Microbial mediation of plant competition and community structure


Correspondence author. E-mail: angela.hodge@york.ac.uk


  1. The drivers behind plant competition and diversity have long been debated, and there is acceptance that soil micro-organisms may act as key drivers in plant interactions and community structure. The evidence for a microbial role in shaping plant interactions and communities will be considered here with emphasis on symbionts and pathogens.
  2. Microbial populations are themselves strongly influenced by the plant community via the ‘rhizosphere’ effect. The rhizosphere community includes microbes both beneficial and detrimental to the plant. Both the ability of plants to cultivate different rhizosphere microbial populations and the resulting impact upon other plant species have largely been studied via ‘plant–soil’ feedback studies, a proxy necessitated by the fact the majority of soil micro-organisms are unculturable, but which nevertheless has rarely been used in conjunction with modern techniques to identify and quantify the micro-organisms involved.
  3. Both microbial symbionts and pathogens can affect plant diversity and productivity, but the direct evidence for impacts on competitive interactions is surprisingly scarce. Evidence comes from biological invasions, the unintentional introduction of microbial pathogens to native plant communities and manipulative experiments under both field and controlled conditions. Pathogens generally have direct effects on plants, reducing their growth and so rendering them less competitive, whereas other symbionts may act by altering the availability of resources, with more subtle effects on competitive interactions.
  4. Some of the best evidence for indirect effects comes from studies on arbuscular mycorrhizal (AM) fungi. New developments in our understanding of nutrient exchange in the arbuscular mycorrhizal symbiosis emphasize the need to view the fungal partners not as mere extensions of the plant. The suggestion that AM fungi may act to share resources among connected plants however remains to be proved.
  5. Although plant competitive interactions can be driven by key microbial groups including symbionts and pathogens, knowledge gaps in the basic biology of these micro-organisms has hindered a full mechanistic understanding of these processes. If ecologists now embrace new technologies, significant advances in this area should be forthcoming.


Much of the debate as to what drives plant diversity–productivity relationships has focused upon how competition for resources may influence relative abundance. If resource availability does control plant interactions, plants should respond to the distribution of these resources, including those that are soil-based. Certainly, plant roots do respond, sometimes spectacularly so, to nutrient heterogeneity in the soil by morphological and physiological plasticity (reviewed by Hodge 2004; Cahill & McNickle 2011). Moreover, the ability to proliferate roots within the nutrient-rich zones or patches can confer a competitive advantage (Hodge et al. 1999, 2000a). However, nutrient availability in soil is closely tied to microbial activity. In the complex organic patches that occur in soil, plant roots rely on microbial decomposition processes for nutrient release (Hodge et al. 2000b). Some plant species can acquire amino acids (Näsholm et al. 1998; Inselsbacher & Näsholm 2012) or even short peptides (Hill et al. 2011) produced during the early stages of decomposition without the requirement for further mineralization into inorganic nitrogen (N), although competition between plants and microbes will occur for these N forms also (see Hodge, Robinson & Fitter 2000c).

At greater scales, heterogeneity in abiotic (i.e. physical or chemical) properties of soils influences diversity (Fitter 1982) and vegetation patterns (see Tilman 1982). Roots of coexisting plant species may avoid each other – and hence, competing with each other – through root segregation by occupying different spatial locations, or niches, in the soil profile (Berendse 1981; Veresoglou & Fitter 1984; Schenk, Callaway & Mahall 1999). In addition to permitting coexistence, this can lead to an increase in biomass compared to monocultures (‘over-yielding’), although Mommer et al. (2010) found that overyielding in root biomass (measured using real-time PCR) was due to enhanced growth of one species in the mixture, the grass Anthoxanthum odoratum, rather than any segregation. Plants can also alter their ability to acquire nutrients at depth depending on the other plant species present. Jumpponen et al. (2002) found that when Achillea millefolium was grown in monoculture, it acquired more N from 5 cm compared to 20 cm depth but not when grown with Festuca ovina, and a similar displacement of preferred depth of uptake was shown by Veresoglou & Fitter (1984). Competitively dominant plants may use the most abundant forms of nitrogen in soil (McKane et al. 2002) especially when grown with competitively inferior species (Ashton et al. 2010) that may themselves rely on less abundant N forms (McKane et al. 2002), thus enabling coexistence.

These studies demonstrate that the differing responses of plant species to neighbours and to abiotic factors can influence coexistence and enhance plant diversity. However, interactions between soil organisms and plants may be even more important in determining the outcome of plant competitive interactions and plant diversity (e.g. Lankau et al. 2011; Schnitzer et al. 2011). Although the main emphasis of this review is on microbial influences on plant competitive interactions in the widest sense (including pathogens), trying to disentangle cause and effect of specific microbes (or microbial groups) on plant communities is fraught with difficulties particularly under field conditions given the large number of possible confounding variables. Further complications arise in that some microbial impacts on plant communities may be density-dependent: for example, bacterial quorum sensing can lead to enhanced enzyme activity and altered nutrient cycling (DeAngelis, Lindow & Firestone 2008), which in turn will likely impact upon plant competitive interactions. However, while particular microbial groups are often reported to be the cause of particular plant competitive outcomes, it is rare that these microbial groups have been isolated, identified or quantified. In consequence, the role of microbes in plant competition is generally inferred from circumstantial evidence rather than empirically demonstrated, especially where soil bacteria are involved. In contrast, studies on symbiotic microbes, such as arbuscular mycorrhizal (AM) fungi, rhizobia and some pathogens, have in some cases involved manipulation experiments (albeit with a relatively low number of bacterial or fungal species), from which firmer conclusions about impacts upon plant competition and community structure are possible.

Plant roots influence the microbial community – the rhizosphere

Plants influence microbial populations in the vicinity of their roots, the rhizosphere, through the release of various substances, collectively termed ‘rhizodeposition’ (reviewed by Jones, Hodge & Kuzyakov 2004). Different plant genotypes can modify the resulting rhizosphere microbial community in divergent ways, presumably because of differences in the quality or quantity of substances released (Rengel 2002; Aira et al. 2010). These substances (or ‘rhizodeposits’) range from chemically simple compounds through to the dying root itself. Most studies have focused upon how exudates, including simple sugars, amino acids and organic acids passively lost from the root, affect the activity or function of the entire microbial community (Hodge et al. 1998; Shi et al. 2011), an approach necessitated by the fact that most of the micro-organisms present have proved to be unculturable. New technologies are, however, beginning to unlock the mysteries of the rhizosphere community ‘black-box’ (e.g. DeAngelis et al. 2009; Uroz et al. 2010). However, exudation itself is a moveable feast and is influenced by a large number of environmental conditions (Hodge & Millard 1998; Paterson et al. 1999), as are rhizodeposition processes in general (Hinsinger et al. 2009), which will likely impact upon microbial community structure dynamics.

Micro-organisms in the rhizosphere can have beneficial, deleterious or even pathogenic impacts upon the plant. While plant roots release signalling compounds that encourage formation of mutualistic symbiotic associations with N-fixing prokaryotes and mycorrhizal fungi, other free-living micro-organisms such as plant-growth-promoting rhizobacteria (PGPR) also apparently benefit the plant in a range of ways including enhanced nutrient acquisition and production of hormones (Richardson et al. 2009). In addition, fungal endophytes present in the roots and shoots of some plants enable the plant to cope with abiotic stress conditions including high temperature, salinity and drought (reviewed by Rodriguez et al. 2009). Plants also actively secrete (as opposed to passively exude) compounds into their rhizosphere, which serve a range of functions including the acquisition of particular nutrients. These compounds differ among plant species resulting in differences in rhizosphere microbial communities (reviewed by Hartmann et al. 2009), and plant species themselves also differ in their response to particular microbes. Complex interactions among microbial groups (Leigh, Fitter & Hodge 2011; Herman et al. 2012) therefore influence nutrient cycling and so affect the plant community. These interactions have important consequences for plant competitive interactions and coexistence, but they are experimentally challenging and have largely been investigated using so-called ‘plant–soil feedback’ studies.

Plant–soil feedback studies

According to Bever (1994), plant–soil feedback involves two stages: (i) a plant (or a plant population) alters the soil microbial community composition, and (ii) this in turn increases (positive feedback) or decreases (negative feedback) the plant population growth rate. In many experimental studies, comparisons are often made between plants of one species grown either in soil it (or a conspecific) previously occupied or in soil occupied by another plant species, and thus of differing microbial community (Bever 1994; Klironomos 2002; Batten, Scow & Espeland 2008), although the microbial differences are usually assumed rather than measured. Such experimental systems are also referred to as ‘home vs. away’ (Bever 1994) or ‘home vs. foreign’ (Klironomos 2002) comparisons.

Although there are a few reports of positive plant–soil feedbacks particularly from investigations into plant invasions (Hawkes et al. 2005; Callaway et al. 2004; but see Andonian et al. 2011), the majority of studies suggest negative feedback occurs, that is, plants are more constrained by the organisms present in their own rhizosphere than that of others, as confirmed by Kulmatiski et al.'s (2008) meta-analysis. Plant–soil feedbacks have been suggested to play a key role in various factors that drive plant community structure including influencing spatial structure (Packer & Clay 2000), succession/temporal dynamics (Van der Putten, Van Dijk & Peters 1993; Kardol, Bezemer & Van der Putten 2006), species composition and diversity (Bever, Westover & Antonovics 1997; Kulmatiski et al. 2008) and plant invasions (Klironomos 2002; Batten, Scow & Espeland 2008).

The interpretation of these experiments is not always straightforward: plant roots may alter chemical and physical properties (including nutrient availability) of the soil as well as the microbial community, and so the soils being compared may be different. Changes in physical and chemical characteristics of the soil will, in turn, influence the microbial community, leading to an additional feedback loop. Any attempt to disentangle these chemical and physical influences from biological, and specifically microbial, differences will be problematic (reviewed by Ehrenfeld, Ravit & Elgersma 2005). Despite this, plant–soil feedback effects are usually attributed to differences in microbial communities, including specific microbial groups, even when these have not been quantified (e.g. Bever 1994; Maron et al. 2011).

In addition to the issues in trying to disentangle the importance of abiotic factors and the subsequent direct vs. indirect effects of these on the microbial community, there are a number of other important considerations when comparing plant–soil feedback studies. Both the experimental design (number of plants grown, duration, location, field vs. glasshouse conditions) and the approaches used for data analysis can influence the ecological interpretation of the observed effect (see Kulmatiski et al. 2008; Brinkman et al. 2010). Moreover, it is generally assumed that plants rapidly cultivate their rhizosphere microbial populations and that this process is dynamic. However, while this may be the case following soil homogenization and sterilization, which is frequently carried out in such studies under controlled conditions, in field soils with established microbial communities, there may be legacy effects from previous plants that can persist for considerable time depending on the soil type (Kulmatiski & Beard 2011). Thus, while plant–soil feedback studies can be eloquent and relatively simple to perform, their interpretation requires caution particularly when artificial conditions are used.

Most feedback studies are undertaken in pots rather than in the field, but Casper & Castelli (2007) planted seedlings of three grass species into field soils adjacent to mature clumps of the same species, in a full factorial design. They also protected half the seedlings from competition by growing them in tubes of soil to which competitor roots had no access and by holding back the shoots of the mature grass clumps. All three grasses grew best in soil adjacent to a mature plant of another species, but for Schizachyrium scoparium and Sorghastrum nutans, this was only the case when they were protected from competition. Andropogon gerardii seemed to be unaffected by competition. There was, therefore, clear evidence for negative plant–soil feedback in a field setting, but the only measurement of soil microbes was of root colonization by AM fungi: in all three species, colonization was greater in ‘home’ soil than in ‘away’ soil. The authors concluded that ‘feedback and competition cannot be viewed as strictly separate forces’, implying that they are likely to interact. However, as with most feedback studies, information on mechanisms is lacking.

Microbial pathogens and plant communities

Because plant–soil feedback seems to be most frequently negative, it may be that pathogenic impacts are widespread and important. There is evidence to support that view from agriculture: yields often decline if a single crop species is grown continuously, an effect usually attributed to a build-up in the rhizosphere of harmful micro-organisms, including pathogens (e.g. Bakker & Schippers 1987; Alström 1992; Olsson & Gerhardson 1992; Cook 2003). Crop rotation is often an effective countermeasure: the best studied example is the control of the ascomycete fungus, Gaeumannomyces graminis var. tritici, which causes ‘take-all’ in wheat, but which can also infect other cereals including barley and rye (Cook 2003). The large areas under crop monoculture typical of many agricultural systems strongly encourage the development of the pathogen population, but pathogens can also have large impacts on native plant communities, an effect best seen where pathogens have been introduced outside their native range. For example, the introduction of Phytophthora cinnamomi to Australia devastated the native Eucalyptus spp. (Weste & Ashton 1994), and other well-known examples include chestnut blight, a canker disease caused by the ascomycete Cryphonectria parasitica, and Phytophthora ramorum, responsible for ‘sudden oak death’ but which can kill a number of different tree species. Demonstrating the role of native pathogens in natural plant communities is more difficult. Microbial pathogens that infect a particular plant species may either prevent it becoming dominant, so promoting diversity in the community, or lead to its extinction, decreasing diversity.

Microbial pathogen–plant interactions have been a major focus of plant–soil feedback studies (e.g. Van der Putten & Peters 1997; Maron et al. 2011). In Dutch coastal sand dunes, Ammophila arenaria frequently loses vigour and is replaced by Festuca rubra. Van der Putten & Peters (1997) demonstrated that when A. arenaria and F. rubra were grown together in soil originally from A. arenaria, F. rubra outcompeted A. arenaria. This replacement was further enhanced by nutrient limitation and was attributed to a build-up of pathogens specific to A. arenaria, which further reduced A. arenaria plasticity in response to low nutrient availability (Van der Putten & Peters 1997). Working with this system of creating plant assemblages, Maron et al. (2011) recently observed that fungicide application increased plant biomass production, but the effect was unpredictable and unrelated to the diversity of the assemblage so that the positive diversity–productivity relationship observed in the controls disappeared under fungicide application. Addition of fungicides can modify soil N dynamics (Chen, Edwards & Subler 2001), but in the Maron et al. (2011) study, there was no difference between soil inorganic N and total microbial biomass N measured in the middle of the growing season and plant-available N across the diversity gradient. Mycorrhizal colonization can influence plant diversity (see section below), but was not the key driver in this experimental system: there was no effect of colonization extent on plant diversity. Maron et al. (2011) therefore suggested that fungal pathogens were responsible for driving the plant diversity–productivity relationship, a conclusion given some support by the finding that fungal infection of roots was low in both fungicide treatments and high-diversity mixtures. However, the evidence from this study is circumstantial because no pathogens were measured. A more detailed study was that by Schnitzer et al. (2011), who also observed that plant disease decreased with increasing plant diversity. In this experiment, artificial soil communities were created by adding back to sterilized soil either AM fungal spores or a ‘pathogen/parasite/saprobe fraction’, both isolated from the field soil, in a factorial design. Measurements of root colonization by AM fungi and other microbes confirmed that the treatments had altered the microbial community as intended, although no microbes were identified. Nevertheless, their results support the view that fungal pathogens affect relationships between productivity and plant diversity (reviewed by Mordecai 2011). Interactions with bacteria antagonistic to plant pathogens are likely also important, which in turn are linked to both the diversity and functional composition of the plant community (Latz et al. 2012), and which warrant further investigation.

Nitrogen fixation

Weathering of rocks produces negligible amounts of N, and thus, N supply is a critical factor in developing plant communities and ecosystems. Most N in soils comes from biological fixation conducted by prokaryotes that may be free-living, loosely associated with plant roots (associative) or in mutualistic symbiotic relationships. Consequently, many N-fixing plants are pioneer species on N-deficient soils. In their classic study at Glacier Bay, Alaska, Croker & Major (1955) proposed that N-fixation by alders (an actinorhizal species) facilitated spruce (Picea) invasion by creating a more favourable environment including accumulation of soil N. However, subsequent work (Chapin et al. 1994; Fastie 1995) demonstrated that other factors are also important, including competition for soil resources and, importantly, plant dispersal. Moreover, chronosequence-based (i.e. ‘space for time’) methods as employed by Croker & Major (1955) and widely used in ecological research need to be treated with caution because concomitant changes in environment (e.g. climate) and plant invasion potential (e.g. rates of spread of potential community members) mean that not all sites in a chronosequence have experienced the same history (Johnson & Miyanishi 2008). Nevertheless, it is commonly observed that soil N content increases sharply during primary succession when N-fixing plants become abundant, and N-fixing symbionts influence plant productivity, community structure and diversity (van der Heijden et al. 2006).

Both leguminous (Karpenstein-Machan & Stuelpnagel 2000) and actinorhizal (Hagen & Jose 2011) plants may increase the proportion of N they derive from their N-fixing symbionts when growing in competition with non-fixing plants. In both these cases, N-fixation rates were not measured directly but rather inferred, either from a reduction in 15N uptake from a fertilizer source (Hagen & Jose 2011) or from the N content of plant material (Karpenstein-Machan & Stuelpnagel 2000). In addition, the competition treatment was not randomly applied in the study by Hagen & Jose (2011), leading to issues of pseudoreplication. If N-fixing plants rely on their symbionts more when in competition, there may be potential for niche differentiation. For example, plant species grown in six-species mixtures including legumes mainly acquired the same N source (NO3) from the same location (shallow soil) rather than displaying spatial niche partitioning, which was attributed to the legume plants increasing availability of soil N by relying on N-fixation instead (von Felten et al. 2009). Similarly, Temperton et al. (2007) found that the N content and concentration of individual plants (‘phytometers’) planted into the plots of a biodiversity experiment were a function of the proportion of legumes in the community, which they interpreted (from δ15N data) as reflecting reduced competition for soil nitrate.

In laboratory conditions, N may also be transferred from legumes to non-legumes via mycorrhizal hyphae (Haystead, Malajczuk & Grove 1988), but the ecological relevance of this transfer is uncertain. Wagg et al. (2011) studied competitive interactions between Lolium multiflorum and Trifolium pratense in low- and high-diversity arbuscular mycorrhizal fungi (AMF) treatments (containing 1 or 4 AMF). The higher AMF diversity generally enhanced T. pratense biomass, and this also enabled coexistence with Lmuliflorum. However, the impact of individual AMF on plant biomass varied with soil type, suggesting that interactions among soil microbes and the soil environment will determine outcomes of plant competition.

Evidence from the oldest ecological experiment in existence (namely the ‘Park Grass’ experiment at Rothamsted Experimental Research Station, UK, set-up in 1856) demonstrates how radically species composition has altered under different fertilizer treatments. Grasses dominate in the fertilized treatments when N is also included, whereas legume species increase in abundance when nitrogen is omitted from the added fertilizer due to their N-fixing symbionts. Thus, intermediate situations favour coexistence particularly if there is spatial heterogeneity in N/phosphorus (P) ratios (Silvertown et al. 2006). As Park Grass demonstrates, inputs of available N are usually disruptive to a plant community. Unsurprisingly, therefore, the consequences for native plant communities when plants that form N-fixing associations are introduced can be serious. One classic example of this is the invasion by Myrica faya (an actinorhizal species) into young volcanic soils in Hawai'i. Native sources of N-fixation in these soils were c. 0·2 kg ha−1 year−1 with precipitation adding <4 kg ha−1 year−1. In contrast, in sites densely colonized by M. faya, N-fixation was estimated to be more than 4 times higher at 18 kg ha−1 year−1 (Vitousek & Walker 1989). Although the main native tree, Metrosideros polymorpha, at these sites is N-limited, the large impact that M. faya alone has on the available N pool creates a more favourable environment for further biological invasions by a broader range of exotic species which in turn may alter the future vegetation composition of the ecosystem. Similarly, the introduction of Australian Acacia spp. into parts of southern Africa, a global biodiversity hotspot, has eliminated many hectares of the native and unique fynbos vegetation (Spent & Parsons 2000). In contrast, introduction of Acacia can help phytoremediate contaminated sites, although inoculation with both N-fixing bacteria and AM fungi is usually required to aid establishment (Spent & Parsons 2000). This example emphasizes the care that must be taken in introducing alien species, even where the apparent benefits in the target ecosystem are great.

Arbuscular mycorrhizal associations, nutrient exchange and networks

Most plant species form symbiotic relationships with mycorrhizal fungi, and the most widespread type is the AM symbiosis. The AM association forms between c. two-thirds of all land plant species and soil fungi in the order Glomeromycota, and AMF may increase plant diversity (Grime et al. 1987; van der Heijden et al. 1998). AMF diversity also influences plant performance as demonstrated by van der Heijden et al. (1998) who added varying numbers (1, 2, 4, 8 or 14 randomly selected from a pool of 23 AMF species) of AMF to an old-field plant community from which the AMF had been isolated. Both plant and fungal performance measures were enhanced by an increase in AM fungal species. Moreover, in a second experiment simulating a European calcareous grassland to which 4 AMF were added singly or in combination, 8 of the 11 plant species were almost completely dependent on the presence of the AMF. In contrast, Carex flacca, the only non-AM species examined, performed best in the non-AM control treatment (van der Heijden et al. 1998). Such experimental studies have been criticized on the basis that increasing the number of species present increases the likelihood of the more effective species being included (‘sampling effect’; Wardle 1999). However, the various AMF species had very different effects on the plant species when added singly, which suggests that in this case, there was no one fungus that was most effective in enhancing growth of all or most plant species (see O'Conner, Smith & Smith 2002).

An alternative approach to reconstructing plant communities is to eliminate AMF from an existing community. This has most frequently been performed using the fungicide benomyl, which of course has large non-target effects on other fungi. Newsham et al. (1995a) sprayed a lichen-rich grassland with benomyl and observed large changes in the plant community: unsurprisingly, lichens largely disappeared from the sprayed plots, but the changes in the plant community were best explained by the changes in AMF colonization of their roots (Fig. 1). The inevitable occurrence of non-target effects, notably on pathogenic fungi (Newsham, Fitter & Watkinson 1995b), complicates the interpretation of such experiments, but they have the advantage of being performed on real communities under otherwise undisturbed conditions.

Figure 1.

The effect of benomyl application on the frequency of principal lichen, moss, mycorrhizal and non-mycorrhizal higher plant species in a lichen-rich plant community at Mildenhall, U.K. Significant responses in frequency to benomyl application are indicated by *< 0·05; **< 0·01; ***< 0·001. Percentage frequencies of each species in control plots are shown in brackets [reproduced from Newsham et al. (1995a), Functional Ecology].

Arbuscular mycorrhizal fungi cannot be host-specific because only c. 200 species have been described, and yet these can colonize around 200 000 potential host plant species, although the true number of species and even the meaning of the term species for these obligately asexual organisms remain obscure. While specificity per se has not been established, AMF species can have contrasting impacts on various hosts (van der Heijden et al. 1998; Helgason et al. 2002): some AMF therefore appear to be better partners for particular plant species than others under a given set of experimental conditions. Because the roots of plants can contain several species of AMF, the ‘best’ partner may vary over time. Because AMF colonization may sometimes result in host growth depressions (at least under controlled conditions, e.g. Klironomos 2002; Rooney et al. 2011), the view has developed that AMF may act as ‘cheats’ under some circumstances, obtaining benefits (sugars) from their hosts without offering reciprocal benefits in exchange, leading to a ‘mutualism–parasitism continuum’ across which AMF operate (Johnson, Graham & Smith 1997).

The evidence that AMF can be parasitic is, however, weak, because almost no studies have measured host fitness, most relying on short-term growth responses under artificial conditions. AMF perform a range of functions for their host in addition to nutrient acquisition, including enhanced pathogen resistance and improved water relations, and no experimental designs allow for the complexity of the environmental conditions that might allow full expression of symbiotic benefit. Observing a growth depression in an experiment under a particular set of cultural conditions is insufficient evidence to prove that an AMF is cheating, especially because in most such experiments, there is no pre-existing mycorrhizal mycelium and the plant must first provide the carbon (C) for its construction before it can receive benefit. Recent evidence suggests that if one AMF transfers more nutrients than another, it also obtains more C in exchange (Kiers et al. 2011), which would reduce the possibility of cheating, as suggested by the conceptual model of Fitter (2006). How the other ‘benefits’ conferred by the AMF on the plant are regulated is unknown, but the fact that the symbiosis has apparently remained largely unchanged over 400 million years suggests that it is remarkably resistant to invasion by cheats.

One possible benefit obtained by plants colonized by AMF is their apparent ability to reduce the growth of species that are normally non-mycorrhizal. This phenomenon – a direct reduction on non-host growth as opposed to the widely observed but non-specific growth depressions that are assumed to be caused by C drain – was first noted by Francis & Read (1995), who measured damage to root growth in ruderal species. Such a mechanism might lie behind the results of Rinaudo et al. (2010) who grew a number of non- or weakly mycorrhizal agricultural weeds in competition with sunflowers in pots. They found that weed growth was lower in the mycorrhizal treatments, whether or not sunflower was present, although sunflower yield was unaffected. The AMF increased sunflower P (but not N) uptake and reduced weed P uptake whether in competition or not. AMF can potentially therefore alter the outcome of plant competition either by a deleterious effect on one competitor or by enhancing the growth of the other.

There is increasing evidence that the composition of AMF communities is determined more by soil conditions than by the composition of the plant community (Dumbrell et al. 2009), and testing this hypothesis requires experiments on communities in the field because of the problems associated with culturing AMF. Only around 70 AMF species have been successfully cultured, and the fungal taxa that are more easily trapped and maintained in pot culture, and so are used in controlled experiments, are almost certainly unrepresentative of the true functional diversity of the AMF population because they are by definition the ones most easily able to adapt to new (pot culture) conditions. Field studies are, however, by their very nature difficult to interpret due to the large number of uncontrolled variables, with the result that most studies either involve reconstruction approaches, fungicides or the ‘plant–soil feedback’ approach. All these approaches have limitations but offer some insights into the potential of AMF in shaping plant community dynamics through their interactions with their hosts.

Johnson et al. (2010) recently used a combination of the feedback and reconstruction approaches to examine the relationship among different ecotypes of the late successional AM plant species, A. gerardii, soil and AMF communities. The soil and AMF communities were collected from three different prairie field sites in the U.S.A., one of which had high levels of available P but low N, while the other two sites show the opposite pattern (i.e. low P but high N). Soil from the three sites was sterilized, and native microbial communities were re-introduced directly below a transplanted A. gerardii seedling. All possible plant ecotype–soil–AMF community combinations were screened, including set-up of non-AM controls to determine whether the native AMF conferred the best advantage to host ecotypes when in their ‘home’ combination of AMF–soil–ecotype. The results did largely support this view: arbuscule frequency was highest when AMF, host ecotype and soil were all matched in ‘home’ combinations. However, the AMF produced more extra-radical (ERM) hyphal length densities in their ‘own’ soil than other soil types, regardless of the host plant ecotype, suggesting fungal adaptation to soil of origin (Johnson et al. 2010). Similarly, Leigh, Fitter & Hodge (2011) found addition of a general soil bacterial filtrate suppressed AM hyphal growth and reduced root P content, demonstrating under some circumstances (i.e. if other than ‘home’ microbial combinations), bacterial populations can reduce AM symbiotic effectiveness. One other feature of this complex experiment by Johnson et al. (2010) was that, while AMF increased N uptake from the N-limited soil, shoot biomass was negatively related to both ERM and arbuscule frequency. Johnson et al. (2010) proposed that this effect was due to the AMF being able to meet the shoots’ demand for P in the two low P soils but not the demand for N in the low N soil as shoot N requirement is higher. This agrees with the results of other studies where, although AMF N transfer to the plant occurs, this does not always result in an increase in plant biomass (Leigh, Hodge & Fitter 2009) or total plant N acquisition (Hodge, Campbell & Fitter 2001).

Arbuscular mycorrhizal fungi can therefore transfer both P and N to the plant, but the results for N are less clear-cut than those for P: in some cases, substantial amounts of N can be transferred (e.g. Leigh, Hodge & Fitter 2009), but Hodge & Fitter (2010) found evidence for competition between plant and fungus for N, perhaps unsurprisingly given that these fungi have a high N requirement and so require substantial amounts for their own needs (Hodge & Fitter 2010). Certainly, similar to roots, AMF can show substantial proliferation in organic matter patches (Hodge & Fitter 2010) suggesting these sources are important for AMF nutrition. This may also explain why an added AMF inoculum increased N capture by Brassica napus, a non-mycorrhizal plant species, from an added organic matter patch both in monoculture and when interspecific competition with Plantago lanceolata. AMF hyphae were observed on the roots of B. napus indicating initial AMF growth but in the absence of a potential host (B. napus monocultures) or a greatly reduced root system to colonize (when in interspecific competition), increased AMF hyphal turnover may have recycled N back into the root-soil system for root capture, the majority of which were B. napus (Hodge 2003a). In contrast, AMF decreased N capture for both Lolium perenne and P. lanceolata from added organic material when grown in monoculture but increased N capture for both plant species when grown in interspecific competition. In the latter case, this effect was likely due to an impact on increased root length in the presence of the AMF (Hodge 2003b).

A more reductionist approach to the role of AMF in N capture for their host was recently taken by Fellbaum et al. (2012). Using root organ cultures grown on compartmented Petri dishes split into a hyphal-only or colonized root compartment, they demonstrated that host plant C supply to the AMF triggers the uptake and transport of N via a change in fungal gene expression particularly in the ERM. Moreover, the addition of C (as acetate) directly to the hyphal-only compartment did not stimulate N uptake. The advantage of using this approach is that N, C and P levels could be tightly controlled and were near exhausted before additional C, N or P were added to the various compartments.

Several investigators have proposed terms to explain why some AMF appear more beneficial to their hosts than others; including the AMF mutualism–parasitism continuum (Johnson, Graham & Smith 1997) and cooperative vs. less cooperative AMF (Kiers et al. 2011). As discussed above, it is almost impossible to demonstrate that AMF are actually not providing some unmeasured benefit, unless actual fitness is measured (it almost never is in such studies), especially because plants may down-regulate P uptake in favour of an AMF uptake pathway even when no net nutritional gain is observed (Smith, Grace & Smith 2009). AM fungi themselves differ in their nutritional demands (Hodge & Fitter 2010) and must balance their need for C supplied by the host, which must be ‘paid for’ by transfer of P or N, with their requirements for N and P for their own growth; these differences may be interpreted as a greater or lesser degree of ‘cooperation’ by the fungus. The fact that the AMF are obligate biotrophs and so require C from the host plant in order to complete their life cycle has traditionally led to the conclusion that the plant is therefore more in control of the symbiosis than the fungus and that the fungus must act to maximize plant fitness, if it is to be successful itself. However, the more recent evidence (Hodge & Fitter 2010; Kiers et al. 2011; Fellbaum et al. 2012) questions these assumptions and emphasizes the need to consider the evolutionary pressures that act on the fungus as well as the plant in the symbiosis (Helgason & Fitter 2009).

Because the fungi are non-specific in terms of host plant colonization, the AM mycelium can link different plants and plant species together via a common mycelial network (CMN). Plant C supply for the fungus is a volatile currency, and its exchange rate can fluctuate widely; the CMN may therefore enable the fungus to obtain a better exchange rate for its nutrient resources from its various hosts. Several studies have demonstrated that C can be moved from one plant to the roots and, in mycoheterotrophic associations also to the shoots, of other plant species via this CMN (reviewed by Robinson & Fitter 1999). It has been suggested that these transfers show that plants may support each other via their fungal partner acting as the broker, with seedlings, for example, being able to survive competition from mature plants. There is some evidence to support this view, but the results are highly variable. In a survey of the literature, van der Heijden & Horton (2009) found that growth of seedling species grown near larger plants responded positively to AM networks in 16 of 37 species (42% of cases) negatively in 15 species (33% of cases) and was not significantly affected in 12 species (29% of cases) compared to in the absence of AM networks. The ‘cases’ value include species that were tested repeatedly and the results come from a total of 13 studies. Such a phenomenon would radically alter our view of plant competition, but there are technical issues with many of these studies, and plant-to-plant C transfer via the fungal partner (except in mycoheterotrophic plants) has still to be convincingly demonstrated. The concept also runs counter to the recent evidence of reciprocal exchange among plants and AMF (Kiers et al. 2011). Certainly, in the case of AM associations, the C appears normally to be retained in the roots, and therefore, probably still retained in the fungal tissue. In other words, the fungus is moving C for its own requirements. The most likely benefit to plants in these instances is the ability to tap into an existing CMN and avoid its construction costs; that could be of great value to seedlings in particular. Utilizing the benefits of intact mycorrhizal mycelia also implies that using AMF in sustainable agriculture will require reduced cultivation to avoid their destruction.

However, nutrients such as N and P are indubitably moved via the CMN, which could still potentially reduce the impact of nutrient heterogeneity for linked plants, opening the possibility of competition among plants not just for nutrient ions in soil but for those in the CMN as well. Despite the potential importance of the CMN, little experimental field research has been conducted. In one such study, Chiariello, Hickman & Mooney (1982) applied 32P to the leaves of a donor Plantago erecta plant present in a serpentine annual grassland community. High levels of 32P were detected in the shoots of neighbouring plants after 6–7 days, but the distances between plants were small (ca 45 mm) and neither plant size or type or the distance between the receiver or donor plant were indicators of the amount of P transferred. This study, however, is based on what must be an atypical event, namely the acquisition of P from a plant by the fungus and its transfer to other plants. We need much better understanding of how N and/or P acquired by the fungus from soil is partitioned among hosts.

In the natural environment, plants typically have several AMF taxa present in their roots. If some AMF are consistently less effective at promoting plant fitness than others, those species should become functionally redundant over evolutionary time. However, fungal species apparently differ principally in their response to the soil niche (Dumbrell et al. 2009) and in their nutrient demands (Cavagnaro et al. 2005; Hodge & Fitter 2010), rather than in symbiotic effectiveness per se. As far as we know, all of the AMF in a community will form their own CMN with different sets of individual plants, although the identity of those networks may be unpredictable and idiosyncratic (Dumbrell et al. 2010).


There is unequivocal evidence that some soil microbes can influence, often decisively, plant competitive interactions, at least in some conditions, and that interactions among those microbes can determine the outcome of competition. However, the bulk of this evidence comes from experiments in which the identity of the microbes is unknown, or the experimental conditions were highly artificial, or in many cases, both. Perhaps unsurprisingly, therefore, few studies or models of competition take this effect into account. There is an urgent need for new experiments, carried out under realistic conditions and with properly characterized microbial communities, using modern techniques to demonstrate which microbial taxa are involved and what are the mechanisms by which they influence plant competition. The technical challenges are large (but not insuperable), and most of the existing approaches have important limitations, but new technologies allow the diversity of the microbes involved to be characterized, and in some cases, the species to be identified, as well as pinpointing likely mechanisms of interaction. Experiments using cultured taxa have the severe drawback of restricting the species pool to those that can grow well in culture, which are almost by definition likely to be atypical. Almost none of the existing studies demonstrate the mechanism of the putative interaction unequivocally. Where interactions are held to be direct (e.g. reduced plant growth caused by pathogens or the apparent direct effect of AM fungi on some non-mycorrhizal plant species), there is a need to apply Koch's principles: the microbial agent needs to be identified in the field, the damage to be quantified and its ability to cause the damage to be recreated. Many interactions are, however, likely to be indirect, often acting via an alteration in resource supply, in which case the resource in question, and the mechanism for the reduction in the availability of the resource, must be identified, using both quantitative (e.g. stable isotopes) and qualitative (e.g. metabolomics) techniques. Engagement with new technologies could rapidly alter the relatively small progress that has been made in this field in the last 20 years.