High diversity of arbuscular mycorrhizal fungi in a boreal herb-rich coniferous forest


Author for correspondence:
Maarja Öpik
Tel: +372 7376224
Fax: +372 7376222
Email maarja.opik@ut.ee


  • • Here, the diversity of arbuscular mycorrhizal (AM) fungi was determined in a boreal herb-rich coniferous forest in relation to environmental variables.
  • • Root samples of five plant species (Fragaria vesca, Galeobdolon luteum, Hepatica nobilis, Oxalis acetosella and Trifolium pratense) were analysed from stands differing in age and forest management intensity.
  • • Thirty-four Glomeromycota taxa (small-subunit ribosomal RNA gene (SSU rDNA) sequence groups) were detected from 90 root samples (911 clones), including eight new taxa. Sequence groups related to Glomus intraradices were most common (MO-G3 and MO-G13). Samples of H. nobilis were colonized by more AM fungal taxa (3.68 ± 0.31) than those of O. acetosella (2.69 ± 0.34), but did not differ significantly in this respect from those of F. vesca (3.15 ± 0.38). Effects of forest management, host plant species (except above) or season on the number or composition of fungal taxa in root samples were not detected, and neither were they explained by environmental variables (vegetation, soil and light conditions).
  • • This is the most taxon-rich habitat described to date in terms of root-colonizing Glomeromycota. The data demonstrate the importance of temperate coniferous forests as habitats for AM fungi and plants. Lack of obvious fungal community patterns suggests more complex effects of biotic and abiotic factors, and possibly no adverse effect of common forest management practices on AM fungal diversity.


Arbuscular mycorrhizal (AM) fungi (Ph. Glomeromycota) are ubiquitous plant root symbionts that can be considered as ‘keystone mutualists’ in terrestrial ecosystems, forming a link between biotic and abiotic ecosystem components via carbon and nutrient fluxes that pass between plants and fungi in the soil (O’Neill et al., 1991). AM fungal diversity affects plant community diversity and productivity (van der Heijden et al., 1998). There can be large differences in functional complementarity between coexisting plants and AM fungi (Helgason et al., 2002; Moora et al., 2004a,b). Therefore it is essential to understand the fine-scale structure and dynamics of AM fungal communities in natural and managed ecosystems. However, possible approaches to link the taxon diversity of Glomeromycota communities with functional significance are still under debate because of the practical difficulties of working with such obligate symbiotic organisms (van der Heijden & Scheublin, 2007).

The diversity and composition of intraradical AM fungal communities vary among habitat types around the globe (Öpik et al., 2006a). When comparing studies using the small subunit ribosomal RNA gene (SSU rDNA) region to identify AM fungi, the highest taxon richness in a single site has been reported from tropical rain forest in Panama – 30 taxa in the roots of three host species (Husband et al., 2002a,b; 29 according to the taxon synonymy used by Öpik et al., 2006a). Taxon-rich fungal communities (over 20 taxa) are also known from temperate grassland and forest locations (Vandenkoornhuyse et al., 2002; Saito et al., 2004; Helgason et al., 2007). Tropical forest may exhibit significantly higher mean fungal richness than other ecosystems (expressed as number of fungal taxa per plant species: 18; Öpik et al., 2006a). By contrast, human-impacted habitats such as arable fields (Helgason et al., 1998; Daniell et al., 2001) or sites polluted with heavy metals (Whitfield et al., 2004) may exhibit low AM fungal taxon diversity. However, recent evidence of higher richness in such habitats (Hijri et al., 2006; Vallino et al., 2006) suggests that the relationship with management is complex.

AM fungal communities in deciduous forests of the temperate zone have been described from the UK by Helgason and colleagues (Helgason et al., 1998, 1999, 2002, 2007). They have recorded > 20 AM fungal taxa from roots of six host plant species in a single forest location. Nine taxa, most of them unique, were detected in 25 pooled root samples from a warm-temperate broadleaved forest in Japan (Yamato & Iwase, 2005). However, the presence of AM fungi in coniferous forests has been largely ignored. Only two AM fungal taxa were detected in the roots of Taxus baccata in a Norway spruce (Picea abies) forest in Germany (Wubet et al., 2003) and 10 taxa in the roots of two Pulsatilla species in a Scots pine (Pinus sylvestris) forest in Estonia (Öpik et al., 2003). To our knowledge, no further information (including that from spore surveys) is available relating to AM fungal richness in boreal forest ecosystems. The land area covered by boreal forest constitutes one-third of the world's forest and provides important ecological functions as well as 20–50% of the world's pulp, newsprint, sawn wood, paper and paper board (Reich et al., 2001). There is a considerable wealth of information regarding the diversity of ectomycorrhizal fungi – the symbionts of the dominant overstorey plants in temperate/boreal coniferous forests (Horton & Bruns, 2001; Johnson et al., 2005; Tedersoo et al., 2006). However, in herb-rich temperate/boreal coniferous forests, where there are plentiful AM plant species, Glomeromycota should not be overlooked as an ecosystem component. In order to obtain a good overview of biodiversity in coniferous forests, more information about AM fungi is required.

Different management practices may have significant impacts on the diversity and composition of boreal forest plant communities (Reich et al., 2001; Ramovs & Roberts, 2003). The impact of logging can directly influence vegetation via the disturbance of soil or forest floor, altered habitat structure, removal of nutrients, or altered microclimate (Roberts & Gilliam, 1995; Bergeron & Harvey, 1997). Clearcut logging may have a significant impact on ectomycorrhizal fungal communities (Jones et al., 2003). There is, however, no information about the impact of boreal forest management on AM fungi.

In the present paper, we aimed to investigate the taxon composition and community structure of AM fungi in a herb-rich boreal coniferous forest. In particular, we asked: (1) what is the taxon richness and composition of the AM fungal communities of the studied coniferous forest, and (2) what is the impact of local environmental conditions and forest management (clear-felling and cultivating clear-cut areas versus old growth) on the taxon richness and composition of AM fungal communities?

Materials and Methods

Study area

The study area is located at Koeru, central Estonia (58°58′N; 26°03′E). The landscape of the region is flat, consisting of a mosaic of cultivated areas and forest. The climate is transitional between maritime and continental. The mean annual precipitation is 700–750 mm. The mean annual air temperature in the region is 4.3–6.5°C, ranging between –7°C in January and 17.4°C in July (Jaagus, 1999). The study area is a 130-ha patch of Hepatica nobilis Mill. site type spruce forest (typification after Lõhmus, 2004). The soil is a calcaric cambisol (typification after Food and Agriculture Organization of the United Nations (FAO)). Soil conditions vary little across the study area (Zobel et al., 2007). The predominant tree species is Norway spruce (Picea abies (L.) H. Karst.) with Corylus avellana L. prevailing in the shrub layer. Altogether 70 herbaceous vascular plant species have been recorded in the field layer: Oxalis acetosella L., Fragaria vesca L. and H. nobilis were the most abundant plant species; Dicranum scoparium Hedw. and Cirriphyllum piliferum (Hedw.) Grout were the most common bryophytes (Moora et al., 2007).

The area has been maintained as forest since at least 1828 (from map evidence). The forest has been managed with clear-cutting in patches of approx. 1–2 ha. However, areas of the forest can still be classified as old growth, with different age classes present, and the oldest spruce trees being 130–140 yr old. In these areas selective felling has been practised.

Root sampling

We sampled forest ecosystems of different age and management intensity. Mature spruce forests with a heterogeneous canopy represented old-growth stands, where the intensity of forest management has been low and the ecosystems are close to their natural state. Early successional stages were represented by young dense stands in areas that were clear-cut 20–25 yr ago and then replanted with Norway spruce. Young stands have been thinned repeatedly since planting.

Both old and young forest stand types were replicated three times on similar soil conditions. From each of the six stands, plant roots were sampled from a 10 × 10 m plot divided into 1 × 1 m subplots. 1 × 1 m subplots were further divided into six equal parts for six sampling times: the beginning of June, end of July, and beginning of October 2003 and 2004.

Thus, a 1 × 1 m subplot could be sampled six times consecutively without disturbing the soil and breaking the fungal mycelium. The following five vascular plant species were sampled for this study: O. acetosella (Oxalidaceae), H. nobilis (Ranunculaceae) and F. vesca (Rosaceae), which were the most frequent species in the field layer and present in abundance in all succession stages; Galeobdolon luteum Huds. (Lamiaceae, syn. Lamium galeobdolon), which was patchily distributed and present only in two stands of old forest; and Trifolium pratense L. (Fabaceae), which was present only in young stands. Entire plants (several individuals if roots were very small) of each species were excavated from the 1/6-m2 subplot if present and placed in plastic bags. In the laboratory, roots were cleaned, dried with silica gel and stored dry until analysis. Please note that only two samples of each plant species per plot per sampling time and the first three sampling times were used in this study (Table 1).

Table 1.  Sampling scheme showing the number of plants successfully analysed for molecular diversity of arbuscular mycorrhizal (AM) fungi in roots
PlotSampling timePlant speciesTotal
  1. Two plants were subjected to analysis per sampling time, species and plot, except for one sample each of HN from plot W at sampling times 1 and 3. Elsewhere values < 2 indicate no success in PCR or cloning. FV, Fragaria vesca; GL, Galeobdolon luteum; HN, Hepatica nobilis, OA; Oxalis acetosella; TP, Trifolium pratense. NP, plant species not present in this plot; NA, not analysed. Plots (10 × 10 m) are designated T, R, P in young stands and Z, W, Y in old stands.

Young stands
T1 0NP 2 0NA 2
2 0NP 2 2NA 4
3 1NP 2 1NA 4
T total 1 63 10
R1 0NP 2 20 4
2 0NP 2 22 6
3 0NP 2 22 6
R total   66416
S1 2NP 1 00 4
2 0NP 2 21 5
3 2NP 2 02 6
S total 4 52314
Young stands total5 1711740
Old stands
Z1 20 2 2NP 6
2 22 2 2NP 8
3 21 2 2NP 7
Z total 6366 21
W1 2NP 1 2NP 5
2 1NP 2 1NP 4
3 2NP 1 2NP 5
W total 5 45 14
Y1 00 2 0NP 2
2 21 0 2NP 5
3 22 2 2NP 8
Y total 4344 15
Old stands total1561415 50
Plant samples total2063126790

Vegetation analysis and environmental conditions

In all 1 × 1 m subplots within six 10 × 10 m plots (600 altogether) we recorded the per cent coverage of all vascular plant species in the field layer, but not the shrub layer, and the total cover of all bryophytes before root sampling. In each subplot, local environmental conditions were characterized as follows. Topsoil samples (1–10 cm) were taken from the centre of each subplot for the determination of pH and the content of mineral nitrogen (N) (inline image-N, inline image-N and total N), phosphorus (P) and dissolved organic material (DOC). Soil pH was measured in 0.01 M CaCl2 (10 g of soil in a 50-ml solution). DOC and mineral N were extracted from 10 g of soil with 1 M KCl (soil:extractant ratio 1 : 4) and filtered through Whatman No. 1 filter paper (Wheatley et al., 1989). Available soil P was extracted using the sodium bicarbonate (Olsen) method (Olsen et al., 1954). N, P and DOC concentrations were determined colourimetrically on a segmented flow autoanalyser (Skalar Analytical, Breda, the Netherlands).

Light availability was estimated using photographs taken at the height of 30 cm at the centre of each subplot with a Nikon CoolPix 950 digital camera equipped with a hemispherical lens. All photographs were taken at times when the sun was blocked by clouds to ensure homogeneous illumination of the overstorey canopy and correct contrast between canopy and sky. We calculated an indirect site factor (ISF) and a direct site factor (DSF) by using WinSCANOPY software (Regent Instruments Inc., Québec, Canada) assuming the standard overcast sky model (Anderson, 1966). ISF and DSF are defined as the proportion of diffuse and direct radiation received below the tree canopy as a fraction of that received above the canopy (Rich, 1990).

Molecular analyses

AM fungi were identified on the basis of sequence variation within an SSU rDNA region in two individuals from each of five plant species, six stands and three sampling times (Table 1); each plant individual was sampled once. A representative subsample (approx. 20 cm) of a root system of a plant was pulverized with 1.1-mm tungsten carbide beads (BioSpec Products, Inc., Bartlesville, OK, USA) with Mixer Mill 301 (Retsch GmbH, Haan, Germany). DNA was then extracted using the Nucleospin® 96 Plant kit (Macherey-Nagel, Düren, Germany) eluting in a final volume of 100 µl. Following optimization for template quantity, 5 µl of the DNA extraction was used in a 25-µl volume PCR reaction containing Expand High Fidelity Buffer (Roche Applied Science, Mannheim, Germany) with 15 mM MgCl2, 100 nM of each of the dNTPs, 200 nM of each of the primers NS31 and AM1 (Simon et al., 1992; Helgason et al., 1998), 20 mg ml−1 bovine serum albumin (BSA) and 0.7 units of Expand High Fidelity enzyme mix (Roche Applied Science). Thermocycling conditions were as follows: 94°C for 2 min; 10 cycles of 94°C for 15 s, 58°C for 30 s and 72°C for 45 s; 20 cycles of 94°C for 15 s, 58°C for 30 s and 72°C for 45 s + 5 s per cycle; 72°C for 7 min using a DNAEngine PTC Dyad thermocycler (MJ Research, Reno, NV, USA). Positive PCR products were purified using the MinElute PCR Purification kit (Qiagen, Crawley, UK), cloned and sequenced following the method of Griffiths et al. (2006). Purified PCR products were inserted into the pGEM-T Easy vector (Promega, Madison, WI, USA) and transformed into Escherichia coli DH10B electrocompetent cells prepared in the laboratory following the method of Tung & Chow (1995). From each sample, 16–32 colonies were grown in 1 ml of 2 × Luria-Bertani (LB) broth with 0.15 mg ml−1 ampicillin in deep-well microtitre plates. Plasmids were purified with a Multiscreen Plasmid Minipreparation Kit (Millipore, Bedford, MA, USA) following the manufacturer's instructions. Sequencing was performed in a volume of 10 µl with a 1 : 8 dilution using the BigDye® Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Warrington, UK) with vector primers directed against the SP6 or T7 promoter regions. Sequencing reactions were purified using 96- or 384-well Geneclean plates (Genetix, Queensway, New Milton, UK) following the manufacturer's instructions and run on an ABI Prism 3700 DNA Analyzer (Applied Biosystems).

Phylogenetic sequence analyses

Raw sequences from 911 clones were aligned using the freeware poa (Lee et al., 2002). First, all root-derived sequences were submitted to neighbour-joining analysis (F84 model with gamma substitution rates) implemented in TOPALi version 1 (Milne et al., 2004). From the apparent sequence groups second strands of representative clones were sequenced; the double-stranded sequences were submitted to a blast search. Retrieved sequences of closely related ‘known’ fungi and environmental samples, sequences of major clades of Glomeromycota, and double-stranded sequences obtained in this study were then aligned automatically using the MAFFT multiple sequence alignment web service in JalView version 2.3 (Clamp et al., 2004). Representative sequences of detected sequence groups were submitted to the European Molecular Biology Laboratory (EMBL) Nucleotide Sequence Database (accession numbers AM849253–AM849327).

Phylogenetic analysis of rRNA alignment can be significantly improved by taking into account RNA secondary structure. Analysis methods that model stem region as doublets (i.e. 16 possible states) taking base pairing into account are better than simple four-state nucleotide models (Telford et al., 2005). We therefore annotated each position in our alignment as belonging to a loop or to a stem based on structure coordinates for the Geosiphon pyriformis X86686 structure obtained from the European ribosomal RNA database (http://bioinformatics.psb.ugent.be/webtools/rRNA/). These coordinates were processed into MrBayes nexus format using the Ystem python script (Telford et al., 2005) which identified position type (stem or loop) and the coordinates of pairs of nucleotides in the stem.

Loop regions were then modelled according to a standard 4 × 4 nucleotide substitution model. Model selection was based on the six nucleotide substitution matrix models available in the MrBayes software (Ronquist & Huelsenbeck, 2003) with or without rate heterogeneity modelled by the gamma distribution. Rather than the conventional approach of comparing these 12 models based on a single phylogenetic tree, comparisons were based on a PhyML maximum likelihood tree (Guindon & Gascuel, 2003) estimated for each model. Based on all model selection criteria, the general time-reversible model with gamma distribution of remaining sites (GTR + G) was chosen. Stem regions were modelled according to the doublet model available in MrBayes which includes a rate heterogeneity term.

A preliminary Bayesian inference (BI) analysis using MrBayes software revealed that the Markov Chain Monte Carlo (MCMC) steady state was reached after less than 50 000 generations. A conservative burn-in of 250 000 generations was chosen and a full analysis of 750 000 generations was carried out with sampling every 1000 generations, resulting in 1000 trees from two independent runs. The potential scale reduction factor (PRSF) values of all 35 parameters were less than 1.11 (31 had values < 1.03) suggesting good convergence (i.e. less than a PRSF threshold of 1.2 as suggested by Gelman et al., 1995) of the two runs. Another convergence diagnostic, the standard deviation of split frequencies between simultaneous runs, was close to zero (< 0.03), confirming convergence. The tree was rooted with Geosiphon pyriformis, Paraglomus brasilianum and Paraglomus occultum.

Statistical data analyses

Effects of forest management intensity (stand type), host plant species identity and sampling time on the number of AM fungal taxa in root samples were estimated using linear mixed models with the residual maximum likelihood method (REML). Forest management (two levels), plant species (three levels), and sampling time (three levels) were included as fixed factors and site (three levels) as a random factor nested in forest management type. Raw counts were analysed. A square root transformation improved the residual plot characteristics only slightly and did not affect significance patterns.

The variation in AM fungal community composition was analysed by principal coordinates analysis (PCoA, or metric multidimensional scaling). PCoA is a multivariate technique that allows placing of nonmetric or semimetric distances (sample similarities) into Euclidean space, so that a linear ANOVA model can be applied to the obtained PCoA axis values (see Legendre & Anderson, 1999). Sample similarity matrixes were calculated using ‘ecological’ (1 – |xi – xj|/range unless xi = xj = 0) and Jaccard (if xi = xj = 1, then 1; if xi = xj = 0, then 0; if xi ≠ xj, then 0) similarity measures for relative abundance and presence/absence data, respectively. Relative abundances of fungal taxa were obtained by dividing the number of clones belonging to the given taxon by the number of clones sequenced in the sample. The experimental factors (forest management type, site, host plant species, and sampling time) were visualized on the plots of PCoA axis scores using different colours. As no clear patterns were observed, no further statistical analyses were applied.

The relations of the environmental variables (ISF, DSF, soil inline image-N and inline image-N, P, pH and DOC), cover of bryophytes and cover and richness of vascular plants in a 1 × 1 m subplot with the number of AM fungal taxa or the fungal community composition in root samples were assessed by plotting the values for the number of AM fungal taxa or scores of the first five PCoA axes against the above variables, again visualizing the experimental factors using colours. Again, no clear patterns were observed and no further statistical analyses were applied. All above statistical analyses were implemented in GenStat version 10 using only samples from the three plant species (F. vesca, H. nobilis and O. acetosella) that occur in all study sites; site-specific G. luteum and T. pratense were not included in these analyses.

The effect of sampling effort on fungal taxon accumulation was assessed by calculating the number of detected fungal taxa (Sobs) as a function of the number of samples using EstimateS version 7.5.1 (Colwell, 2005) based on presence/absence of fungal taxa in individual samples of the five plant species.


AM fungal taxa

We recorded 34 AM fungal SSU rDNA taxa in the roots of five host plant species (90 plant samples; 911 clones sequenced) in the Koeru boreal forest. These taxa comprised: five Acaulospora, two Glomus group C, and 23 Glomus s. str. sequence groups (Fig. 1, Table 2). Eleven sequence groups clustered with a known species or isolate, with two of these not having been registered from environmental samples (roots) previously; 15 groups have been previously detected from plant roots, but were not represented in sequence databases by any known species or isolate; eight groups represented previously unknown taxa. Altogether 23 sequence groups detected in this study have been previously recorded from environmental samples.

Figure 1.

Glomeromycota sequences detected in this study and database sequences of known Glomeromycota and of those from environmental samples of the small-subunit ribosomal RNA gene (SSU rDNA) fragment between the NS31 and AM1 primers. A Bayesian analysis with a general time-reversible model with gamma distribution of remaining sites (GTR + G) is shown.

Table 2.  Number of clones of arbuscular mycorrhizal (AM) fungal taxa detected in studied plant species in old and young forest stand types
Plant speciesG3G13G7G4G5G20G33G18G2G21G27G17A5G30G15G26G28A6A7GC1GC2G16A3G24G25G31G22G10G19G32A4G11G23G29Total
  1. Fungal taxa are ordered by frequency of occurrence across all samples. FV, Fragaria vesca; GL, Galeobdolon luteum; HN, Hepatica nobilis, OA; Oxalis acetosella; TP, Trifolium pratense. Fungal taxon codes are as in Fig. 1, excluding the MO prefix.

Old stands
FV 57 21  0 620 2 016 0 7 0 5 0 0 10110220010000000000142
GL 19 14  8 415 8 0 0 0 0 0 0 0 0 0010000100000202010075
HN 30  9 36 9 4 8 5 5 1 0 0 0 9 0 33100103201100000000131
OA 21 19 12 315 6 0 420 4 0 0 0 0 25005210003000000000122
Young stands
FV 32 28  8 0 0 1 0 0 0 2 0 0 1 4 0000000000000000100077
HN 31 20 2620 01618 2 2 111 0 2 8 40340021130200200001180
OA 59 25 13 9 0 2 1 0 0 1 3 6 0 0 00100000000030000010124
TP 25 15  5 3 0 2 4 0 1 0 1 2 0 0 0000000010010000000060

Variation in AM fungal taxon richness

In total we identified 26 AM fungal taxa from the roots of H. nobilis (31 individuals), 20 taxa from F. vesca (20), 21 taxa from O. acetosella (26), 11 taxa from T. pratense (7) and 11 taxa from G. luteum (6). Sampling effort curves (Fig. 2) indicate a trend for the actual number of fungal taxa associated with H. nobilis to be higher than that of other host species. One can expect that more samples of T. pratense and G. luteum would have revealed more fungal taxa associated with these plant species, as the curves follow those of F. vesca and O. acetosella, showing higher total numbers of AM fungal taxa.

Figure 2.

Expected arbuscular mycorrhizal (AM) fungal taxon accumulation curves (Mao Tau) of the studied plant species. One may observe that a larger number of root samples could have increased the number of fungal taxa detected in the host species Galeobdolon luteum and Trifolium pratense. Plant species codes: FV, Fragaria vesca; GL, Galeobdolon luteum; HN, Hepatica nobilis, OA; Oxalis acetosella; TP, Trifolium pratense.

A mean of 3.17 (± 0.24; SE of the estimate) AM fungal taxa colonized a root sample for the three abundant plant species (F. vesca, H. nobilis and O. acetosella). The effect of plant species identity on the number of fungal taxa per plant individual was marginally nonsignificant (P = 0.056). The estimated mean number of fungal taxa per sample of H. nobilis (3.68 ± 0.31; mean ± SE of the estimate) was higher than that per sample of O. acetosella (2.69 ± 0.34), but did not differ significantly from that per sample of F. vesca (3.15 ± 0.38). Forest management intensity (stand type), sampling time, and their interaction had no significant effect on the AM fungal taxon richness in samples. We could not detect any relations between the number of AM fungal taxa and explanatory environmental variables (data not shown). The two stand-specific plant species, G. luteum and T. pratense, were not included in this model; a mean of 3.5 (± 1.38, SD) and 3.71 (± 1.60) AM fungal taxa were observed per sample for these species, respectively.

Variation in AM fungal community composition

Principal coordinate analysis (PCoA) of fungal community composition based on presence/absence and relative abundance data for AM fungi in root samples did not yield obvious groupings of samples; neither were there patterns relating to forest management type, site, plant species or sampling time. The percentage variation described by the first five PCoA axes was 39.91 and 37.64% for presence/absence and relative abundance data, respectively (Table 3). We could not detect any relations between the fungal community composition and explanatory environmental variables. Example plots of first PCoA axis against plant cover, plant species richness, soil P and N content, soil pH and light availablity (ISF) are presented in Supplementary Material Fig. S1.

Table 3.  Percentage variation explained by the first five axes of principal coordinate analyses of arbuscular mycorrhizal fungal community composition in root samples of Fragaria vesca, Hepatica nobilis and Oxalis acetosella using relative abundance or presence/absence data
Relative abundance14.057.036.495.684.39

Five of the 34 detected AM fungal taxa occurred in the roots of all studied plant species and in all plots: Glomus sp. MO-G3 (in 60% of samples/30% of clones, related to the Glomus intraradices group), Glomus sp. MO-G13 (60/17%, related to Glomus vesiculiferum), Glomus sp. MO-G4 (30/6%), Glomus sp. MO-G7 (30/12%, related to Glomus hoi), and Glomus sp. MO-G20 (20/5%). G4 and G7 were half as frequent (in terms of the proportion of samples colonized) in spring as in summer and autumn (data not shown). Eight fungal taxa were detected from one host species only, but were all represented by only one or two clones or samples. Seven and eight taxa were detected from young and old stands only, respectively. Most frequent among these, occurring in > 5% of root samples, were Glomus sp. MO-G27 (young stand), Glomus sp. MO-G5 (old stand), and Glomus sp. MO-GC1 (old stand).


Communities of AM fungi colonizing the roots of five understorey plant species in an Estonian herb-rich boreal coniferous forest were found to be remarkably rich. We recorded 34 AM fungal taxa in total, comparable to the fungal richness described in tropical rain forests in Panama (Husband et al., 2002a,b) and higher than that in temperate grassland and broad-leaved forest locations (Vandenkoornhuyse et al., 2002; Saito et al., 2004; Helgason et al., 2007). To our knowledge the only other boreal or temperate forest systems where AM fungal community dynamics has been studied are a Scots pine forest in Estonia, a Norway spruce forest in Germany, a warm-temperate broadleaved forest in Japan, and a broadleaved forest in the UK (Helgason et al., 2002, 2007; Öpik et al., 2003; Wubet et al., 2003; Yamato & Iwase, 2005).

The number of samples in our study is higher than (90 vs 20–54), and the number of sequenced clones comparable to or lower than (911 vs 558–2001), those in the studies cited above. Forty-one taxa were reported from roots of three host species from tropical forest and pasture locations in Costa Rica, although forest-specific richness cannot be deduced from these data (Aldrich-Wolfe, 2007). The number of root samples processed can significantly affect the number of root-colonizing AM fungi detected (Öpik et al., 2006a). Furthermore, because of the patchy distribution of colonization units, small subsamples of a root system can contain different AM fungi (Öpik et al., 2006b). Here the individual samples were large (20 cm length) but contained on average three AM fungal taxa in an ecosystem supporting at least 34 taxa. This indicates a trade-off between the number of samples and the size of a root sample needed to reasonably describe the diversity of root colonizers in an ecosystem, an issue that needs to be addressed in future field surveys. In conclusion, care is required when comparing AM fungal richness data acquired with different methodologies, including differences in sample number, screened/sequenced clone number, spatiotemporal sampling design, and number of sampled plant species.

In this study we identified eight previously undescribed AM fungal taxa and 23 taxa that are known from molecular analysis of plant roots only. This high proportion of ‘known-as-sequence-only’ taxa reflects the accumulation of molecular diversity data of AM fungi as a result of the increasing number of studies of Glomeromycota in natural ecosystems. More matches with known fungal species would be expected if or when more intensive sequencing, of relevant genome regions, of identified isolates maintained in culture collections occurs.

The taxon richness of AM fungi per root sample was higher for H. nobilis than for F. vesca and O. acetosella. However, there were no differences among forest management types or seasons (sampling times). It has been shown that co-occurring plant species can be colonized by AM fungal communities of different composition (Helgason et al., 2002; Vandenkoornhuyse et al., 2002), but differences in the number of fungal taxa associated with host plant species have not been previously demonstrated. We can hypothesize that these trends are linked to the host preferences of AM fungi, different symbiont ranges of AM plant hosts, or different sizes of fungal colonization units in roots resulting in variable fungal taxon densities. Variable symbiont ranges of plants could be related to plant functional types, for example life forms, plant growth rates, rooting traits and types of clonal growth. However, why fungi or plants should ‘prefer’ one host to another requires further research.

There is evidence that disturbance can decrease AM fungal taxon richness (Helgason et al., 1998; Whitfield et al., 2004). Therefore, we expected to observe smaller numbers of AM fungal taxa in the more intensively managed young forest stands. However, these ecosystems have preserved a high richness and the common management practices have not had an adverse impact on the AM fungal biodiversity. In contrast to the findings of Helgason et al. (1998) and Whitfield et al. (2004), habitats with moderate management intensities may still display rather high numbers of AM fungal taxa (Hijri et al., 2006; Vallino et al., 2006). In the studied ecosystem the soil disturbance associated with clearcut logging and planting of tree saplings is less intense than that associated with recurrent ploughing, which in combination with no soil disturbance during subsequent years may aid maintenance of fungal diversity through management activity. Thus we propose that the severity and recurrence of disturbance events influence the magnitude of the reaction of AM fungal communities following disturbance. Furthermore, it is worth noting the lack of clear management- or environment-related patterns of AM fungal communities in soil despite the obvious differences among these forest ecosystems apparent to the naked eye. The (lack of) variability in diversity of soil micro-organisms in relation to common disturbances and natural environmental gradients warrants further investigation.

The two most common fungi in our study, Glomus sp. MO-3 and MO-13, are related to the G. intraradices group, which has been detected from world-wide locations of both stable and disturbed ecosystems (Öpik et al., 2006a) and many host species (Helgason et al., 2007). Glomus intraradices is sometimes considered to be an aggressive species. It can depress plant growth even if it provides all the P acquired by the plant (Smith et al., 2003). Glomus intraradices isolates from the same or geographically distant locations can affect plant growth differentially (Hart & Reader, 2002; Koch et al., 2006). Such variation may be explained if this group contains several functionally different taxa as proposed by van der Heijden et al. (2004). Thus, it is reasonable to hypothesize that the G. intraradices species group, as identified using the NS31/AM1 primer pair, contains several cryptic taxa with differences in various ecological properties such as disturbance tolerance, mycelial growth, root colonization rate and sporulation traits. Even if it is considered as a group of multiple taxa, this is the most common group in both the studied boreo-nemoral forest and many ecosystems world-wide, and deserves further attention.

The third most common fungus in the studied ecosystem was G. hoi (here Glomus sp. MO-G7), detected in one-third of samples and all plots and host species. A host specificity of G. hoi towards Acer pseudoplatanus has been demonstrated in one ecosystem (Helgason et al., 2002), but otherwise the taxon appears to be widespread with no specialization in regard to habitat type or host species (Öpik et al., 2006a; Helgason et al., 2007).

An old stand-specific taxon, Glomus sp. MO-G5, has been previously reported from forests and grasslands, but not from disturbed habitats (Öpik et al., 2006a), and from 11 host species (as Glo2; Helgason et al., 2007). MO-G5 was dominant in experimental plants inoculated with boreal Scots pine forest soil (Öpik et al., 2003), and in natural grassland habitats (Öpik et al., 2006a). Here, the taxon was detected in c. 10% of samples.

As regards the methodology, it is obvious that the SSU rDNA gene region used in this study does not separate all Glomeromycota groups very well because of limited nucleotide variation among some clades, including the G. intraradices and G. caledonium clades (Fig. 1). Investigations of mitochondrial large subunit (LSU) sequences, including that of G. intraradices, suggest that better marker regions with more interspecific variability might be available (Raab et al., 2005). However, these have not been rigorously tested on a range of related taxa, on isolates of the same taxa and on environmental samples. Apart from these imperfections the SSU region provides comparability of data for ecological studies as a result of the accumulation of database entries and publications based on this gene region (Öpik et al., 2006a).

In conclusion, the observed unexpectedly high richness of Glomeromycota in a temperate coniferous forest indicates the need to obtain comparable descriptive soil fungal community data from a more diverse range of ecosystems. This almost unique richness of Glomeromycota could be speculatively attributed to the relative stability of the ecosystem and/or a high diversity of host species. Comparative data from a range of ecosystems along disturbance and plant richness gradients and from a range of plant hosts would help to test these hypotheses. The described taxon richness patterns and apparent lack of taxon composition patterns deserve further evaluation in order to establish the role of host plant identity, plant species richness, light availability, and soil conditions as determinants of Glomeromycota taxon distribution at a small scale. Furthermore, it is essential to evaluate the observed diversity patterns in functional terms.


MÖ received short-term scholarship from the European Molecular Biology Organisation (EMBO) and Kristjan Jaak scholarship from the Archimedes Foundation (Estonia) for visits to SCRI, UK. The study was supported by Estonian Science Foundation grants 6533, 7366 and SF0180098s08, EU FP6 integrated project ALARM (GOCECT-2003-506675) and EU Marie Curie Fellowship grant MEIF-CT-2005-024657 (MÖ). TJD acknowledges the support of the Scottish Government Rural and Environment Research and Analysis Directorate (RERAD). We are grateful to Lauri Laanisto and Eve Eensalu (UoT) for their help with taking fish-eye photographs and calculating the light parameters of the study sites, and to James McNicol (BIOSS) for help with statistical analyses.