Specificity and resilience in the arbuscular mycorrhizal fungi of a natural woodland community


*Correspondence: Thorunn Helgason (fax +44 0 1904 328505; e-mail th7@york.ac.uk).


  • 1The composition of communities of arbuscular mycorrhizal (AM) fungi can have a large effect on the performance of their plant hosts. The role of individual fungal species in shaping this response is as yet unresolved.
  • 2We have used the fungicide benomyl to alter the community of AM fungi in undisturbed monoliths of soil in a natural community. Changes in the community were characterized by root colonization (%RLC), cloning, sequencing and tRFLP of a partial SSUrDNA fragment. Eleven plant species were sufficiently abundant in the monoliths to be examined.
  • 3In the highly mycorrhiza-dependent perennial herb Ajuga reptans, phosphate concentration was significantly reduced after benomyl treatment over a full growing season. The other plant species showed low colonization and no significant difference in phosphate concentration after benomyl treatment.
  • 4Although colonization in A. reptans was reduced, many mycorrhizal fungi survived in the roots. Some became more abundant following fungicide treatment, suggesting competitive release. Fungi that increased were generalists that have been identified in field samples from published studies colonizing a wide range of plant species. Those that declined were specialists with a narrow host range; five types had not been recorded previously in field samples.
  • 5AM fungi in this study differed greatly in their response to perturbation, independent of the identity of the host plant. If such functional diversity is widespread, then elucidating the part played by AM fungal diversity in regulating plant community structure will be key to our understanding and management of ecosystems.


Uncultured microorganisms dominate soil ecosystems (Fitter 2005). One key group in which many species may be unculturable is the Phylum Glomeromycota that forms arbuscular mycorrhizas (AM). These fungi are among the most abundant and ecologically important symbionts on earth, forming mycorrhizas with around two-thirds of all plant species and occurring in virtually all ecosystems. The Glomeromycota were traditionally viewed as a species-poor phylum. Relatively few AM fungal species have been successfully grown in culture, with all of them observed only in symbiosis with a host plant. Almost all AM fungi are non-specific symbionts, readily colonizing the roots of most plant species they encounter (Smith & Read 1997). This fact, combined with the apparent low diversity of this taxon (150–200 described species, Morton & Benny 1990) and its enormous host range (> 2 × 105 plant species; Fitter & Moyersoen 1996), has led to the view that AM fungi are non-specific. If so, we would predict that AM communities would not show a high degree of differentiation among host plants. However, results of spore-based morphological studies (Sanders & Fitter 1992; Bever et al. 1996; Eom et al. 2000) and molecular analyses of the occurrence of AM fungi within the root systems of plants collected in natural communities (Clapp et al. 1995; Helgason et al. 2002; Husband et al. 2002; Öpik et al. 2003; Vandenkoornhuyse et al. 2003; Johnson et al. 2004; Santos et al. 2006) demonstrate not only that the genetic diversity of these fungi is much greater than the morphological taxonomy suggests, but also that different plant species often harbour quite distinct AM fungal communities. This suggests that even where AM fungi display a high degree of non-specificity, other factors are influencing community structure. As the majority of the fungi revealed by molecular techniques are apparently unknown in culture, the question has been raised as to whether these uncharacterized species play distinct ecological roles and may be responsible for some of the profound ecological impacts of AM fungi (Newsham et al. 1994; van der Heijden et al. 1998; Hartnett & Wilson 1999; Dhillion & Gardsjord 2004).

Moora et al. (2004) recently showed that different communities of AM fungi had large effects on the performance of two closely related, coexisting plant species. What remains to be resolved is the extent to which particular members of the AM fungal community associated with a plant species have distinctive impacts on plant performance. The null hypothesis must be that all the fungi act in the same way, transporting P to the plant, protecting it from pathogens, binding roots to soil and hence promoting drought resistance, and performing any of the other functions that have been ascribed to mycorrhizas (Newsham et al. 1995). However, many of these activities are mutually incompatible: enhancing P uptake depends on growth of the extra-radical mycelium outside the rhizosphere (Avio et al. 2006; Oliveira et al. 2006), promoting drought resistance requires mycelial growth in the rhizosphere, and pathogen protection must depend on the internal mycelium. It seems inherently more likely therefore that coexisting fungi have distinct functional capabilities (van der Heijden et al. 2003; Oliveira et al. 2006). Those fungal taxa that display a degree of specificity towards particular host plants are likely to offer those plant species the greatest symbiotic benefit, as there is likely to have been a co-evolutionary response by both partners (Fitter 2005).

In an earlier laboratory study (Helgason et al. 2002), we showed that plants from a semi-natural woodland varied both in their overall response to AM colonization, and to specific AM fungi. The same study suggested that AM fungi display a range of colonization strategies. In particular, the taxa that were able to colonize all the host plants were those that resulted in a consistent but small benefit to the plant, measured as P acquisition; these we identified as essentially opportunist taxa. Here, we have used the fungicide benomyl to alter the fungal community colonizing roots of the same natural plant community used previously by Helgason et al. (2002). We hypothesized that the fungicide would not affect all fungi equally, and that the fungi most likely to recover quickly from this environmental perturbation would be r-selected fungi that display ‘weedy’ characteristics of broad host range and weak symbiotic effectiveness. We also predicted that a change in the fungal community favouring generalist types would be accompanied as before by a reduction in plant phosphate uptake, at least in the most mycorrhiza-dependent plants, because the less effective symbionts would be more abundant.

Materials and methods

experimental design

The experiment was conducted at Pretty Wood near York, UK (grid ref. SE 732 867, altitude 50 m), under a mature canopy of Quercus petraea (Mattuschka) Liebl. and sycamore Acer pseudoplatanus L. We used the fungicide benomyl ([1-[(Butylamino)carbonyl]-1H-benzimidazol-2-yl] carbamic acid methyl ester) at a concentration of 1.125 g L−1 to disrupt the AM fungal community associated with the ground flora in this woodland. Benomyl has been widely used to reduce root colonization by AM fungi (Fitter & Nichols 1988; Newsham et al. 1994; Hartnett & Wilson 1999; Smilauer & Smilauerova 2000; Smith et al. 2000; Callaway et al. 2004). However, it penetrates soil poorly and we adapted the technique we previously used at the same site (Merryweather & Fitter 1996) in which entire monoliths of soil (with growing plants) were lifted intact, placed in mesh baskets, replaced in situ and raised regularly for immersion in a bath of fungicide for 3 min, found to be sufficient time for the monolith to be thoroughly drenched. Control baskets were immersed in water at the same time.

The experiment was set up during September and October 1996. Twenty-four cubic monoliths of soil (30 × 30 × 30 cm; volume 27 L) were created in three blocks of eight in adjacent sites in the wood, no more than 100 m apart, selected to represent distinct subcommunities to allow for variation in plant community structure and to ensure that some plant species were well represented in the experiment. Each basket contained between three and six plant species; in block 1 all had Primula vulgaris Huds. and Glechoma hederacea L., block 2 all had Ajuga reptans L. and Teucrium scorodonia L., and block 3 all had Mercurialis perennis L. Across all 24 baskets, the most abundant species were G. hederacea and Poa trivialis L. (17/24), A. reptans (18/24), Oxalis acetosella L. (13/24) and Lysimachia nummularia L. (12/24). The treatments (plus and minus fungicide) were allocated randomly but equitably within these subsets of eight baskets, to give 12 of each (four in each subcommunity). Treatments were applied monthly from 18 March 1997 to 18 June 1998.

Leaf samples of the principal species were taken on three occasions (28 May 1997; 21 May 1998; 20 July 1998), dried for 48 h at 70 °C and a subsample digested for phosphorus concentration assay by the molybdenum blue method (Allen 1974). Root samples were taken at the final harvest (20 July 1998); part of each sample was analysed for P as above and part stained with 0.01% acid fuchsin (Kormanik & McGraw 1982) (using a phenol-free modification) and percentage root length colonized by AM fungi (%RLC) assessed using the magnified intersection method (McGonigle et al. 1990). Soil phosphate concentration was measured by extracting 1 g soil in 20 mL deionized water for 30 min in a rotating shaker; P concentration in the supernatant was measured after centrifugation as above. Leaf N concentration was measured on ground material using a CHN NA2100 Brewanalyser (CE Instruments, Milan, Italy). There were no treatment effects on leaf nitrogen or soil phosphorus (data not shown).

molecular methods

DNA was extracted from subsamples of the final harvest roots using a PEX extraction method (Edwards et al. 1997). As only one species, A. reptans, showed either a high degree of root colonization (%RLC) in the controls, or a significant decrease in %RLC and tissue P concentration, only A. reptans roots were analysed in detail. Eighteen root samples, nine from control and nine from benomyl-treated monoliths were analysed. Partial SSU DNA fragments (c. 550 base pairs) were amplified using a universal eukaryotic primer NS31 (Simon et al. 1993) and a fungal primer AM1 (Helgason et al. 1998). PCR products were cloned and digested with the restriction enzymes Hinf I and Hsp92II (Promega, Southampton, UK) after Helgason et al. (2002). The abundance of each restriction fragment length polymorphism (RFLP) profile was scored from 2% agarose gels. Representative clones were sequenced; sequencing was performed by Lark Technologies Ltd (Essex, UK), using ABI BigDye™ chemistry (Applied Biosystems, Warrington, UK).

Those samples from all host species sampled that gave a sufficient product yield in the initial PCR were amplified using NS31 and AM1 primers labelled with WellRED Beckman Dye D4-PA and D3-PA, respectively (Sigma-Proligo, Gillingham, UK). The NS31-D4-PA primer was diluted 1 : 3 with unlabelled NS31 to compensate for the dye's greater intensity, resulting in equivalent peak heights in the subsequent tRFLP analysis. PCR products were checked on an agarose gel and 5-µL aliquots were digested with Hsp92II as described above. Digested PCR products were resuspended in deionized formamide following ethanol precipitation. A 15-µl aliquot was added to 30 µL SLS loading solution (containing 0.16 µL Size Standard600; Beckman Coulter UK Ltd, High Wycombe, UK). Fluorescently labelled fragments were separated by capillary electrophoresis and detected by laser-induced fluorescence using the CEQ 8000 automated gene sequencer (Beckman Coulter UK). Samples were denatured at 90 °C for 2 min before injection at 2 kV for 25 s and separation for 65 min. Fragments were compared with the internal standard using CEQ 8000 software for fragment size analysis. Samples were rejected if maximum peak height was outside a range of 10–100K fluorescence units; four samples exceeded the maximum and were diluted 1 : 1 with loading solution and re-run. Each sample was scored for the presence/absence of two peaks diagnostic for specific AMF groups: Acaulosporaceae and Glomus intraradices/G.mosseae (Table 2).

Table 2.  Sequence types identified in A. reptans roots using NS31-AM1 PCR
Identity*Fragment sizesClone numbers (relative abundance)Host rangeNearest similar sequence (%)§
HinfIHsp92IIControl Benlate 
  • *

    Sequence types were given names derived from the isolates with greatest similarity. The three sequence types with low similarity to databank sequences (Glo54–56; the numbering is consistent with previous publications) were judged to be new types.

  • †RFLP fragment sizes reflect the order in which they occur, and are given for a single representative sequence without primers; they may vary within a group by 1–3 bp in total length. Acau2 and Glo2 have two RFLP patterns as the resulting sequences do not resolve into two groups. Only sequence types that fell into the groups resolved by NS31-AM1 PCR (in bold), i.e. the Diversisporales and Glomerales (Schüßler et al. 2001), were assessed for host range, as amplification of the Archaeosporales and Paraglomales is unpredictable with this primer pair. Italicized fragments are those used in the t-RFLP analysis.

  • ‡Host range was estimated by the number of plant genera in which the sequence type has been found in other published studies and data base entries. This was restricted to studies using field material (Helgason et al. 1999; Daniell et al. 2001; Helgason et al. 2002; Husband et al. 2002; Kowalchuk et al. 2002; Vandenkoornhuyse et al. 2002; Regvar et al. 2003; Oba et al. 2004; Scheublin et al. 2004; Whitfield et al. 2004; Wirsel 2004; Santos et al. 2006; Vallino et al. 2006). (Table S1 and Table S2 in Supplementary Material.)

  • §%Similarity is given where a full alignment was generated. Representative sequences have been deposited in EMBL; accession numbers AJ854081AJ854105.

Ascomycete50390 146 26729 41  U32414 100
Archaeosporaceae478,23,5237 26910 22  AF131054 98
Archaeosporaceae474,23,5147,90,260,5 0  1  AJ563887(96)
Archaeo/Acau17474,23,5147,90 265 0  4  AJ563887(96); AF074346(96)
Archaeosporaceae301,177,22,6102 137 267 0  1  AF131054 (93)
Archaeosporaceae485,26,5246 27010 14  AJ563866 (–)
Archaeosporaceae485,26,5238,260,516 13  AJ563887 (98)
Archaeosporaceae419,55,23,5147,90,260,5 0  1  AJ563887 (95)
Acau1177,428,5148,90,23 249 00.0 10.008 1AJ716002 (99)
Acau2121,384,5/505,5148,90,23 249250.197 60.050 1AF074346 (99)
Acau7121,384,5148,90 272110.087110.092 7Y17633 (99)
G. etunicatum261 249148,90 272 40.031 00.0 3Y17644 (99)
G. hoi503,5144,93 271 90.071200.16711AJ716012 (100)
G. intraradices120,383,595 142 271330.260230.19222AJ309462 (99)
G. mosseae260,244,595 143 271 30.024 80.06714AY635833 (100)
G. spurcum120 141 249148,90 271 00.0 90.075 1AJ315525 (99)
Glo14120,49,90,241,5237 268 00.0 10.008 5AF437664 (100)
Glo18168,90,244,5143,93 271 10.008 20.017 8AF437675 (99)
Glo2169,331,5/120,49,331,590 147 268240.189 60.05011AJ418882 (99)
Glo3(UY1227)120,49,334,5237 271100.079190.01615AF437719 100
Glo4168,90,243,5236 270 00.0 50.04210AJ716005 (99)
Glo54(new)120,380,5237 268 70.055 10.008 1AJ496085 (97)
Glo55(new)502,5237 270 00.0 30.025 1AY129612 (97)
Glo56(new)168,333,5235 271 00.0 50.042 1AJ716007 (96)

fungal host range

In order to identify the host ranges, a reference sequence of each type was used as a query in a BLAST search (Altschul et al. 1997), and the nearest hits used to identify published studies where the sequences have been identified from field roots. Sequences from unpublished studies were excluded because the methodology could not be verified from the data base entries alone. Laboratory experiments using cultured isolates and trap cultures using field soils as inoculum were likewise excluded on the grounds that these represent artificial situations (additional information in Table S1 and Table S2 in Supplementary Material).

statistical methods

Data regarding root colonization by AM fungi were arcsine-square root transformed and tested using the GLM command for bivariate analysis of variance for blocked design of unbalanced data using SPSS 11. All other data were tested using standard bivariate analysis of variance for a blocked design. Raw sequence data were checked by eye for quality and forward and reverse sequences assembled and checked using Lasergene v.6 software (DNASTAR Inc., Madison, WI, USA). Alignments were generated using ClustalX alignment (Thompson et al. 1997), and NJ tree building (Saitou & Nei 1987). Correlations were performed using SPSS 11.


The intensity of mycorrhizal colonization was unusually low in this community: only one species, A. reptans, was heavily colonized. Four other species had > 10% root length colonized in control plots and in all these benomyl reduced colonization, though not significantly (Table 1). Benomyl application reduced the mean community root length colonized by AM fungi, averaged across 11 plant species, from 15% to 4% at the final harvest (F1,101 = 8.73, P = 0.004). Among individual species the reduction was only significant for A. reptans, which had the highest colonization in control plots (45 ± 12% of root length), reduced to 12 ± 7% after benomyl treatment. The phosphorus (P) concentration of shoots and roots was affected by benomyl treatment only in A. reptans (Fig. 1), where it was reduced by 37–40% except at harvest 1; at the final harvest, overall leaf P concentration was reduced by one-third in benomyl-treated A. reptans (1.8 vs. 2.7 mg g−1). There was no significant relationship between shoot P concentration and root length colonized, either for all plants together, or for A. reptans on its own (Fig. S1 in Supplementary Material). Although benomyl treatment reduced colonization of roots by AM fungi, that reduction was not simply responsible for the marked decline in P concentration in A. reptans.

Table 1.  Colonization of roots by arbuscular mycorrhizal fungi in the principal species sampled in the plots. Only a single plant of M. perennis was sampled from a control basket. The only species in which the reduction in colonization was significant was A. reptans: F1,14 = 9.54, P = 0.008
SpeciesControl (%RLC)Benomyl (%RLC)
Poa trivialis1.4 3.0
Primula vulgaris2.3 2.4
Ajuga reptans44.612.3
Lysimachia nummularia13.0 2.7
Teucrium scorodonia16.7 6.8
Oxalis acetosella0.0 6.7
Glechoma hederacea16.1 1.7
Mercurialis perennis(19.0) 3.0
Silene dioica0.0 0.0
Circaea lutetiana2.8 0.0
Urtica dioica0.0 0.0
Figure 1.

Leaf and root P concentrations of A. reptans in benomyl-treated and control baskets. H1–H3 are harvests 1–3.

Morphological analyses can give only limited information on the species of fungi involved. We therefore obtained 408 clones from 18 roots of A. reptans, from both control and benomyl plots (mean 22.7 clones per root). RFLP analysis of the clones revealed a total of 26 RFLP types using HinfI and Hsp92II. Sequencing confirmed that RFLP types sampled by the NS31-AM1 PCR could be grouped into 24 sequence types. Of these, one was a group of Fusarium-like Ascomycetes, one was a mix of Archaeosporaceae and an Acaulospora sp. that could not be separated using this combination of restriction enzymes, and the remaining 23 were AM fungal sequence types that could be unambiguously resolved (Glomeromycota; Table 2). Individual roots had between 2 and 12 sequence types (mean 7.2). This rich diversity of AM fungi from a single plant species at one site is comparable with that recorded in other recent detailed studies (Vandenkoornhuyse et al. 2003; Wirsel 2004).

Most of the fungi (17/24; 71%) increased in frequency in the benomyl-treated plots, including the Ascomycete group. There was great variation in the response of the AM fungi to benomyl treatment: while some types were almost eliminated, others were more abundant than in the control. The variable response was reflected in a lack of correlation between the abundance of each sequence type in the control and treated plots.

Five sequence types that were abundant in A. reptans had not been previously identified in any other host species in the field, either at this site or others; only one of those is known in culture (Glomus spurcum). The values of the host range, defined as the number of genera from which the sequence type had been recorded, were examined in relation to the abundance of each type in the controls and benomyl-treated plots. While there was no significant correlation between host range and abundance in control plots (Fig. 2a), the treated plots showed a significant correlation between the abundance of sequence type and host range (Fig. 2b). Sequence types with wide host range were, on average, more resilient to this treatment, probably reflecting an ability to recover quickly and colonize any available root.

Figure 2.

(a) Relationship between relative abundance of clones of each AM sequence type isolated from roots of A. reptans in the control baskets and their host range. Total number of AM clones (n = 127) was used as the reference point (Kendall's tau correlation = 0.365, P > 0.05; number of clones used rather than RA). (b) Relationship between number of clones of each AM sequence type isolated from roots of A. reptans in the benomyl-treated baskets and their host range. Total number of AM clones (n = 120) was used as above, Kendall's tau correlation = 0.450, P < 0.05; clone numbers used as above. Open symbols are taxa used in the studies discussed in the text; ▵, G. hoi; □, G. mosseae; □, G. intraradices; ○, Glomus sp. Glo3.

In the tRFLP analysis, low colonization resulted in poor amplification yields, and only 35 roots yielded sufficient PCR product for tRFLP analysis (Fig. 3). The abundance of the two AM group-specific fragments showed a response to the treatment similar to that observed in A. reptans alone. The proportion of roots showing Acaulosporaceae and G. intraradices/G. mosseae fragments was similar in the control treatment. However, the Acaulosporaceae were reduced whereas G. intraradices/G. mosseae doubled in frequency after benomyl application, suggesting that the response of the fungi to the treatment is independent of the identity of the host plant.

Figure 3.

Proportion of terminal restriction fragments (TRF) that could be ascribed to Acaulospora spp. and G. intraradices/G. mosseae on the 35 roots from seven species from which amplification was obtained, viz. 8 (control)/6 (benlate) A. reptans, 2/1 G. hederacea, 0/2 Circaea lutetiana L., 3/1 M. perennis, 3/3 L. nummularia, 1/2 O. acetosella, 3/0 P. trivialis; 20 plants from control treatments and 15 from benlate.


There has been much debate about the ecological role of individual AM fungal taxa (Hart et al. 2003). Molecular studies of AM associations in natural communities have begun to reveal that AM fungal taxa vary greatly in specificity and host range (Clapp et al. 1995; Helgason et al. 2002; Husband et al. 2002; Klironomos 2003; Öpik et al. 2003; Vandenkoornhuyse et al. 2003; Johnson et al. 2004). We show here that this variation can be linked to the biology of the fungi. Treating A. reptans growing in a natural community with the fungicide benomyl reduced both P concentration and colonization, but there is not a straightforward correlation between colonization and tissue P concentration, as would be predicted if all AM fungi were functionally equivalent. The fungal taxa least affected by the fungicide were widespread taxa that have been recorded with many other plant genera, not only at this field site but elsewhere (Öpik et al. 2006). The rare and specific taxa, some of which have not been recorded from field samples previously, may therefore have been principally responsible for promoting P uptake in A. reptans. Cultures of Glomus sp. UY1225 (Glo3) and G. hoi, two of the widespread taxa identified in A. reptans roots in this study (Fig. 2b), have been used as symbionts in a previous laboratory experiment with A. reptans and several other co-occurring plant species from the same field site. Both fungal taxa were successful colonisers, and both resulted in only a small increase in P uptake by A. reptans relative to uncolonized controls (Helgason et al. 2002). The same effect was seen with most other plant-fungus pairs in the experiment. G. mosseae and G. intraradices, two commonly cultured taxa, are also among the most commonly identified sequence types in field studies. G. mosseae increased in frequency, and G. intraradices remained the most abundant AM type after benomyl treatment, at the expense of Acaulospora spp. That result is consistent with Fig. 2(b) (open symbols) and reinforces the view that these fungi are indeed generalists. This interpretation of course depends on demonstrating that the generalist taxa are less able to promote uptake of P, something this study was not able to do. It is possible that this response is due to a differential effect of benomyl on the AM fungi that may not necessarily reflect a difference in function. Clearly, the next step is to determine whether or not there is any functional diversity among taxa in the field. Diversity in P-uptake has been demonstrated even within a single species in laboratory experiments (Munkvold et al. 2004), so it would be reasonable to predict such diversity in a field community.

There are a number of underlying assumptions made in equating relative abundance of clones with abundance of root colonization, e.g. that all targets amplify equally and in proportion at all abundances (Helgason et al. 1999). It is notable that no AM taxa found in the control samples are absent in the treatment, suggesting that the increased diversity of taxa may be a function of the reduced dominance of the abundant AM sequence types. Whether or not this is the case, the fact remains that the overall abundance of AM sequence types relative to other fungi declines and a number of taxa with apparently restricted host range decreased more. Our data provide the framework then for explicit testing of the hypothesis that host range is linked to the biology of the fungus, in this case to disturbance in the form of fungicide application.

Benomyl treatment does not eliminate fungi in the Glomeromycota but it can be used to control them by regular treatment (Fitter & Nichools 1988), as undertaken here. It is likely that benomyl kills the extra-radical mycelium, which is then regenerated from mycelium surviving in roots. This interpretation is consistent with the finding that taxa that we have found consistently to produce extensive external mycelium (e.g. G. intraradices, G. hoi) recovered most effectively when regularly treated with benomyl, as compared with Acaulospora spp. It is also likely that taxa that are most effective at producing and maintaining a large extra-radical mycelium will have a large host range, as they will encounter a wider variety of roots. Host range is therefore likely to be linked, as shown here, with other aspects of fungal biology.

Most studies of AM fungi are conducted with a limited number of species, such as Glomus intraradices, G. mosseae and Gigaspora margarita, because they are easy to culture and will colonize almost any host (Smith & Read 1997). Even some of these species display preferential patterns of colonization when offered different host plants (van der Heijden et al. 1998; Helgason et al. 2002; Hart et al. 2003). However, AM fungal species that are abundant in the field, whether revealed by spore collections (Merryweather & Fitter 1995) or molecular techniques (Helgason et al. 2002), may be difficult or impossible to bring into culture. Our data imply that these uncultured species may be responsible for important symbiotic functions. An increasing number of studies have shown that AM fungi are not randomly distributed among plant hosts in natural communities (e.g. Sanders & Fitter 1992; Clapp et al. 1995; Bever et al. 1996; Eom et al. 2000; Helgason et al. 2002; Husband et al. 2002; Öpik et al. 2003; Vandenkoornhuyse et al. 2003; Johnson et al. 2004; Öpik et al. 2006), but none of these have shown whether this pattern is caused by host specificity. An alternative explanation would be control by environmental factors. Abiotic niche differentiation among the fungi might lead to these patterns, as the plant species will also respond differentially to the same factors. This study demonstrates that AM fungi differ in both host-range (i.e. the biotic niche) and symbiotic function, and that these two characteristics may be linked. Therefore, the discovery of high diversity of AM fungi in a wide range of natural communities, and notably in the roots of single plant species (Helgason et al. 2002; Vandenkoornhuyse et al. 2003; Wirsel 2004) as here, seems likely to reflect diversity both in function (Newsham et al. 1995) and host-range, which will play a major role in controlling the diversity of the plant communities with which they interact (Hart et al. 2003).

This study represents evidence from a natural ecosystem that the high levels of AM fungal diversity found associated with individual plant species allow diversity in functional properties of individual coexisting AM fungal taxa. This functional diversity may provide the mechanism by which AM diversity might regulate plant community diversity: plant species may be expected to persist best in communities where their specialist symbionts occur, an outcome that would have large implications for the conservation of endangered species, for the ability of communities to respond to rapidly changing environments, and for our ability to manage invasive plant species. A key priority in rhizosphere biology is therefore to elucidate these patterns of diversity, colonization and function in natural communities.


This project was funded by the Natural Environment Research Council grant No. GR3/10253 to A.H.F. and J.P.W.Y. We gratefully acknowledge Castle Howard Estate for permission to work in Pretty Wood. We would also like to thank Jean Denison, Irene Watson, Phyllis Wilson and David Sherlock for technical assistance, and the editor and two anonymous referees for their helpful comments on earlier versions of the manuscript.