Establishment, persistence and effectiveness of arbuscular mycorrhizal fungal inoculants in the field revealed using molecular genetic tracing and measurement of yield components


Author for correspondence:
Manuela Giovannetti
Tel: +39 050 2216643


  • Inoculation of crop plants by non-native strains of arbuscular mycorrhizal (AM) fungi as bio-enhancers is promoted without clear evidence for symbiotic effectiveness and fungal persistence. To address such gaps, the forage legume Medicago sativa was inoculated in an agronomic field trial with two isolates of Funneliformis mosseae differing in their nuclear rDNA sequences from native strains.
  • The inoculants were traced by PCR with a novel combination of the universal fungal NS31 and Glomeromycota-specific LSUGlom1 primers which target the nuclear rDNA cistron. The amplicons were classified by restriction fragment length polymorphism and sequencing.
  • The two applied fungal inoculants were successfully traced and discriminated from native strains in roots sampled from the field up to 2 yr post inoculation. Moreover, field inoculation with inocula of non-native isolates of F. mosseae appeared to have stimulated root colonization and yield of M. sativa.
  • Proof of inoculation success and sustained positive effects on biomass production and quality of M. sativa crop plants hold promise for the role that AM fungal inoculants could play in agriculture.


Recently, considerable attention has been paid to the management of soil biota as providers of key ecological services (Myers, 1996; Barrios, 2007). Such organisms of particular anthropogenic interest are often referred to as ‘ecosystem engineers’ and ‘biofertilizers’. Indeed, more and more biota are deliberately released to the environment with the aim of restoring ecosystems, combating pollution and pests, or profiting from plant growth-stimulating effects of root–microbial symbioses. One of the most important plant–microbe mutualisms is the association formed between plant roots and arbuscular mycorrhizal (AM) fungi of the phylum Glomeromycota. The large majority of land plants, including many important agricultural fodder and grain crops, form arbuscular mycorrhizas (Smith & Read, 2008), whose main beneficial effect for plants lies in much improved soil exploitation by the extraradical hyphal network (Giovannetti et al., 2001; Avio et al., 2006). Active uptake, translocation, and transfer of mostly poorly plant-available soil mineral nutrients, such as phosphorus (P) and zinc (Zn), by these fungi are thought to make the greatest contributions to improved plant growth (Smith & Read, 2008), in addition to contributions to soil aggregate formation and protection of their host plants against biotic and abiotic environmental stressors (Newsham et al., 1995). Based on their supportive function in plant nutrition, AM fungi are often imprecisely referred to as ‘biofertilizers’, but unlike N2-fixing rhizobia, they do not actually contribute new mineral nutrients and thus are better referred to as ‘bio-enhancers’ of plant performance. Bio-enhancement through deliberately released AM fungal strains could profitably be used in low-input agriculture, provided that the fungal inoculants are effective in promoting crop yield and quality and are able to persist among local residents of the indigenous AM fungal assemblages.

Initiatives towards improvements in low-input and organic farming have attempted to implement agricultural management strategies favouring crop plant-beneficial AM fungi. In recent years, much effort has been dedicated to finding suitable formulations for AM fungal propagules and appropriate means for their application to the field (Gianinazzi & Vosatka, 2004). Such efforts seem justified, as a recent meta-analysis showed that plant biomass production and P uptake in the field are usually positively correlated with AM fungal root colonization, which can even be improved by inoculation (Lekberg & Koide, 2005). However, few attempts have been made to verify the success and effectiveness of inoculated AM fungi in the field (Azcón-Aguilar et al., 1986; Farmer et al., 2007; Ceccarelli et al., 2010; Mäder et al., 2011; Pellegrino et al., 2011; Sýkorováet al., 2012). Given the cost of producing, and the effort of applying, AM fungal inocula on a large scale, it is important to seek ways to verify whether inoculated AM fungi establish functional symbioses, and whether they contribute to yield improvements. Only sustained increases in crop yield and fungal persistence among native community members after inoculation justify any investment in applying artificially propagated AM fungal inocula. However, as AM fungal inoculants are most often of foreign origin, concerns have been raised about the possibility of invasive spreading after release to the field and further negative ecological consequences (Schwartz et al., 2006). Therefore, before inoculation of non-native AM fungi can be promoted for low-input agriculture, effects on soil microbial biodiversity and the functioning of natural ecosystems will have to be carefully studied. Clearly, there is an urgent need for thorough multifaceted investigations, involving molecular genetic monitoring of fungal establishment and persistence after inoculation, as well as monitoring of the sustainability of symbiotic benefits for crop yield.

In the past, consequences of inoculating AM fungi could only be assessed via microscopic measurements of root and soil colonization and via biometric and nutritional analyses of the inoculated host plants (McGonigle, 1988; Lekberg & Koide, 2005). However, the molecular genetic tool box now enables highly specific and sensitive tracing of AM fungi in any type of environmental sample (Helgason et al., 1998; Redecker, 2000; Krüger et al., 2009) by either sequencing (Öpik et al., 2009; Jumpponnen et al., 2010; Tedersoo et al., 2010) or molecular genetic fingerprinting (Vandenkoornhuyse et al., 2003; Gollotte et al., 2004), or a combination of both (Börstler et al., 2010; Kivlin & Hawkes, 2011).

The aim of this study was to monitor the success of inoculating two non-native isolates of the AM fungus Funneliformis mosseae (new classification by Krüger et al., 2012; formerly known as Glomus mosseae) in the field.

In the present molecular ecological field investigation, we verified fungal establishment and persistence via molecular genetic tracing in roots, and bio-enhancement in terms of crop yield and nutritional quality measurements. Mycorrhizal root colonization and plant yield were measured 3 months and 2 yr after inoculation. AM fungal inoculation benefits for crop yield and quality were also measured after the first year. Furthermore, the first evidence of effects of non-native AM fungal inoculants on the community of native AM fungi was recorded.

The specific questions addressed in this study were as follows.

  • Can non-native isolates of F. mosseae be distinguished phylogenetically from native strains of this AM fungal species by means of a novel nuclear rDNA marker spanning the variable 3’ end of the small subunit (SSU) gene, the highly polymorphic internal transcribed spacer 2 (ITS2) and the variable 5’ end of the large subunit (LSU)?
  • Is it possible to combine previously established polymerase chain reaction–terminal restriction fragment length polymorphism (PCR-(T)-RFLP) community fingerprinting with phylogenetic strain identification, based on concatenated sequence stretches of one single long environmental PCR amplicon that covers all known phylogenetically informative parts of the nuclear rDNA cistron?
  • Do non-native inoculants of F. mosseae colonize roots of the forage legume Medicago sativa when in competition with indigenous strains of the same phylospecies, and do they persist and contribute to improved crop yields over more than one season?
  • Is inoculation of non-native AM fungal isolates altering the composition and structure of the indigenous AM fungal community?

Materials and Methods

Field site

The agronomic field inoculation trial was situated at the Rottaia Experimental Station of the Department of Agronomy and Agroecosystem management of the University of Pisa in Pisa, Italy (43°30′N, 10°19′E, 1 m asl). The soil at the site is a sandy loam with 65.5% sand, 23.9% silt, 9.6% clay, 17.3 g kg−1 organic matter (Walkley–Black) and a pH (H2O) of 8.3 with the following total nutrient concentrations: 0.9 g kg−1 N (Kjeldahl), 508.6 mg kg−1 total P, and 15.6 mg kg−1 available P (Olsen). Climatic conditions at the experimental station are typical for Mediterranean regions.

Evaluation of a novel DNA marker for molecular genetic tracing of a common AM fungus

The universal fungal PCR primer NS31 (Simon et al., 1992) and the AM fungus-specific primer LSUGlom1 (Renker et al., 2003) were used to generate c. 2200-bp-long PCR amplicons of the nuclear ribosomal DNA cistron, which is rich in information for reliable phylogenetic strain discrimination (Stockinger et al., 2010).

Fungal material  Inocula of the non-native isolates AZ225C (collector J. C. Stutz) and IMA1 (collector B. Mosse) of Funneliformis mosseae (T. H. Nicolson & Gerd.) C. Walker & A. Schlüβler, with origins in the USA and UK, respectively, were produced in seven pots (18 l each) filled with a steam-sterilized 1 : 1 volumetric mixture of loamy soil and TerraGreen (calcinated clay; OILDRI, Chicago, IL, USA) with the addition of 500 g of crude inoculum from the International Microbial Archive (IMA collection) at the Department of Crop Plant Biology, University of Pisa, Pisa, Italy. To produce mock inoculum, an additional seven pots were set up by mixing the substrate with 500 g of a sterilized mixture of equal quantities of the crude non-native inocula. Ten individuals of Zea mays L. were grown as host plants for 4 months. All pots received 1.5 l of a microbial wash of the two inoculants.

In order to propagate native F. mosseae isolates for characterization, soil-trap cultures were set up with soil collected from the rhizosphere of 72 mycorrhizal plants from the experimental field site. Six plastic pots (600 ml) per host plant species (Cicer arietinum L., Lolium multiflorum Lam., Matricaria chamomilla L., Medicago sativa L., Plantago major L., Rudbeckia hirta L., Sorghum halepense (L.) Pers., and Trifolium alexandrinum L.) were set up, totalling 48 trap cultures, which were grown at a day : night temperature of 22 : 15°C and under a 14-h light period in a glasshouse for 4 months.

Extraction of genomic DNA, PCR amplification, cloning and sequencing  DNA was extracted from 50 spores of the non-native F. mosseae isolates AZ225C and IMA1, and from the roots of plants grown in the trap cultures. Spores were crushed in microtubes on ice and the DNA extracted using 50 μl of extraction buffer (100 mM Tris-HCl, 100 mM NaCl, 2 mM MgCl2 and 2% Triton-X100, pH 8). Genomic DNA from the roots from the 48 trap culture pots was extracted from 100-mg fresh root samples, using the DNeasy® Plant Mini Kit (Qiagen, Germantown, MD, USA).

The choice of PCR primers was guided by an in silico analysis of the public database sequences FN547474-76, FN547482-84, and FN547486-93 of isolate BEG12 of F. mosseae (which had the same origin as IMA1), aiming for coverage of the variable regions in the nuclear rDNA cistron of the AM fungi. As nucleotide sequence polymorphism is highest at the 3’ end of the SSU rRNA gene, at ITS2, and at the 5’ end of the LSU rRNA gene, such partitions of the rDNA cistron were chosen for sequencing.

NS31/LSUGlom1 PCR amplicons were generated from 10 ng μl−1 genomic DNA in volumes of 25 μl with 2.5 U of HotStarTaq DNA Polymerase (Qiagen), 0.2 μM of each primer (NS31/LSUGlom1), 0.2 mM of each dNTP, 2 mM of MgCl2 and 1× reaction buffer, using touchdown thermal cycling on a PTC100 DNA Engine (MJ Research Inc., Waltham, MA, USA). The temperature profile was as follows: denaturation and enzyme activation at 95°C for 15 min, 20 cycles with denaturation at 95°C for 30 s, primer annealing for 1 min starting at 62°C and decreasing by 0.5°C per cycle to 52°C, extension at 72°C for 135 s and 20 cycles with denaturation at 95°C for 30 s, primer annealing at 52°C for 1 min, extension at 72°C for 135 s, and a final extension at 72°C for 10 min.

The QIAquick (Qiagen) purified PCR amplicons of DNA from the spore samples and Wizard® SV (Promega Corporation, Madison, WA, USA) gel-purified amplicons of DNA from the root samples were ligated into the pGem®-T Easy vector (Promega) to transform XL10-Gold® Ultracompetent Escherichia coli cells (Stratagene, La Jolla, CA, USA).

For the initial characterization of the natural AM fungal community, the roots of all trap-culture plants were screened using T-RFLP (Vandenkoornhuyse et al., 2003). Those Fabaceae root samples (= 6) that yielded T-RFLP profiles showing high relative abundance of AM fungal patterns affiliated to F. mosseae were selected for cloning and sequencing. Fifty recombinant clones per amplicon library were screened for RFLP in the c. 550-bp-long NS31-AM1 fragment (Helgason et al., 1998) on agarose gels (2% MetaPhor agarose (BMA, Rockland, ME, USA) stained with 0.5 μg ml−1 ethidium bromide for UV imaging), employing separate restriction digestions with the endonucleases HinfI and Hsp92II (Promega; Vandenkoornhuyse et al., 2003). The variable SSU, ITS2 and LSU fragments of 52 vector inserts (2.2 kb long) that gave rise to an RFLP pattern typical for F. mosseae (NS31/AM1 fragment of the SSU gene; HinfI: 282, 244, 24; Hsp92II: 290, 143, 117) were sequenced from plasmids, using the GenElute™ Plasmid Miniprep Kit (Sigma-Aldrich, St Louis, MO, USA). Sequencing reactions were set up with the vector primers SP6 and T7 using BigDye® version 3.1 chemistry on an 3730XL Genetic Analyser automated sequencer (Applied Biosystems, Foster City, CA, USA) at the High-Throughput Genomics Unit in Seattle (WA, USA).

Set-up of the AM fungal inoculation trial in the field

Nine experimental field plots of 5 × 3 m were ploughed to 10 cm depth and harrowed to 5 cm depth in September 2004, before inoculation with 10.5 kg of crude soil-based inoculum or mock inoculum (Pellegrino et al., 2011). The inoculation trial followed a completely randomized design with three inoculation treatments (non-native F. mosseae isolates AZ225C and IMA1 and mock inoculum) in three replicates. The forage legume Medicago sativa L. cv Messe was sown to a density of 5 g seeds per m2. The mycorrhizal inoculum potential (MIP) of the two F. mosseae inoculants, AZ225C and IMA1, and of the field site was 1.9 ± 0.4, 2.9 ± 0.5 and 0.19 ± 0.02 infection units per centimetre of root and spore densities were 3.5 ± 0.8, 3.9 ± 0.5 and 5.2 ± 0.6 spores per gram of dry soil, respectively.

Molecular genetic tracing of non-native F. mosseae inoculants in roots of M. sativa plants from the field and measurement of inoculation effects on plant growth

Three months and 2 yr after inoculation, 100 mg of fresh roots was collected from each replicate plot, and genomic DNA was extracted and amplified using NS31/LSUGlom1 primers, as described in the section ‘Extraction of genomic DNA, PCR amplification, cloning and sequencing’. The structure and composition of the AM fungal communities were determined using PCR-RFLP screening of clone libraries (NS31/AM1 primers and HinfI and Hsp92II restriction enzymes), and representative clones of each RFLP pattern were sequenced using T7 and SP6 vector primers, which yielded partial sequences of the SSU and LSU genes and complete ITS2.

Seven to 41 clones were screened by PCR-RFLP analysis per clone library (Tables S3, S4) and 126 plasmids were sequenced in total to cover all different RFLP patterns and phylogenetically identify the vector inserts that gave rise to an RFLP pattern typical for F. mosseae.

Root colonization levels were determined microscopically using the gridline intersect method (Giovannetti & Mosse, 1980). Three months after inoculation, the plant heights and the dry weights of the leaf, stem and root fractions were measured from six individuals of M. sativa, per experimental plot. In the first and second years after fungal inoculation, samples were taken from the root systems of three plants per replicate plot to assess mycorrhizal colonization. Furthermore, the cumulative aboveground dry matter and the shoot N and P concentrations (Jones et al., 1991) of two annual cuttings were measured in sampling areas of 1 m2.

Data analysis

The glomeromycotan affiliation of the sequences was verified in similarity searches against the international sequence databases, using BLASTn version 2.21 (Altschul et al., 1997). No chimeric sequences were detected among the 122 newly generated AM fungal SSU, the 53 ITS2 and the 53 LSU sequences using Chimera Check version 2.7 (Cole et al., 2003; Twenty newly generated partial SSU (≈ 550 bp), ITS2 (≈ 200 bp) and LSU (≈ 380 bp) sequences of the isolates AZ225C and IMA1 and of the native F. mosseae were aligned together with two sequences of each isolate from the public sequence databases and with 17 public sequences of isolate BEG12 of F. mosseae using MAFFT ( as implemented in SeaView version 4.2.5 (Gouy et al., 2010; Multiple alignments were first separately computed for each partition of the nuclear rDNA cistron, before concatenation and manual fine-editing in SeaView. One further multiple sequence alignment was derived from this, including 53 newly generated sequences of F. mosseae from field root sampled 3 months and 2 yr after inoculation. Another multiple sequence alignment (122 new sequences plus 16 published ones), using only partial SSU rRNA gene sequences of the natural AM fungal root community, was performed in order to assess community changes after inoculation.

Phylogenetic trees were inferred by Bayesian analysis in MrBayes version 3.2 (Ronquist et al., 2005;, using the General Time Reversible sequence evolutionary model. Details are given in Supporting Information, Methods S1. The phylograms were drawn in FigTree version 1.3.1 ( and edited in Adobe Illustrator CS4.

Phylogenetic discrimination of non-native AM fungal strains relied on branch support of ≥ 92% Bayesian posterior probability. Pairwise distances of different sequences, based on the Kimura-2-parameter sequence evolutionary model, were calculated in the ape package of R (Paradis et al., 2004). These complementary approaches of sequence analysis enabled a comparison of the ability to discriminate native from inoculated non-native strains of F. mosseae, based on pure nucleotide sequence similarity, as practised in molecular barcoding, and phylogenetic divergence, as used in phylogenetics.

Details of sampling effort curves are given in Methods S1. All new sequences were submitted to the EMBL nucleotide sequence database ( and are available under the accession numbers FR715045-51, FR715919-57, FR773835-53, FR715524-32, FR717152-7205, FR751247-1309, HE578017-49, and HE578143-61. The sequences affiliated to F. mosseae are listed in Table S1.

Possible influences of inoculation on crop yield and quality were assessed by the redundancy analysis (RDA) in Canoco for Windows version 4.5 (ter Braak & Šmilauer, 2002) using root colonization as a covariable. All 18 measurements of plant traits (first and second years after inoculation and three replicate plots per field treatment) were used in the same analysis by treating records from the same plot as a split-plot and permuting the data only within the nine whole plots (Monte Carlo test; Lepš & Šmilauer, 2003).

Mycorrhizal colonization and plant parameters after 3 months, 1 yr and 2 yr were analysed by one-way analysis of variance (ANOVA) with the inoculation treatment as fixed experimental factor. Data were loge- or arcsin-transformed when necessary to fulfil the assumptions for ANOVA. Post hoc Tukey significant difference tests were used for comparisons among treatments. To assess the effect of inoculation on AM fungal phylotype abundances, Kruskal–Wallis nonparametric tests, followed by Mann–Whitney U post hoc tests, were used because of unequal variances. Linear correlation analysis was used to identify relationships between AM fungal root colonization and parameters of plant performance. All univariate analyses were performed in spss version 17.0 (SPSS Inc., Chicago, IL, USA).


A novel DNA marker for molecular genetic discrimination among non-native and native strains of F. mosseae

The c. 2200-bp-long central stretch of the nuclear rDNA cistron, flanked by the PCR primers NS31 and LSUGlom1 (Supporting Infomation Fig. S1), provided sufficient phylogenetic resolution to discriminate the inoculated isolates from native strains of F. mosseae, even if only concatenated sequences of the variable 3’ end of the SSU rRNA gene, ITS2, and the variable 5’ end of the LSU rRNA gene were used (Fig. 1). The sequences (obtained from spores) of the inoculated isolates clustered distinctly separately from those of the native strains of F. mosseae, amplified from roots of trap plants (Fig. 1, Table S1). The c. 1130-bp-long concatenated sequences were sufficient to distinguish between those from the native AM fungal community and those from the inoculated isolates of F. mosseae (Table S2). The pairwise Kimura-2-parameter distances were on average 0.20% for the isolate AZ225C, 1.07% for the isolate IMA1-BEG12, and 0.54% for the native sequences with affiliations to F. mosseae from the experimental site. ITS2 showed the highest intraspecific sequence divergence, with mean values ranging between 1.17 and 4.83%, while the 5’ end of the LSU rRNA gene and the 3’ end of the SSU rRNA gene showed mean values ranging between 0 and 0.54% and 0 and 0.49%, respectively (Fig. S1, Table S2). This was expected based on the overall nucleotide variability across the rDNA cistron and AM fungal taxa. Character-based phylogenetic analysis (MrBayes) resolved distinctive phylogenetic sequence clusters for the inoculated isolates and the native strains of F. mosseae (Fig. 1). Near-exhaustive characterization of the ribotypes of the two inoculants and the native strains of F. mosseae at the experimental site (Fig. S2a), together with the discriminative power of the rDNA marker, gives confidence that the two isolates of F. mosseae can accurately and successfully be traced in the field.

Figure 1.

Bayesian consensus tree inferred from nuclear ribosomal rDNA sequences of Funneliformis mosseae, originating from native strains and the non-native inoculated strains IMA1 and AZ225C. Partial sequences of the entire nuclear ribosomal rDNA cistron (≈ 1130 aligned sites) were used, composed of the 3’ end of the small subunit ribosomal RNA gene (SSU; ≈ 550 bp; NS31/AM1 fragment), internal transcribed spacer 2 (ITS2; ≈ 200 bp) and the 5’ end of the large subunit ribosomal RNA gene (LSU; ≈ 380 bp; LSUGlom1), which were concatenated. The concatenated sequences FR717169, FR717187 and FR717105 of isolate BEG75 of Rhizophagus irregularis (N. C. Schenck & G. S. Sm.) C. Walker & A. Schüβler was used as an outgroup to root the tree. Branch support values are posterior probabilities. Clades formed by sequences of native and inoculated F. mosseae strains are labelled with brackets. (a) and (b) indicate different clusters of F. mosseae native sequences. Newly generated sequences are highlighted in boldface, and their accession numbers are prefixed by the isolate/clone identifier. Sequences from spores of the non-native strains of F. mosseae are followed by an oval symbol; sequences from colonized roots of trap plants of the native F. mosseae strains are followed by a tick symbol.

Some reference sequences of isolate BEG12 of F. mosseae from the public sequence databases clustered together with sequences of isolate IMA1 (Fig. 1), as could be expected, based on these isolates’ common origin. In addition, the partial sequences of the nuclear rDNA cistron of both these isolates clustered into two phylogenetic subclusters, as did the sequences of F. mosseae of the strains native to the experimental site. This is strong evidence that there are at least two major types of divergent nuclear rDNA loci within the genomes of F. mosseae strains.

AM fungal communities in roots of field-inoculated M. sativa plants as determined by RFLP fingerprinting and sequencing

Semi-nested PCR-RFLP analysis using the PCR primer pair NS31/AM1, which target the 3’ end of the SSU rRNA gene, was used to screen clone libraries from PCR amplicons of nine crude DNA extracts of the field root samples, collected 3 months post inoculation, and of six extracts from the sampling 2 yr post inoculation. From roots of the first sampling, 211 of a total of 317 clones tested positive in the RFLP analyses for having inserts of putative AM fungal origin; from the second sampling there were 141 out of 637 clones with possible AM fungal inserts. In total, 20 different RFLP patterns were found, which were verified by sequencing 126 recombinant plasmids with the SP6 and T7 primers. Except for 12 sequences, corresponding to five RFLP patterns, all 114 remaining new sequences were of glomeromycotan origin (Tables 1, S3, S4). The non-target sequences were of basidiomycotan and ascomycotan origin.

Table 1.   Types of nuclear ribosomal small subunit RNA gene sequence of arbuscular mycorrhizal (AM) fungi, as characterized by cloning and sequencing and restriction fragment length polymorphism (RFLP) analysis of the approx. 550-bp-long fragment flanked by the NS31-AM1 PCR primer pair, which was re-amplified from inserts of recombinant bacterial clones
Sequence phylotypes*RFLP typesFragment sizes (bp)% identityPutative taxonomic affiliation
HinfI Hsp92II
  1. The analysed mycorrhizal root samples of Medicago sativa plants were collected 3 months and 2 yr after inoculation with two non-native isolates, AZ225C and IMA1, of the AM fungus Funneliformis mosseae and a mock-inoculum (control).

  2. *Names denote the most similar AM fungal species of sequenced clones: Rirreg_Pi, Rhizophagus irregularis-Pisa; Rintra_Pi, Rhizophagus intraradices-Pisa; Fmos_Pi, Funneliformis mosseae-Pisa; Glo_P1, glomeromycete-Pisa1; Glo_Pi2, glomeromycete-Pisa2.

  3. **Representative sequences have been deposited in EMBL (accession numbers FR715048-51, FR715919-31, FR717165-68, FR751247-59, FR751286-1309, FR773835-53, FR715045, FR715524, FR715527, FR715530, FR717152, FR717153, HE578017-49 and HE578143-49).

Rirreg_Pi**RFLP1384, 142, 24291, 142, 117FJ009618 (99) Rhizophagus irregularis
RFLP3408, 142291, 142, 117FJ009618 (99) Rhizophagus irregularis
RFLP11244, 142, 140, 24290, 143, 117FJ009618 (99) Rhizophagus irregularis
RFLP14383, 141, 25433, 116FJ009618 (99) Rhizophagus irregularis
Rintra_PiRFLP2334, 142, 49, 25291, 142, 117AJ536822 (99) Rhizophagus intraradices
RFLP7251, 142, 81, 49, 27293, 140, 117FJ009604 (97) Rhizophagus intraradices
RFLP9173, 161, 142, 49, 25291, 142, 117AJ536822 (99) Rhizophagus intraradices
RFLP13382, 141, 25290, 258EU232660 (99) Rhizophagus intraradices
RFLP15334, 141, 49, 25291, 258FJ009604 (97) Rhizophagus intraradices
Fmos_PiRFLP8282, 244, 24290, 143, 117AY635833 (99) Funneliformis mosseae
RFLP10282, 268290, 143, 117AY635833 (99) Funneliformis mosseae
RFLP11244, 142, 140, 24290, 143, 117AY635833 (99) Funneliformis mosseae
RFLP12244, 234, 47, 25290, 143, 117AY635833 (99) Funneliformis mosseae
Glo_Pi1RFLP4244, 190, 90, 26292, 258AM946850 (100)Uncultured glomeromycete
RFLP5244, 98, 92, 90, 26292, 258EU123438 (99)Uncultured glomeromycete
Glo_Pi2RFLP6334, 191, 25291, 259EF177525 (98)Uncultured glomeromycete

Root samples taken 3 months and 2 yr post inoculation yielded highly similar sequences, which were either phylogenetically affiliated with the species F. mosseae, Rhizophagus intraradices and Rhizophagus irregularis, or represented two additional phylotypes (Glo_Pi1 and Glo_Pi2), only known as environmental sequences from the public databases (Fig. 2, Table 1). PCR-RFLP fingerprinting expedited a targeted retrieval of sequences that could be ascribed to the focal species F. mosseae, highlighting the advantage of pre-screening clone libraries by RFLP analysis before plasmid sequencing. Despite a partial compositional similarity of the AM fungal communities in M. sativa roots 3 months and 2 yr post inoculation, community structure differed significantly. Whereas AM fungal ribotype richness in the mock inoculation control still increased steadily beyond 60 fingerprinted clones, saturation was reached in the two inoculated experimental plots for the root samples taken 3 months post inoculation (Fig. S2b).

Figure 2.

Bayesian consensus tree of glomeromycotan sequences derived from roots of Medicago sativa, sampled 3 months after inoculation with two non-native Funneliformis mosseae isolates, AZ225C and IMA1, and a mock inoculum (control). The analysis is based on partial nuclear small subunit ribosomal RNA gene sequences (SSU; ≈ 550 bp; NS31/AM1 fragment), and the tree is rooted with a reference sequence of Geosiphon pyriformis (X86686). Branch lengths of the root and of the reference sequences of Acaulospora rugosa WV935 (Z14005) and Scutellospora calospora BEG32 (AJ306446) are shortened, so as not to compromise the display of the phylogram. Branch support values are posterior probabilities. Clades of sequences affiliated to the genera Rhizophagus and Funneliformis and additional taxa are indicated with brackets on the right: Rirreg_Pi, R. irregularis-Pisa; Rintra_Pi, R. intraradices-Pisa; Fmos_Pi, F. mosseae-Pisa; Glo_Pi1, glomeromycete-Pisa1; Glo_Pi2, glomeromycete-Pisa2. Sequences obtained in the present study are shown in bold and their accession numbers are prefixed with the treatment/field plot clone identifiers and RFLP type. Sequences from spores of the non-native F. mosseae isolates are followed by an oval symbol; those from roots sampled 3 months or 2 yr after inoculation are followed by an open triangle, or filled rectangle, respectively.

Three months post inoculation, only two sequence types (Fmos_Pi and Rintra_Pi) and one sequence type (Fmos_Pi) could be found in the roots collected from the plots to which IMA1 and AZ225C had been applied, respectively (Table 2). This contrasted with four sequence types (Rirreg_Pi, Rintra_Pi, Glo_Pi1 and Glo_Pi2) that were recovered from roots sampled in the mock-inoculated plots. The sole occurrence of F. mosseae sequences in roots inoculated with the American isolate AZ225C clearly shows the dominance and possibly rapid colonization ability of this foreign mycorrhizal fungal strain. The dominance of the Fmos_Pi RFLP/sequence-phylotype and lack of it in the control inoculation treatment (Table 2) suggest strongly that both F. mosseae inoculants established successfully as root symbionts of M. sativa, despite competition from members of the natural AM fungal community, particularly strains affiliated with R. intraradices.

Table 2.   Results of Kruskal–Wallis tests comparing the relative abundances of arbuscular mycorrhizal (AM) fungal sequence phylotypes in clone libraries of PCR amplicons from colonized roots of Medicago sativa plants collected in the field 3 months after inoculation with the two non-native isolates AZ225C and IMA1 of Funneliformis mosseae and a mock-inoculum (control)
Sequence phylotypes*Relative abundance in the roots** (%)χ2 value P-value
  1. Clone libraries were screened by combined RFLP and sequence analyses.

  2. *Rirreg_Pi, Rhizophagus irregularis-Pisa; Rintra_Pi, Rhizophagus intraradices-Pisa; Fmos_Pi, Funneliformis mosseae-Pisa; Glo_P1, glomeromycete-Pisa1; Glo_Pi2, glomeromycete-Pisa2.

  3. **Mean values of three field replicates of each treatment.

  4. ***Values not followed by the same letter are significantly different (Z ≥ − 2.087; P = 0.037) according to the post hoc Mann–Whitney U test.

Fmos_Pi***0.0 a100.0 c48.4 b7.7140.021

However, 2 yr post inoculation, the relative proportion of ribotypes of F. mosseae decreased in favour for members of the native AM fungal community. Under inoculation of the isolate AZ225C, the relative abundances of clones of F. mosseae dropped from 100% to only 16.3%, with R. intraradices and R. irregularis accounting for 37.2% and 46.5%, respectively. In addition, the roots collected from the plots inoculated with IMA1 were colonized by Rirreg_Pi (42.6%,), Glo_Pi1 (38.2%), Rintra_Pi (17.6%), and Glo_Pi2 (1.5%), an AM fungal community considerably richer than that 3 months after inoculation and lacking the inoculant.

Tracing non-native and native strains of F. mosseae in the field

Non-native and native strains of F. mosseae were successfully traced in roots of M. sativa from the field by character-based sequence analysis. Bayesian phylogenetic inferences showed a clearly separate clustering of the sequences of the two inoculants and those of F. mosseae strains from the resident community (Fig. 3), both 3 months and 2 yr post inoculation. This shows that the polymorphic regions of the SSU and LSU rRNA genes together with ITS2 provided sufficient phylogenetic information for resolution at the level of F. mosseae strains. At the first sampling, both inoculated F. mosseae strains were retrieved, while only isolate AZ225C of F. mosseae was successfully traced as an active root colonizer, 2 yr post inoculation (Fig. 3). No sequence similar to that of the IMA1 isolate was found at the second sampling, a finding that could be explained by low fungal abundance as well as an insufficient sampling effort, as growth stimulation and improvement of mineral nutrition persisted in the M. sativa plants (Table 3; see subsection below).

Figure 3.

Bayesian consensus tree of native Funneliformis mosseae sequences originating from trap cultures before inoculation and of sequences recovered from colonized roots of Medicago sativa, sampled in the field 3 months and 2 yr post inoculation. For reference, sequences generated from genomic DNA extracts of spores of the inoculated, non-native isolates AZ225C and IMA1 of F. mosseae are also included. Concatenated sequences (≈ 1130 aligned sites) of partitions of the entire nuclear ribosomal DNA cistron were used, composed of a fragment of the 3’ end of the ribosomal small subunit RNA gene (SSU; ≈ 550 bp; delimited by the PCR primers NS31/AM1), internal transcribed spacer 2 (ITS2; ≈ 200 bp) and a fragment of the 5’ end of the ribosomal large subunit RNA gene (LSU; ≈ 380 bp; delimited by the PCR primer LSUGlom1 at the 3’ end). The tree is rooted with the concatenated sequences FR717169, FR717187 and FR717105 of the isolate BEG75 of Rhizophagus irregularis (N. C. Schenck & G. S. Sm.) C. Walker & A. Schüβler. Branch support values are posterior probabilities. The clades formed by the sequences of the non-native isolates of F. mosseae (AZ225C and IMA1) and native strains of F. mosseae are highlighted with brackets on the right. (a) and (b) indicate different clusters of native F. mosseae sequences. Sequences obtained in the present study are shown in bold and their accession numbers are prefixed by the isolate/clone identifier. Sequences from spores of the inoculated isolates are labelled with an oval symbol; those recovered from roots of trap plants, grown in soil collected before inoculation, with a tick symbol. Labels of sequences from root samples of M. sativa collected 3 months post inoculation are followed by an open triangle and labels of sequences from root samples taken after 2 yr are followed by a filled rectangle.

Table 3.   Shoot dry weights and shoot nitrogen (N) and phosphorus (P) concentrations of Medicago sativa plants 1 and 2 yr after inoculation with two non-native isolates of the arbuscular mycorrhizal (AM) fungus Funneliformis mosseae, AZ225C and IMA1, and a mock-inoculum (control)
InoculumShoot dry weight (g m−2)N concentration (%)P concentration (‰)
  1. *Values are means ± SE of samples from three replicate plots per inoculation treatment; values not followed by the same letter in the same column are significantly different according to post hoc Tukey–Kramer honestly significant difference tests (< 0.05).

  2. **Cumulative values of two annual cuttings.

After 1 yr
 Control377.4 ± 26.7*,** a2.11 ± 0.11 a2.53 ± 0.22 a
 AZ225C768.8 ± 110.4 b4.15 ± 0.02 b3.80 ± 0.05 b
 IMA1589.1 ± 77.5 ab4.34 ± 0.08 b3.80 ± 0.10 b
After 2 yr
 Control826.0 ± 63.7 x3.43 ± 0.13 x1.32 ± 0.26 x
 AZ225C1371.9 ± 67.9 y5.10 ± 0.48 y2.77 ± 0.06 y
 IMA11137.6 ± 73.7 y4.12 ± 0.30 xy2.35 ± 0.25 y

Long-term effects of non-native AM fungal inoculants on crop plant performance

Inoculation with the isolates AZ225C and IMA1 of F. mosseae increased overall mycorrhizal colonization significantly, compared with the mock-inoculation, 3 months after inoculation (F2,9 = 37.9, = 0.001; Table S5). However, root colonization did not differ between the two treatments. Whereas the two non-native F. mosseae inoculants consistently improved crop yield and mineral nutrient contents, compared with the mock-inoculation treatment (Table 3), differences in symbiotic effects between the isolates IMA1 and AZ225C were slight (data not shown).

Colonization of roots by AM fungi remained significantly higher in the plots to which inoculants of F. mosseae had been added 1 and 2 yr post inoculation (Table S5). Yield increases were also sustained, with 104 and 36% higher cumulative annual aboveground biomass after the first cropping season, and 66 and 38% higher biomass after the second cropping season for the inoculants AZ225C and IMA1, respectively (Table 3). Increased crop biomass production was accompanied by higher shoot N and P concentrations (Table 3), again suggestive of positive effects of inoculation with the two non-native F. mosseae isolates. Shoot biomass and N and P concentrations, measures of yield quantity and quality, were all positively correlated with total AM fungal root colonization (data not shown), suggestive of direct positive symbiotic effects of the inoculated non-native F. mosseae strains.

The RDA analysis of the growth-physiological data over the two cropping seasons showed that 45.2% of total variance in yield quantity and quality could be explained by inoculation, although the two isolates of F. mosseae did not differ in their benefits for the host plant (Fig. 4).

Figure 4.

Ordination biplot of a redundancy analysis, carried out on the growth-physiological parameters of the inoculated Medicago sativa crop plants, harvested in the field 1 and 2 yr after inoculation with the two isolates of Funneliformis mosseae, AZ225C and IMA1, and a mock inoculum (control). The inoculation treatment was used as the explanatory variable, depicted using bold arrows, and the plant growth-physiological parameters were used as dependent variables, which are depicted using dotted arrows. SDM, shoot dry matter; LDM, leaf dry matter; StDM, stem dry matter; [SN], shoot N concentration; SNC, shoot N content; [SP], shoot P concentration; SPC, shoot P content. The first and second axes explain 45.2% of total variance.


Molecular phylogenetic discrimination between non-native inoculants and native F. mosseae isolates based on a novel marker, covering the central polymorphic regions of the nuclear rDNA cistron

The PCR primers NS31 and LSUGlom1 were successfully combined in PCR amplifications from genomic DNA extracted from multiple spores and mycorrhizal roots of M. sativa from the field. The phylogenetic information contained in the polymorphic 3’ end of the nuclear SSU rRNA gene, ITS2, and the 5’ end of the LSU rRNA gene proved sufficient for phylogenetic discrimination among non-native and native strains of F. mosseae, despite the well-known intra-individual rDNA sequence polymorphism and partially overlapping ranges of pairwise sequence distances.

The partial SSU sequence has been widely used for environmental studies at higher taxonomic levels, from genus to order (Öpik et al., 2006). The ITS region, highly variable among different species and isolates and also polymorphic within single individuals of AM fungi, has also previously been used as a molecular genetic marker in environmental studies (Redecker, 2000; Renker et al., 2003). However, the focus solely on the highly polymorphic ITS region (Sanders et al., 1995; Lloyd-McGilp et al., 1996; Lanfranco et al., 1999; Redecker, 2000) may have been the reason why markers of the nuclear rDNA cistron were supposed to be unsuitable for environmental molecular genetic tracing in AM fungi. Recent work using the SSU rRNA gene, the ITS region, and the LSU rRNA gene in combination clearly shows that species- or even isolate-level resolution is reached (Krüger et al., 2009; Stockinger et al., 2010). In fact, there is little reason left for doubt about the appropriateness of long nuclear rDNA sequence markers to discriminate phylogenetically among closely related taxa of AM fungi.

Nevertheless, the rDNA marker developed by Krüger et al. (2009) has the clear disadvantage that it does not cover the widely employed variable region V4 of the SSU rRNA gene (Öpik et al., 2010). This was the reason for our decision to combine the PCR primers NS31 and LSUGlom1, which extends the c. 1500-bp fragment covered by the nested PCR approach developed by Krüger et al. (2009) upstream to an overall length of c. 2200 bp. Use of the universal fungal NS31 primer, instead of that suggested by Krüger et al. (2009), results in inclusion of the variable and most widely sequenced V4 region of the nuclear SSU rRNA gene. For work on colonized roots, it was thus possible to screen the AM fungal communities by PCR-(T)-RFLP fingerprinting, as done previously (Helgason et al., 1998; Vandenkoornhuyse et al., 2003), relying on the c. 550-bp-long relatively conserved fragment of the SSU rRNA gene. One RFLP pattern, namely RFLP8, corresponded to F. mosseae, which made targeted sequencing of environmental clone libraries possible, reducing manual and financial expenses of environmental molecular genetic tracing.

Consistent with Stockinger et al. (2010), we found that combining the highly polymorphic ITS2 region with variable SSU and LSU regions paved the way for phylogenetic species- and strain-level resolution. ITS2 and the 5’ end of the LSU rRNA gene provide sufficient variation to achieve resolution among closely related AM fungal strains. Although we did not sequence the entire c. 2200-bp-long partial rDNA cistron, amplifiable using the NS31/LSUGlom1 primer pair, the concatenated 3’ end of the SSU and 5’ end of the LSU rRNA together with ITS2 contained sufficient phylogenetic signal to discriminate among inoculated and native strains of F. mosseae at our field site. Of course, molecular genetic tracing of our inoculants may have been facilitated at our field site as a result of the lack of more closely related F. mosseae strains, because the applied inocula, IMA1 and AZ225C, had distant origins in the UK and the USA. Moreover, it is not known whether the NS31-LSUGlom1 marker shows sufficient phylogenetic divergence also among strains of other AM fungal species and whether it could also be used on DNA extracts from soil with many more different nontarget organisms. Obviously, future studies should aim to sequence the entire c. 2200-bp-long fragment, or even try to combine the NS31 PCR primer with the reverse primers developed by Krüger et al. (2009) to also include the variable domain D2 of the LSU rRNA gene.

Despite efforts to develop mitochondrial markers for environmental studies (Raab et al., 2005; Börstler et al., 2008, 2010), which would be advantageous because they lack intra-individual sequence polymorphism, for the time being, nuclear rDNA markers remain the only viable choice for reliable environmental tracing, because of their phylogenetic resolution and sufficient reference sequences across a broad systematic range of fungi. Intra-individual subclustering of rDNA sequences, as observed for the inoculant IMA1 (Figs 1, 3), has been observed previously (Clapp et al., 2002; Stukenbrock & Rosendahl, 2005) and indicates the existence of different rDNA loci with divergent sequences of the rDNA cistron.

AM fungal communities in roots of M. sativa 3 months and 2 yr after inoculation with two non-native isolates of F. mosseae

Successful establishment of inoculant AZ225C was demonstrated by the fact that it was the dominant colonizer of roots of M. sativa after 3 months, while inoculant IMA1 accounted for half of the symbiotic AM fungal community at this stage. This clear dominance of the inoculants in the roots of the 3-month-old M. sativa plants had the consequence that fewer native AM taxa could colonize the same root systems simultaneously and thus that roots sampled from the mock-inoculated experimental field plots hosted taxonomically richer AM fungal communities. The most dominant species in the native AM fungal community at the site of the inoculation trial was R. intraradices, which was a good competitor. At the second sampling 2 yr post inoculation, only isolate AZ225C could be successfully traced, although it had massively declined from 100 to 16%, confirming its survival as a root symbiont. IMA1 could not be found after 2 yr, despite sustained stimulatory effects on crop plant performance. Our findings suggest that both isolates AZ225C and IMA1 of F. mosseae represent fast and prompt root colonizers, at least relative to the resident AM fungi colonizing M. sativa at this experimental field site. However, the American isolate AZ225C was more competitive than the British isolate IMA1, showing longer persistence as a root symbiont, possibly as a result of greater functional complementarity to European strains of F. mosseae. It remains to be seen whether the non-native strains of F. mosseae will persist as rare AM fungal taxa and whether they will hybridize with native strains (Croll et al., 2009), which would obviously complicate future molecular genetic traceability and determination of symbiotic growth effects by inoculation.

Very little is still known about successful introductions of foreign AM fungi to natural communities and their possible effects in symbiosis with plants. This is mainly because tools have been lacking for reliable molecular genetic tracing at the strain level. The first attempt to track an AM fungal inoculant was made by Tonin et al. (2001) in a pot trial in the glasshouse, using T-RFLP. Farmer et al. (2007), Douds et al. (2011), Antunes et al. (2009), Koch et al. (2011), and Sýkorováet al. (2012) all attempted to trace AM fungal inoculants among members of complex natural AM fungal communities, with variable success. Whereas, for instance, Antunes et al. (2009) with a strain of R. intraradices could only detect minor inoculation effects on the richness and structure of the natural AM fungal community in microcosms, a glasshouse study in pots (Koch et al., 2011) and a field inoculation trial (Douds et al., 2011) with other strains of the same AM fungal species found that, at least in the short term, inoculants disturbed natural AM fungal communities. Mummey et al. (2009) also reported observations that natural AM fungal communities had been altered after AM fungal inoculation. Using sequencing to trace the inoculant, Ceccarelli et al. (2010) recovered sequence types that were related to those of the inoculant from roots of pre-inoculated artichokes (Cynara cardunculus L. var. scolymus) 24 months after transplantation to the field.

Relationship between AM fungal root colonization and yield of M. sativa in response to inoculation with two non-native isolates of F. mosseae in the field

Microscopic measurements 3 months after inoculation showed that field inoculation with the two foreign F. mosseae inoculants, AZ225C and IMA1, produced seven- to ten-fold higher root colonization, as compared with the mock-inoculation control. Particularly, shoot biomass and N and P concentrations appeared to have been positively affected by the inoculation with the non-native isolates of F. mosseae, as evident from the multivariate analysis of all growth physiological parameters of the crop legume. Root colonization by AM fungi remained elevated over the two following cropping seasons in the inoculated M. sativa plants and was positively correlated with shoot biomass production and N and P concentrations as well as contents. Growth promotion and improved mineral nutrition had been previously observed for mycorrhizal legume plants in glasshouse and field trials (Azcón-Aguilar & Barea, 1981; Monzon & Azcón, 1996; Smith et al., 2000; Avio et al., 2006; Mäder et al., 2011), which could also arise from additional synergistic interactions with native N2-fixing rhizobia (Azcón-Aguilar & Barea, 1981; Mortimer et al., 2009).

Growth stimulation by the same two isolates of F. mosseae was previously observed for T. alexandrinum and Z. mays, grown in the same soil (Pellegrino et al., 2011). Effects on symbiotic performance differed little between the two inoculants of F. mosseae, despite the fact that IMA1, compared with AZ225C, was less competitive against members of the natural AM fungal community, as evident from its shorter persistence as an active root symbiont. Only shoot N content and stem dry matter were significantly higher in AZ225C-inoculated as compared with IMA1-inoculated plants 2 yr after inoculation. A greater difference in the symbiotic effects of the two inoculants might have been expected, given their previously determined morphological, molecular, physiological, and phylogenetic distinctiveness (Giovannetti et al., 2003; Avio et al., 2009; Bedini et al., 2009; Pellegrino et al., 2011) and the generally high functional diversity among strains of the same AM fungal species (Munkvold et al., 2004; Koch et al., 2006; Angelard & Sanders, 2011). One possible explanation for more similar symbiotic effects of the two non-native inoculants is that the members of the natural community might have had their effects on host plant performance too, attenuating their influence. Strong isolate- or strain-specific symbiotic effects could thus be something that is preferentially detected in only pairwise symbiotic associations and under highly controlled growth conditions.


This field inoculation trial is one of the first to show successful establishment and persistence of an AM fungal species as root symbionts of a crop inoculated in the field. The crop was, moreover, stimulated in its growth and mineral nutrient uptake up to 2 yr post inoculation. The two foreign isolates of F. mosseae that were used as inocula appear to have successfully competed with members of the natural AM fungal community as root colonizers. However, the exclusion of native AM fungi from colonizing roots raises concerns about the bio-invasion potential of non-native inoculants that will need further study in light of serious concerns about possible biodiversity losses and homogenization as a result of anthropogenic translocations of biota between biogeographic regions.

At the same time, this study showed the molecular genetic traceability of individual strains of F. mosseae under field settings, where conspecific native strains co-occur. The partial sequence of the nuclear rDNA cistron, flanked by the PCR primers NS31 and LSUGlom1, was found to be suitable for combined community-level RFLP fingerprinting and species-level phylogenetic analysis of PCR amplicon libraries.


This work represents part of E. P.’s PhD thesis project, which was jointly funded by Scuola Superiore Sant’Anna and the University of Pisa. The molecular genetic analyses were enabled by E. P.’s visit to the Biology Department of the University of York, UK. Special thanks go to Prof. Petr Šmilauer for statistical advice, and Prof. David Swofford, Dr Michael Matschiner, and Dr Anna Maria Fiore-Donno for assistance with phylogenetic analyses. We are also grateful for the technical support of the staff of the Rottaia Experimental Station of the University of Pisa in setting up and managing the field experiment.