New insights into the mycorrhizal Rhizoscyphus ericae aggregate: spatial structure and co-colonization of ectomycorrhizal and ericoid roots


  • Gwen-Aëlle Grelet,

    1. Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, AB24 3UU, UK
    2. The Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen, AB15 8QH, UK
    Search for more papers by this author
  • David Johnson,

    1. Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, AB24 3UU, UK
    Search for more papers by this author
  • Trude Vrålstad,

    1. Department of Biology, University of Oslo, Box 1066, Blindern, N–0316 Oslo, Norway
    Search for more papers by this author
  • Ian J. Alexander,

    1. Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, AB24 3UU, UK
    Search for more papers by this author
  • Ian C. Anderson

    1. The Macaulay Land Use Research Institute, Craigiebuckler, Aberdeen, AB15 8QH, UK
    2. Centre for Plants and the Environment, University of Western Sydney, Locked Bag 1797, Penrith South DC, NSW 1797, Australia
    Search for more papers by this author

Author for correspondence:
Gwen-Aëlle Grelet
Tel: +64 3 3219999


  • Fungi in the Rhizoscyphus ericae aggregate have been recovered from the roots of co-occurring ericaceous shrubs and ectomycorrhizal trees. However, to date, there is no evidence that the same individual genotypes colonize both hosts, and no information on the extent of the mycelial networks that might form.
  • Using spatially explicit core sampling, we isolated fungi from neighbouring Pinus sylvestris (ectomycorrhizal) and Vaccinium vitis-idaea (ericoid mycorrhizal) roots and applied intersimple sequence repeat (ISSR) typing to assess the occurrence and extent of shared genets.
  • Most isolates were identified as Meliniomyces variabilis, and isolates with identical ISSR profiles were obtained from neighbouring ericoid and ectomycorrhizal roots on a number of occasions. However, genet sizes were small (< 13 cm), and several genets were found in a single soil core. Genetic relatedness was independent of spatial separation at the scales investigated (< 43 m) and M. variabilis populations from sites 20 km apart were genetically indistinguishable.
  • We conclude that individual genets of M. variabilis can simultaneously colonize Scots pine and Vaccinium roots, but there is no evidence for the formation of large mycelial networks. Our data also suggest significant genotypic overlap between widely separated populations of this ubiquitous root-associated fungus.


Ericaceous plants are the dominant understory vegetation of Boreal and Northern temperate forests. Most of these ericaceous species form ericoid mycorrhizas, while the overstorey trees in these ecosystems form ectomycorrhizas. The fungi involved in the formation of ectomycorrhizas and ericoid mycorrhizas are generally thought to be different (Smith & Read, 2008). However, recent molecular studies have shown that the DNA of many fungi commonly classified as ectomycorrhizal (ECM) can be detected in roots of ericoid mycorrhizal (ERM) shrubs (Bougoure et al., 2007) and the DNA of supposedly ERM fungi can be amplified from tree ectomycorrhizas (Vrålstad et al., 2002a; Bergero et al., 2000; Collier & Bidartondo, 2009). Ascomycetous fungi in the Rhizoscyphus ericae aggregate (syn. Hymenoscyphus ericae) are particularly important in this respect. They have been reported to colonize up to 38% of ECM pine root tips (Heinonsalo et al., 2007) and are always found in hair roots of Northern hemisphere ericaceous plants, whether detected by molecular methods or by direct culturing (Bougoure et al., 2007; Allen et al., 2003).

The R. ericae aggregate contains four main clades (Vrålstad et al., 2000, 2002a) that are genetically closely related (at least 84% sequence similarity in the internal transcribed spacer (ITS) region of the nuclear ribosomal gene cluster). The structure of the aggregate was further refined by Hambleton & Sigler (2005) who raised three new species mapping onto the clades initially described by Vrålstad et al. (2000): Meliniomyces variabilis and Meliniomyces vraolstadiae, each forming a clade of their own, and Meliniomyces bicolor, forming a subclade closely related to Cadophora finlandica (Wang & Wilcox, 1987). Meliniomyces vraolstadiae has so far only been found to form ectomycorrhizas, or to be nonmycorrhizal (Vrålstad et al., 2002b; Hambleton & Sigler, 2005) while M. variabilis, C. finlandica and M. bicolor occur in both ERM and ECM roots (Hambleton & Sigler, 2005; Bougoure et al., 2007) in nature. Meliniomyces bicolor forms both ericoid mycorrhizas and ectomycorrhizas in vitro, sometimes simultaneously, with either a beneficial or neutral effect on host plant growth (Vrålstad et al., 2002b; Villarreal-Ruiz et al., 2004). The ectomycorrhizas formed by M. bicolor correspond to the morphotype Piceirhiza bicolorata sensu Agerer (1987–2002) (Vrålstad et al., 2000, 2002a).

Isolates of M. variabilis and M. bicolor obtained from P. bicolorata ectomycorrhizas form typical ericoid mycorrhizal structures and engage in reciprocal transfer of carbon and nitrogen with Vaccinium vitis-idaea (Grelet et al., 2009a). This suggests that Meliniomyces spp. isolates obtained from ectomycorrhizas can function as ERM fungi, and raises the possibility that mycorrhizal connections exist between ericoid and ectomycorrhizal plants in boreal and temperate forests. To date, however, field evidence of such connections is restricted to the detection of identical rDNA-ITS sequences in ericoid and ectomycorrhizal roots (Vrålstad et al., 2002a). Because ITS sequences do not discriminate below species level, this approach cannot provide evidence for the simultaneous colonization of ericoid and ectomycorrhizal roots by the same individual fungal genet. In addition, while we know that genets of some ectomycorrhizal fungi can extend for tens of metres and can colonize roots of several plants simultaneously (e.g. Beiler et al., 2010), we have no idea of the size of individual genets of Meliniomyces spp. in the field. The only study reporting on the spatial distribution of ericoid genotypes is restricted to individual plant root systems (Midgley et al., 2004). Therefore, the extent of individual ericoid mycorrhizal fungal genotypes, and whether they can simultaneously colonize several plants, are unknown.

In this study we used intersimple sequence repeat (ISSR) typing to establish whether identical Meliniomyces spp. genotypes occur simultaneously in P. bicolorata ectomycorrhizas and neighbouring ericaceous hair roots in the field. By taking samples in a spatially explicit way at three locations across two different forests, we were able to determine the extent of individual genotypes, and thus their potential for forming large mycelial networks, and to gain insight into the population structure of Meliniomyces spp in the field.

Materials and Methods

Root sampling and isolation of fungal isolates

Soil cores (120 mm deep × 45 mm diameter) were collected over a period of 6 wk in autumn 2006, in two Scots pine forests in north-east Scotland with a Vaccinium vitis-idaea understorey. There were no other woody plants present in the plots. The first was a mature 100–250-yr-old native stand at Glen Tanar (57°02′ N, 2°52′ W) while the second was a 60-yr rotation plantation at Hill of Fare (57°07′ 03″ N, 2°26′ W). The two forests were 20 km apart, separated by a major river and hills up to 450 m above seal level (a.s.l.).

A 10.5 × 12 m plot was set up at Hill of Fare, and cores were taken at 1.5 m grid intersects, giving 56 cores in total. At Glen Tanar, two composite sampling grids were set up in stands 2 km apart. Each composite grid comprised four subgrids. Within each subgrid, nine cores were taken at 50-cm intervals giving 72 cores in total. The spatial structure was designed to give intersampling distances varying between 50 cm and 2.8 m within a subgrid, whereas intersampling distances varied from 5 m to 37 m between the different subgrids (Fig. 1). The location and size of the sampling grids were constrained by site topology (i.e. area where the only woody plants present were P. sylvestris and V. vitis-idaea) and the need to optimize both number of sampling points and range of distances between points. Ten additional cores were also taken outside the sampling grids across both Hill of Fare and Glen Tanar forests.

Figure 1.

 Sampling grids in the forests at Glen Tanar and Hill of Fare, UK. Soil cores where at least one Piceirhiza bicolorata tip was found are represented by closed circles, and those in which none were found are represented by tinted circles.

After collection, cores were sealed in polypropylene bags and stored at 4°C for a maximum of 72 h until root sorting. Any ECM root tips with a morphology consistent with P. bicolorata ectomycorrhizas (Agerer, 1987–2002) were collected from each core, cleaned from all soil organic debris, placed in sterile 1.5 ml Eppendorf tubes and surface sterilised by successive gentle vortexing in Tween 20 at 0.2% (v : v) for 1 min, sterile H2O (30 s), H2O2 at 30% (30 s) followed by three further rinses in sterile water (30 s each). Individual P. bicolorata root tips were then plated on Modified Melin-Norkrans agar (MMN) media containing (NH4)2HPO4 (500 mg), KH2PO4 (300 mg), MgSO4.7H2O (140 mg), CaCl2.6H2O (50 mg), NaCl (25 mg), thiamine (0.01 mg), FeNaEDTA (12.5 mg), ZnSO4·7H2O (3 mg), glucose (10 g), agar (15 g), streptomycin (15 mg), gentamycin (15 mg) and tetracycline (12 mg) per litre. Whenever ericaceous hair roots were found directly adjacent to P. bicolorata root tips, they were assumed to belong to V. vitis-idaea, the only ericaceous species present on the plot. These hair roots were also collected and surface-sterilized by successive vigorous vortexing in commercial bleach (4.5% available chlorine, 0.1% Tween 20 v : v) for 30 s, sterile H2O (1 min), 70% ethanol (30 s) followed by five rinses in sterile H2O (30 s each). Hair roots were then cut into 2–5 mm segments and plated on the same culture media as for P. bicolorata root tips. Emerging fungal colonies were incubated in the dark at 20°C and were transferred to MMN media without antibiotics when they reached 1 cm in diameter (c. 1–2 months after plating).

Molecular identification of fungal isolates

After 6–8 wk, 15 × 5 mm strips of mycelium were sampled from the leading edge of each fungal colony. DNA was extracted using the MoBio PowerSoil®-htp 96 Well Soil DNA Isolation Kit (MO BIO Laboratories, Carlsbad, California, USA) according to the manufacturer’s instructions. The rDNA-ITS region was amplified and sequenced in both directions for each DNA extract as described in Grelet et al. (2009b). Contigs were assembled in sequencher v 3.1.1. (Gene Codes Corporation, Ann Arbor, MI, USA). All ITS sequences corresponding to isolates obtained from different root sample and/or corresponding to different ISSR genotypes (see Genotyping section below) were deposited in the EMBL nucleotide sequence database (accession codes FN678797FN678888 and FN811925FN811927– see the Supporting Information, Table S1). The ITS sequences obtained in this study were matched against fungal ITS sequences using the fasta tool search in EBI ( with default options. The position of relevant isolates within the R. ericae aggregate was confirmed by conducting a neighbour-joining (NJ) (Saitou & Nei, 1987) phylogenetic analysis with 1000 bootstrap replicates in mega4 (Tamura et al., 2007), using the Maximum Composite Likelihood method (Tamura et al., 2004) to compute evolutionary distances, assuming homogeneous/uniform rates of transitions and transversions among lineages/sites. All positions containing alignment gaps and missing data were eliminated only in pairwise sequence comparisons (pairwise deletion option). Representative ITS sequences from each clade/species described by Hambleton & Sigler (2005) were also included in the analysis for reference.


We used ISSR PCR (ISSR-PCR) in order to test whether isolates with identical ITS sequences also represented identical ISSR genotypes. Three anchored, and one nonanchored ISSR primers were used to determine the presence of identical genotypes in different root fragments: CGA5 (5′-DHB (CGA)5), ACA5 (5′-BDB (ACA)5), CCA5 (5′-DDB (CCA)5), and GACA4 ((GACA)4). Amplification reactions were carried out in 25-μl volumes containing 5 μM primer, 250 μM each dNTP, 2 mM MgCl2, 1× green GoTaq reaction Buffer, 1.25 U GoTaq DNA polymerase (Promega) and 0.5 μl DNA extract. Negative controls were included in each PCR. The PCR conditions were as follows: 1 min at 94°C, followed by 30 cycles of 1 min 30 s at 94°C, 45 s at 55°C and 1 min at 72°C, followed by 10 min at 72°C. In order to test for reproducibility in ISSR typing, several control checks were conducted. First, mycelial samples were taken at a 2-yr interval for a selected number of colonies to assess whether ISSR profiles were constant over time and to confirm that any differences were not caused by somatic variability. Second, the effect of DNA concentration on the reproducibility of ISSR-PCR was tested because, although DNA concentration in each extract was broadly similar (as assessed visually by gel electrophoresis), it was not adjusted precisely in each PCR reaction. Third, for each primer and fungal isolate, ISSR-PCR amplification reactions were performed in duplicate to confirm that the presence/absence of ISSR-bands could be reproduced and scored systematically in two independent reactions.

Two isolates of R. ericae (Read 101 supplied by DJ Read and UAMH6735 supplied by AFS Taylor) were also subjected to ISSR-typing for inclusion in relevant analyses as outgroups. Sixteen PCR samples were run simultaneously on a 1.5% agarose gel (15 × 15 cm) in Tris-acetate buffer (TAE) at 110 volts for 2.5 h. Hyperladder I and II (Bioline, London, UK) were included at both ends and in the middle of the gel to aid in band sizing (Fig. S1). The size and presence/absence of bands were determined for each isolate, and converted into binary data using GelCompar II (Applied Maths NV, Sint-Martens-Latem, Belgium) after visual confirmation of the presence/absence of each band. When identical ISSR profiles were obtained from fungal colonies emerging from the same root fragment, these were taken to represent multiple sampling of the same isolate and only one representative entry was kept in the dataset.

To further assess the reliability of our ISSR dataset, we tested for congruency between ISSR primers. The binary data exported from Gelcompar II were used to calculate a multiloci pairwise genetic Euclidian distance matrix using the software package genalex version 6 (Peakall & Smouse, 2006), following the method of Huff et al. (1993). The correlation between each pair of distance matrices generated for each individual ISSR primer were tested by a Mantel test (Mantel, 1967), following the method of Smouse et al. (1986) with 9999 random permutations.

Phenetic analyses of ISSR data

Binary data were exported from Gelcompar II and each possible ISSR fragment size was defined as a locus with two alternative alleles, coded either present (1) or absent (0). The dikaryotic phase of ascomycete fungi is transient and their vegetative mycelium is haploid (Kendrick, 2000). Therefore, we assumed in all further genetic analyses that all isolates were haploid and we refer to ISSR types as haplotypes. Data were analysed in two different ways, to avoid potential misinterpretation of single phenetic analysis of neutral arbitrary markers conducted in isolation (Hollingsworth & Ennos, 2004). First, a hierarchical cluster analysis was conducted in primer v6 using the Jaccard coefficient index to calculate pairwise genetic similarity (Jaccard, 1908) and using the average group linkage option (Clarke & Gorley, 2006). Second, the same dataset was subjected to a NJ analysis with 1000 bootstrap replicates in paup version 4.0b10 (Swofford, 2003) using Nei and Li’s similarity coefficient (Nei & Li, 1979), as a measure of genetic distance between each pair of multilocus haplotypes. We used these approaches to test: whether our ISSR profiling method was capable of distinguishing between clades within the R. ericae aggregate and was congruent with ITS data; whether isolates derived from adjacent ECM and ERM roots shared identical genotypes; and whether the observed clustering pattern related to sampling location.

Groupings within the M. variabilis clade and spatial differentiation

Grouping patterns within M. variabilis isolates were further investigated using multivariate statistical analyses (nonmetric multidimensional scaling analyses/analyses of similarities) of their ISSR profiles. These analyses assume neither normal distribution of, nor independence between, variables and are therefore appropriate for the analysis of multiloci markers irrespective of collision, homoplasy or any other interdependence issue (e.g. linkage) that may be inherent to the type of markers chosen, as long as the profiling protocol adopted (from amplification to scoring) is reproducible and consistent across the entire dataset.

To minimize PCR-based linkage between loci owing to random amplification across multiple contiguous ISSR regions, which may occur when ISSR primers are not anchored, all loci derived from the nonanchored primer (GACA)4 were excluded from the dataset. All nonpolymorphic loci, and those for which the frequency of one of the two alleles was < 5%, were also eliminated from the dataset. This minimized erroneous mutually exclusive clustering of individuals, which can occur even at low levels of population differentiation when a large number of loci are scored (Hollingsworth & Ennos, 2004). To confirm that this restriction in the dataset did not substantially reduce the genetic variability among samples at the population level, the number of genotypes recognized by 100 repeated random samplings of the data was compared between the original dataset and the restricted dataset using multilocus 1.3 (Agapow & Burt, 2001).

A series of nonmetric multidimensional scaling analyses (MDS) were first conducted in primer v6, to identify isolates showing divergent ISSR profiles and confirm clustering patterns highlighted by hierarchical clustering and NJ analyses. Non-metric MDS analyses were conducted with 9999 random starts and repeated with increasing number of dimensions, to maximise the probability that the ordination with the lowest stress value (global optimum) had been found. We applied a low similarity threshold (40%) and a snowball approach to define clusters (i.e. overlapping clusters were amalgamated within the same cluster) to increase the minimum genetic distance required to classify a group of isolates as divergent.

Analyses of similarities (ANOSIM) were then conducted to test whether the populations of M. variabilis differed genetically between forests and/or between sampling grids. The ANOSIM analyses were repeated with or without inclusion of ISSR haplotypes consistently identified as divergent by hierarchical cluster, NJ and MDS analyses. ANOSIM analyses were conducted in primer v6, with a minimum of 9999 random permutations. In addition, to assess whether genetic variation between isolates could be explained at the intrapopulation (within grids), interpopulation (among sampling grids) or inter-region level (among forests), analyses of molecular variance (AMOVA) were conducted with 9999 random permutations in both genalex version 6 (Peakall & Smouse, 2006) and arlequin v3.11 (Excoffier et al., 2005).

Analyses of autocorrelation between spatial and genetic structure

The relationship between genetic and spatial distance among M. variabilis isolates were investigated within each spatially explicit sampling grid to determine the size of each genet, and to assess whether isolation-by-distance could be inferred from our dataset at the scale of the sampling grid (0.13–45 m). Spatial autocorrelation analyses were conducted in genalex version 6 (Peakall & Smouse, 2006). For each sampling grid, autocorrelation coefficients (r) were calculated between pairwise geographical distances and squared pairwise euclidian genetic distances, within user-defined distance classes. Pairwise genetic distance matrices were calculated using the same restricted ISSR dataset as that used in MDS, ANOSIM and AMOVA analyses. Distance classes were chosen to best represent the range of distances observed within each grid. Tests for statistical significance of r were computed for each distance class using 9999 random permutations to assess the 95% interval confidence of the null hypothesis of = 0 (i.e. no significant autocorrelation between genetic and spatial distance) and 1000 bootstrapping for the 95% confidence error, as described by Peakall et al. (2003). Autocorrelation coefficients were taken to be statistically significant only when both tests (permutations and bootstrapping) were significant.

For each grid, three analyses were conducted: the first included all M. variabilis isolates; the second was clone-corrected (i.e. only one representative isolate per ISSR haplotype was included in the dataset); and the third was clone-corrected and excluded ISSR haplotypes consistently identified as divergent by MDS, hierarchical cluster and NJ analyses.


Isolation and identification of fungi

Thirty six out of the 138 cores sampled contained at least one P. bicolorata ECM root tip and 24 of these also contained neighbouring Vaccinium hair roots. From these, 70 P. bicolorata tips and 32 Vaccinium hair root segments were surface-sterilized and plated-out. This yielded 124 fungal isolates that differed in either provenance (different root segment) or colony morphology.

The majority of isolates (91%) belonged to the Helotiales (Table 1) and 63% of these fell within the R. ericae aggregate, comprising 55 isolates from P. bicolorata tips and 21 from Vaccinium hair roots. Thirty-one isolates (23%) fell within the Phialocephala fortinii species complex, 20 of which were derived from P. bicolorata tips. The remaining 17 isolates were a diverse assemblage of basidiomycetes and ascomycetes (Table 1).

Table 1.   Number and taxonomic affinity of isolates obtained from the two Scots pine forests (based on Best fasta match)
TaxonomyNumber of isolatesNumber per root typeNumber per forestBest matchMean overlapSimilarity (%)Authors (year) of Genbank entryIsolates names/ITS type*
PineVacciniumHill of fareGlen Tanar
  1. *Isolate names are based on the following nomenclature: site of collection (FG, Hill of Fare within sampling grid; FT, Hill of Fare outside sampling grid; GF, Glen Tanar within first composite sampling grid; GS, Glen Tanar within second sampling grid; GT, Glen Tanar outside sampling grids), followed by soil core number, type of source root (P, P. bicolorata tip; V, Vaccinium hair root), root number within soil core, and final letter/number distinguishing different isolates derived from the same root piece. Accession codes for internal transcribed spacer (ITS) sequences of representative isolates are given in the Supporting Information, Table S1.

Meliniomyces variabilis22175517EF093173: Meliniomyces variabilis isolate MVA-9522/561 99–100Vohnik et al. (2006)TypeI1 (FG15P1, FG15V1, FG40P1, FG54P1b, GT6P1), TypeI3 (GF3V2, GF4P1b, GF4P2, GF4P3, GF4P5a, GF4P5b, GF4P5c), TypeI4 (GT3P3, GT3V1a, GT3V3), TypeI5 (GF10P1a, GS13P1, GT5P1, FG37P1b), FG34V1, GF3P1b, GF3P3
Meliniomyces variabilis116556AY838785: Meliniomyces variabilis strain UAMH 6826798/259299.6–100Hambleton & Sigler (2005)TypeI2 (GF7P1b, GF7P2b, GF7V1a, GF7V1b, GS15P4), TypeI9 (FG50P1, GF2P1), TypeI10 (FG31V1a, FG31V1b), FG37P1a, FG37V2
Meliniomyces variabilis24168717EF093170: Meliniomyces variabilis isolate MVA-6526/559100Vohnik et al. (2006)TypeI6 (GT02P1 = [Isolate F], FG17P3a, FG28V1a, FG38P1, FG38P2, FG38P3, FT4P1a, FT4P1b, GS14P1b, GF1P1, GF1P2, GF1V1, GF4P4, GF8P2, GF8V1, GF8V1b, GF8V1g1, GF8V1i, GT3P4, GT3V2, GT4P1, GT4P3, GT03V1 = [Isolate Hc], GT9P1b)
Meliniomyces variabilis85308AJ308339: M. variabilis ARON2894.S464/48699.6Vrålstad (2001)TypeI7 (GS11P3, GS11P4, GS11P5, GS11P6, GS11V3a, GS11V4a, GS11V5a, GT7P1)
Meliniomyces variabilis44004AJ430110: M. variabilis ARON2971.S486/48698.8Vrålstad et al. (2002a,b)TypeI8 (GF6P1a, GF6P1b, GS12P1, GT13P1)
Meliniomyces variabilis11001AF149082: Salal mycorrhizal fungus strain UBCtra43488/51899.6 Millar et al. (1999)GS14P1a1
Meliniomyces bicolor11001AY579413: ECM isolate LVR4069562/56399.6 Villarreal-Ruiz et al. (2004)GT01P1 = [Isolate E]
Meliniomyces bicolor11010AJ430147: M. bicolor ARON2805.S476/47699.8Vrålstad et al. (2002a,b)FG54P1c
Meliniomyces vraolstadiae44040AJ292200: M. vraolstadiae ARON2917.S447/47899.2Vrålstad (2001)TypeII (FG2P1, FG17P2, FG17P3b, FG34P1).
Rhizoscyphus ericae Aggregate7655212254     
Phialocephala fortinii complex119283AF214580: P. fortinii strain UAMH6677486/49599.4–99.6Hambleton et al. (1999)FG14P1, FG14P2, FG28P1, FG43P3, FG49P2, FG49V2a1, GS11P1, GS11V1, GS11P2, FG54P1a, FG54P2
Phialocephala fortinii complex21120AY078144: P. fortinii strain CBS 109300485/48599.8Gruenig et al. (2002)FT4P1, FT6V1
Phialocephala fortinii complex18108108AM905084: Fungal sp. ITS product clone JH8485/56799.8–100Heinonsalo et al. (2007)FG31V2, GS15P1, GS15P1c, GS15P3, GF9V1, GT1P1, GT5V1, FG17P1, FG17V1, FG22P2, FG2V1, FG31P1, FG38V3, FT7V1, FT8P2, FT8P3, GF10P1a2, GF10V1
TotalPhialocephala fortinii complex3120112011     
Phialocephala sphaeroides51405AY524844: P. sphaeroides strain UAMH 10279787/312199.2–99.4Hollingsworth & Ennos (2004)GS11V4c, GS14V1, GS15P2a1, GS15V1, GS15V2
Other unclassified Helotiales11001AF462425: Rhabdocline parkeri strain RP-CHMI557/57783.8Catal & Adams (2001)GT3P1
Dermateaceae10101AF141163 : Dermea viburni strain CBS145.46542/82095.8Abeln et al. (1999)GF8V1f
Ascomycete11001AY354278 : Ascomycete sp. Olrim339477/477100Lygis et al. (2004)GF7P1a
Pezizomycete44004EU076954: Soil fungal sp., isolate 33-M-2 from Vietnam519/51996.4–96.6Deng & Zhu (2007)GF3P1a, GF4P1a1, GF6P1c2, GF6P2
Unclassified fungi21102AJ541799: Mortierella sp. Isolate 3441 Finse 15-07-00536/53688.5–88.7Hoiland (2003)GF5P1b, GF5V1a
Basidiomycetes; Agaricales20211AM084421: Agaricales sp. Isolate HK-S168542/60299–99.5Kauserud (2005)FG2V1c, GF8V1e
Basidiomycetes11010AF481369: ECM root tip 81-sepB_Ny1.EB-17.1440/44085.9Rosling (2002)FG15P2b
All isolates12484404480     

Most (93%) of the ‘H. ericae’ isolates fell within the M. variabilis clade (Hambleton & Sigler, 2005) (Fig. 2). The ITS sequence identity among these M. variabilis isolates was at least 97.7%. They originated from both ectomycorrhizas and ericaceous hair roots. Two isolates from ectomycorrhizas were assigned to M. bicolor (Hambleton & Sigler, 2005), and a further four to M. vraolstadiae (Hambleton & Sigler, 2005). We did not recover any Rhizoscyphus ericae isolates from either P. bicolorata ECM tips or Vaccinium hair roots.

Figure 2.

 Neighbour-joining phylogenetic analysis of internal transcribed spacer (ITS) sequences showing the position of Meliniomyces variabilis and Meliniomyces bicolor isolates within the Rhizoscyphus ericae aggregate. The optimal tree with the sum of branch length = 0.5823 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches. The tree is drawn to scale, with the scale bar representing the number of base substitutions per site. The five main clades are as defined by Hambleton & Sigler (2005) and Vrålstad et al. (2000). The evolutionary distances were computed using the Maximum Composite Likelihood method. All positions containing alignment gaps and missing data were eliminated only in pairwise sequence comparisons (Pairwise deletion option). There were a total of 522 positions in the final dataset. Phylogenetic analyses were conducted in mega4.

Within the 76 M. variabilis isolates, 16 different ITS types were detected, 10 of which were shared by 2–24 isolates (Table 1). Isolates with identical ITS sequences were obtained from different soil cores, different root tips, different sampling grids or even from the two different forests (Table 1). In seven instances, identical M. variabilis ITS sequences were isolated from P. bicolorata ectomycorrhizas and Vaccinium hair roots collected adjacent to each other within a single soil core.

Genotyping of isolates using ISSR profiling

All the tests we conducted indicated that ISSR-typing was stable over time and reproducible, even with the nonanchored primer GACA4. The ISSR profiles were identical between duplicate PCR amplifications and duplicate DNA extracts, irrespective of the primer tested, and whether or not the duplicate DNA extracts originated from colonies sampled at the same time or across a 2-yr interval (data not shown). The ISSR profiles were also identical when the concentration of DNA template in the PCR reaction was halved or doubled (data not shown).

Bands were assigned to 37–52 class sizes depending on the primer. On average, each isolate yielded 16–17 bands per primer: 17.6 ± 0.2 for ACA5, 17.4 ± 0.3 for CCA5, 16.0 ± 0.3 for CGA5 and 16.8 ± 0.4 for GACA4. A total of 182 loci were scored using all four primers (133 if only anchored primers were scored). Genotyping was broadly congruent among the four primers, as the pairwise genetic distance matrices were significantly correlated for each pair of primers (< 0.001, 0.15 ≤ R2 ≤ 0.43).

Seventy different ISSR haplotypes were recognized (Fig. 3). This was reduced to 67 if the nonanchored primer GACA4 was excluded. Hierarchical cluster and NJ analyses clearly separated the M. variabilis isolates from the other Meliniomyces species and from the two R. ericae reference strains (Fig. 3). Six of the seven pairs of isolates sharing identical ITS sequences, and obtained from adjacent P. bicolorata ectomycorrhizas and Vaccinium hair roots, had identical ISSR profiles (Fig. 3).

Figure 3.

 Hierarchical cluster analysis based on presence or absence of intersimple sequence repeat (ISSR) fragments amplified. The Jaccard coefficient index was used to calculate pairwise genetic similarity between ISSR profiles. The cluster analysis was conducted with the average group linkage option in primer v6. Bootstrap support obtained from a neighbour-joining analysis conducted in paup 4.0b10 are indicated next to relevant branches. The neighbour-joining analysis included 1000 bootstrap replicates and used Nei and Li’s similarity coefficient to calculate pairwise similarities between ISSR types. Both cluster and neighbour-joining analyses were conducted with or without inclusion of bands derived from nonanchored primer (GACA)4. The ITS-types and forest sites of origin are apposed next to each isolate. Isolates obtained from different root type (Vaccinium versus pine roots) and sharing identical ISSR profile are enclosed by the same solid grey line. Isolates from the same soil core with identical ITS types, but different ISSR profiles, are boxed and interconnected by black (soil core GF4) and grey (soil core GS11) dotted lines. The star indicates the main M. variabilis cluster.

Clustering patterns among M. variabilis ISSR haplotypes

Neighbour-joining and hierarchical cluster analyses identified four ISSR haplotype clusters, composed of one to four haplotypes, which fell outside the main M. variabilis cluster, but which had the same ITS sequence as isolates in the main cluster. Within the main M. variabilis cluster, ISSR data were only partly congruent with ITS data, and not all isolates of the same ITS type clustered together (Fig. 3). Interestingly, the six dual ECM–ERM haplotypes clustered together in the hierarchical cluster tree (Fig. 3), and in the NJ tree (not shown), but with no bootstrap support.

To investigate further the clustering pattern within M. variabilis, the dataset was ‘clone corrected’ and restricted to polymorphic loci derived from the three anchored primers only, that is, excluding all loci derived from the nonanchored primer, all nonpolymorphic loci and all loci for which the frequency of one of the two alleles was < 5%. The restriction in the number of loci did not lead to a significant loss of genotypic diversity, as 25 polymorphic loci were sufficient to identify 95% of the genotypes (Fig. S2). The MDS analyses using the restricted dataset indicated that while most haplotype similarity groups overlapped to form a single cluster, there were four outlying groups (Fig. S3). These outliers contained the same isolates that fell outside the main M. variabilis cluster in both NJ and hierarchical cluster analyses, confirming that these haplotypes were different from the majority of the other M. variabilis isolates (Fig. 3). Because of the small number of isolates in the outlying groups, it was not possible to test the significance of this divergence.

Neither MDS, nor NJ, nor hierarchical cluster analyses provided evidence of clustering according to forest, or sampling grid within Glen Tanar (Figs 3, S3). The ANOSIM analyses indicated that there was no significant genetic difference amongst the three populations sampled within Glen Tanar (GF, GS and GT) (Table 2). When these three populations were treated as a single population there were no significant differences between the Glen Tanar population and that sampled at Hill of Fare (Table 2). Outcome of the ANOSIM analyses were not affected by the inclusion or exclusion of the four groups of divergent ISSR haplotypes in the dataset (data not shown). AMOVA analyses conducted in both arlequin v3.11 and Genalex v6 showed that the majority of the genetic variation (95–100%) occurred within populations. The proportion of the genetic variation explained by difference between populations or between forests was insignificant, and these results were consistent with or without inclusion of divergent ISSR haplotypes.

Table 2.   Analysis of similarity (ANOSIM) between haplotypic composition (based on intersimple sequence repeat (ISSR) profiles) of Meliniomyces variabilis populations isolated from Piceirhiza bicolorata pine ECM root tips and neighbouring ericoid Vaccinium roots collected in four spatially defined locations in two Scottish pine forests
Populations comparedDivergent haplotypes includedaDivergent haplotypes excludeda
  1. aANOSIM analyses were repeated with or without inclusion of divergent ISSR haplotypes.

  2. bNumber of ISSR haplotypes per population.

  3. cCalculated as of the proportion of possible permutations for which R is equal or greater than the overall observed R statistic. When comparisons were made between two populations of the smallest size (n1/2 ≤ 16), P-values equal to 0.05 were taken to be nonsignificant and only differences with < 0.05 were taken to be significant. ns, not significant.

Comparisons within the semi-natural forest
 GFGS22100.26 ns16100.05 ns
 GFGT22130.16 ns16100.07 ns
 GSGT10130.05 ns10100.07 ns
Comparison between forests
 FGGF + GS + GT15450.81 ns15360.46 ns

Extent of M. variabilis genets and spatial structure

Meliniomyces variabilis isolates sampled from the three spatially structured grids were subjected to spatial auto-correlation analyses based on their ISSR profiles, using the restricted dataset with or without inclusion of multiple clone entries. Because not all soil cores contained P. bicolorata ectomycorrhizas, it was not possible to define contiguous distance classes between pairs of M. variabilis isolates. Distances classes were therefore optimized for each sampling grid, so as best to represent the range of spatial distances observed between M. variabilis isolates. The number of ISSR haplotypes within each grid ranged from 10 to 22, and the number of pairwise comparisons within each distance class ranged from 5 to 85. When the dataset included all isolates (i.e. not clone-corrected), there was significant spatial autocorrelation only at the smallest distance classes: within the same core in grid FG (Hill of Fare), within the same pair of adjacent pine and Vaccinium roots in grid GF (Glen Tanar) or even only within the same root in grid GS (Glen Tanar). (Fig. 4). However, when the dataset was clone-corrected (i.e. only one entry per haplotype), there was no spatial autocorrelation at any distance classes, although the highest values of r were recorded at distance classes equal to or smaller than the soil core unit (data not shown). We conducted the analyses both with and without the inclusion of isolates displaying divergent ISSR haplotypes, and found that it did not change the outcomes of autocorrelation analyses (data not shown). Within the limitations inherent to the relatively small sample size of our dataset, these results indicate that the size of individual genets was smaller than the size of the soil core (4.5 cm diameter × 12 cm depth, i.e. maximum 13 cm), and that genetic relatedness was independent of spatial separation, at least within the sampling scale investigated.

Figure 4.

 Relationship between genetic and spatial distance among Meliniomyces variabilis isolates sampled at Hill of Fare (a), Glen Tanar grid GF (b) and Glen Tanar grid GS, UK (c). Data include all isolates (including clones and isolates with divergent ISSR haplotypes). Spatial autocorrelation analyses were conducted in GenAlEx version 6. Bars indicate values of autocorrelation coefficients (r) calculated between pairwise geographical and pairwise squared genetic distance within each user-defined distance class. Error bars correspond to the 95% confidence error (1000 bootstrap replicates) of r, and dashed lines correspond to boundaries of the 95% interval confidence of the null hypothesis of no autocorrelation between genetic and spatial distance (i.e. = 0) (9999 permutations). Asterisks indicate significant autocorrelations between genetic and spatial distance at given distance class (*, P < 0.05).


The majority of the isolates obtained from P. bicolorata ectomycorrhizas and adjacent hair roots belonged to the species M. variabilis sensu lato, which has been reported to form functional ericoid mycorrhizal associations with Vaccinium (Grelet et al., 2009a), and is frequently PCR-amplified from ectomycorrhizal root tips (Vrålstad et al., 2002a). However, we need to be mindful of the fact that our analyses were based on culturable fungi and this has been shown in the past to detect only a subset of the total fungal community (Bougoure & Cairney, 2005). Many of the isolates within a plot that shared identical ITS types represented different ISSR genotypes, emphasizing that identical ITS types cannot be taken as evidence for sharing of mycorrhizal symbionts between roots. However, on six occasions we did find identical M. variabilis genotypes in pine ectomycorrhizas and adjacent ericaceous hair roots, demonstrating coinfection of neighbouring ECM and ERM roots by six different fungal genets. Further, spatial analyses suggested that M. variabilis genets were smaller than 13 cm, based on the greatest dimension within a core.

The potential for fungi to form mycorrhizal networks between ectomycorrhizal and ericoid mycorrhizal plants was first suggested by Bergero et al. (2000) and Vrålstad et al. (2000). Villarreal-Ruiz et al. (2004) later showed that M. bicolor obtained from a P. bicolorata pine ectomycorrhiza could simultaneously form ectomycorrhizas with pine and ericoid structures with Vaccinium roots in vitro, and could induce beneficial effects on host plant growth. We recently showed that M. variabilis and M. bicolor obtained from ectomycorrhizal pine roots could not only form typical ericoid mycorrhizal structures, but could also engage in reciprocal transfer of carbon and nitrogen with Vaccinium seedlings in monoxenic conditions (Grelet et al., 2009a). Curlevski et al. (2009) found identical fungal genotypes in roots of Epacris pulchella, an Australian species forming ericoid mycorrhizas, and Leptospermum polygalifolium (Myrtaceae) which has been reported to form both ECM and arbuscular mycorrhizal associations (Pattinson et al., 2004). There are therefore numerous lines of evidence pointing towards the ability of some fungal taxa, and particularly those with affinity to the R. ericae aggregate, to associate with both ECM and ERM roots. Our data from the present study confirms that identical genotypes of M. variabilis occur simultaneously in pine ECM roots and adjacent ericaceous hair roots. This ‘coinfection’ of adjacent ECM and ERM roots was observed in two different forests, and at multiple locations within Glen Tanar. Given the relatively modest number of samples handled in this study, it appears likely that simultaneous colonization of neighbouring ECM and ERM roots is a common occurrence in the field. While it is possible that these shared genets formed an uninterrupted mycelial network between pine tree and ericaceous roots over small distances, it is not possible to determine if intact mycelial connections were indeed present at the time of sampling, or if the integrity of such mycelia networks is stable over time.

Our study is the first insight into the spatial distribution of ericoid mycorrhizal fungi in plant communities. The only previous spatial investigation was restricted to individual ericaceous root systems (Midgley et al., 2004). Using spatial autocorrelation analyses, we showed that each individual M. variabilis genet was small (< 13 cm). Although small genets have been observed for some ECM fungi (Gryta et al., 1997), many species have been shown to have genets that extend over significant distances (metres to tens of metres), in the case of both ECM basidiomycetes (e.g. Anderson et al., 2001; Kretzer et al., 2004; Carriconde et al., 2008) and ascomycetes (e.g. Wu et al., 2005). Recently, Beiler et al. (2010) showed that ectomycorrhizal Rhizopogon spp. formed genets that could not only occupy an area of up to 135 m2, but could also colonize the root systems of up to 19 different trees. By contrast, we found no evidence for large genets of M. variabilis in roots. It is necessary, however, to bear in mind that our study, like most other spatial studies of fungi, is constrained by our choice of target organ. In our case, we targeted P. bicolorata ectomycorrhizas and restricted our sampling of ericaceous roots to those few found directly adjacent to the P. bicolorata ECM root tip sampled. It is therefore possible that our sampling strategy led to the underestimation of M. variabilis genet size.

There is no clear evidence that M. variabilis can form ectomycorrhizas (Hambleton & Sigler, 2005), so the functional significance of a M. variabilis hyphal network between ECM and ERM roots is not known. We tested whether several of our M. variabilis isolates could form ectomycorrhizas with pine and birch in vitro and although all of them colonized the short roots of both species, with neutral or positive effects on host plant growth, they did not form typical ECM structures (G-A Grelet & J Brodie, unpublished). Similar results were obtained by Vrålstad et al. (2002b): However, the strains tested by Vrålstad did not form ERM structures either. By comparison, both our M. bicolor isolates formed ectomycorrhizas with pine and birch (G-A Grelet & J Brodie, unpublished), as did all M. bicolor isolates tested by Vrålstad et al. (2002b). It may be that M. variabilis is an endophyte in ectomycorrhizas formed by other fungi. This would be compatible with its ability to function as an ERM fungus, and support recent reports of endophytic behaviour in several ECM and ERM host plants (Ohtaka & Narisawa, 2008; Vohnik et al., 2007a,b) or DNA amplification from ectomycorrhizas formed by other fungal species (Tedersoo et al., 2009).

We found several M. variabilis genotypes within the same soil core, including a number of isolates with identical ITS types. Other root-associated ascomycetes display a similar phenomenon, particularly those representing cryptic species complexes (Grunig et al., 2004, 2008). Up to six different multiloci haplotypes of P. fortinii were found at the same sampling grid point (Queloz et al., 2005). Within Cenococcum geophilum, as much genetic diversity has been detected with a single soil sample as between isolates collected from across the USA (Douhan & Rizzo, 2005).

We found no evidence for isolation by distance. Meliniomyces variabilis isolates obtained from the two different forest sites were drawn from populations that were genetically indistinguishable, despite a distance of 20 km. Furthermore, when repeated samplings of the same genet were removed from the dataset (clone correction), no significant autocorrelation between spatial and genetic distance could be observed over distance classes ranging from a few millimetres to 43 m. Spore production by M. variabilis has not been reported. However, our data suggest that either the modes of propagation of this fungus are not constrained by the distances used in this sampling, indicating significant genotypic exchange over 20 km, or that the populations we sampled from the two forests had some common origin. This could result from forestry practices. Further multigene phylogenetic analyses may help assess the occurrence of sexual reproduction in this species and would clarify potential cryptic speciation.

Until further light is shed on the trophic status of M. variabilis sensu lato, its ecological role in mixed heaths and coniferous forests remains unknown. In particular, functional studies are urgently required to clarify the nature of the interaction between M. variabilis and ectomycorrhizal tree hosts.


We thank Pamela Parkin and Alison Williams for technical assistance with the molecular work, and Janis Brodie for assistance with fungal isolation and culturing. We are grateful to D. J. Read and A. F. S. Taylor for the donation of R. ericae cultures. This manuscript greatly benefited from discussion with Annette Kretzer and Valentin Queloz, and comments from three anonymous reviewers: all are gratefully acknowledged. We are indebted to the Glen Tanar Estate, Hill of Fare Estate and SNH for permission to sample in the two forests. This research was funded by a NERC-research grant awarded to IJA, DJ and ICA. The Macaulay Institute receives funding from the Scottish Executive (Rural and Environment Research and Analysis Directorate).