fungal networks from an infested donor plant. The line colours represent the degree of herbivore–plant specificity: the herbivore feeds on only one species (blue); five species (red); or 10 species (green). There is stronger selective pressure to evolve interplant signalling (and response) in species-poor environments, but if the herbivore is a generalist, interplant signalling may still evolve in a more species-rich environment (although the chance of benefit still declines with species richness). The curves are derived by dividing the degree of insect–plant specificity by the number of species present in a community. The model assumes equal mixing of all plant species in the common mycelial network (CMN) and that the signal is not pest species-specific.

Understanding the cost and benefits of interplant signalling must also be placed in a broader framework with the recognition that, for example, plants often undergo challenges from several herbivores or pathogens simultaneously and this can lead to differences in defence responses depending on whether the attacker is above or below ground (Kutyniok & Muller, 2012). Moreover, cost–benefit assessments should ultimately consider the full suite of potential benefits of the establishment of CMNs (i.e. nutritional, disease resistance, herbivore defence).

V. Consequences of interplant signal transfer for ecosystems

How effective is interplant signalling in reducing the effects of insect herbivores on mycorrhizal plants and fungi and how far-reaching are these effects through natural communities? Answering this question requires experiments testing impacts on fitness-related traits of the pests, plants and fungi, as well as field studies (all the studies undertaken on CMN-based signalling, and also most on nutrient transfers, have been undertaken under highly simplified laboratory conditions). One of the consistent outcomes of work undertaken to date is the rapidity of signal transfer with the maximum response by receiver plants occurring just 50 h after donors are attacked or infested (Fig. 3). This analysis suggests a decline in strength of the response through time, but this effect is driven by just one point and needs to be confirmed by additional analyses. We can also speculate on the distance over which signals may have an effect based on previous work. Barto et al. (2011) showed that a CMN could transport allelopathic compounds at least 12 cm, while Song et al. (2010) and Babikova et al. (2013a) measured effects over 15 and 20 cm, respectively. A zone of influence c. 20 cm from plant roots has the potential to suppress the density of herbivores at local scales and it could be possible for signals to be delivered through cohorts of CMNs, so the zone of influence becomes even greater (Babikova et al., 2013d). The signalling mechanisms may also be important, with electrical signals offering the potential to travel further than transport of soluble molecules.

Figure 3.

The speed of responses of ‘receiver’ plants connected via a common mycelial network (CMN) to ‘donor’ plants infested with one of three pests: aphids (Babikova et al., 2013a,c), necrotrophic fungus (Song et al., 2010) and caterpillar (Song et al., 2014). The curve is a best fit of the response times (= −0.0135x2 + 2.367x + 0.47; < 0.001; R2 = 0.914), and the relative responses are calculated from Babikova et al. (2013) by comparing aphid attractiveness of receiver plants either connected via a CMN or unconnected to donor plants infested with aphids; and by comparing the expression of the gene allene oxide cyclase in receiver plants connected to donors infested with either a necrotrophic fungus (Song et al., 2010) or a caterpillar (Song et al., 2014) with expression in receiver plants connected to healthy donors (treatments A and D in their papers).

To determine the cascading effects of CMN-based interplant signalling through the wider ecosystem, we also need to quantify the importance of CMN-based signalling relative to aerial signalling. A recent field-based study suggests that aerial signalling could have effects that are ecologically meaningful. Karban et al. (2013) prevented CMN-based signals by keeping donor plants in pots, and found that plants with neighbours that had been clipped suffered less herbivory than those that were near individuals that remained unclipped. Strikingly, they also found that these effects of aerial signalling were greater when the aerial cues were from kin. Karban et al. (2012) also found that signalling between sagebrush plants in response to clipping benefited fitness-related traits such as the number of inflorescences.

Interplant signalling via CMNs offers a potentially more effective and directed pathway of defence compared with the aerial pathway, because the signals may be able to travel greater distances, limited only by the length of CMNs, which for some species of fungi can be vast, and may be targeted to particular plants and are unaffected by dilution effects caused by wind. One of the key priorities is to determine if, and to what extent, the phenomenon impacts fitness-related traits in the field. These sorts of experiments may be critical for developing effective sustainable solutions to insect pests in agricultural situations (Pickett et al., 2014), although often there can be little overlap between fitness-related and desirable agricultural traits (Denison, 2012).

The role of CMNs in facilitating interplant signalling of defence against insect herbivores needs to be considered alongside wider effects of integration of plants into an active fungal network. For example, the process of becoming colonized by mycorrhizal fungi itself can have major effects on plant–insect herbivore (e.g. aphids) interactions. Often plants colonized by AM fungi are more attractive to aphids (Babikova et al., 2013b) and increase aphid fecundity (Gange et al., 1999), although these responses are thought not to be related to the effects of mycorrhizal fungi on leaf phosphorus concentrations (Babikova et al., 2014).

VI. New questions and how to address them

Interplant signalling of defence-related signals by mycorrhizal fungi can now be regarded as one of several mechanisms by which CMNs influence both the plants and fungi that form them. The findings suggest that multitrophic interactions, particularly among above- and below-ground organisms, occur over greater spatial scales than was previously considered (Wardle et al., 2004), and offer new lines of inquiry towards the development of sustainable controls of pests such as aphids (Babikova et al., 2013d). Key priority questions for future research include the following: what is the mechanism of signal transfer through the fungal mycelium; how ubiquitous is interplant signalling via CMNs in nature (including e.g. in ericoid and ectomycorrhizal systems); and what are the wider eco-evolutionary impacts of interplant signalling for both mycorrhizal plants and fungi, especially in terms of fitness, in the field? Answering these questions will require ecologists to adopt transdisciplinary approaches, including combinations of behavioural, genetic and sophisticated analytical chemistry techniques.


We thank Martin Heil, two anonymous referees and Marc-André Selosse for their comments. L.G. was supported by the Scottish Government's Rural and Environment Science and Analytical Services Division (RESAS).