Molecular diversity of arbuscular mycorrhizal fungi in Prunus africana, an endangered medicinal tree species in dry Afromontane forests of Ethiopia

Summary

• • 
  The molecular diversity of arbuscular mycorrhizal (AM) fungi colonizing roots of Prunus africana and of AM fungal spores obtained from baiting cultures of indigenous soils from two dry afromontane forests of Ethiopia was investigated.
• • 
  The internal transcribed spacer (ITS) rDNA region from colonized roots and single spores of three AM fungal spore types was amplified, cloned and sequenced using AM fungal specific primers.
• • 
  Phylogenetic analysis using the 5.8S rDNA data set revealed that 109 of the sequences obtained belong to members of the Glomeromycota. Subsequent 5.8S/ITS2 rDNA sequence analysis indicated high AM fungal diversity and dominance of Glomus types. Twenty sequence types belonged to the Glomeraceae and one each to the Diversisporaceae and Archaeosporaceae. Two of the three spore types were identified as Glomus etunicatum and Glomus mosseae.
• • 
  Twenty of the AM fungal types identified are new to Ethiopia and to science. The AM fungal community differed between the two sites studied.

Introduction

Prunus africana (syn. Pygeum africanum) was once a geographically widespread tree restricted to Afromontane forest islands in mainland Africa from Ethiopia to South Africa, and from the west coast to the outlying island of Madagascar (Kalkman, 1965; White, 1983). Prunus africana is an important multipurpose tree species (Azene et al., 1993; Cunningham & Mbenkum, 1993; Cunningham et al., 1998); its bark is the only source of an important drug, which is used in the treatment of benign prostatic hyperplasia (Bombardelli & Morazzoni, 1997; Marandola et al., 1997). The increasing market for P. africana remedies intensifies destructive bark harvesting and endangers the long-term sustainability of harvesting, and the conservation of P. africana and associated Afromontane species. As a result, P. africana was added to Appendix II of the Convention on International Trade in Endangered Species (CITES) (Cunningham & Mbenkum, 1993; Cunningham et al., 1998).

In Ethiopia, P. africana is an important component of the dry Afromontane vegetation (Tamrat, 1993; Demel, 1996) and is suffering from overexploitation to satisfy the increasing demand associated with the growing population, leading to devastating consequences for the forest ecosystem. Prunus africana is an ideal species for reafforestation and enrichment plantations. Incorporation into agroforestry systems has been suggested as a solution to decrease the pressure on natural forests (Cunningham & Mbenkum, 1993). However, problems associated with establishment after transplanting need to be solved. Although wild P. africana was demonstrated to be arbuscular mycorrhizal (AM) (Wubet et al., 2003a), information on the diversity of colonizing AM fungal species and the prospects for exploiting them for production of mycorrhizal seedlings is still lacking.

Arbuscular mycorrhizal fungi form intimate symbiotic associations with most land plants (Smith & Read, 1997). Plant diversity and productivity in ecosystems are significantly influenced by AM fungal diversity in the soil (van der Heijden et al., 1998). It is also known that different AM fungi can simultaneously colonize a single root segment, and that the species are functionally diverse (Merryweather & Fitter, 1998; Smith et al., 2000). Arbuscular mycorrhizal fungi that specifically colonize roots of a plant species are therefore likely to be of great relevance to function, and their identification is important to understand the ecology of plant–fungus interactions in natural ecosystems (Helgason et al., 2002). Furthermore, the growing evidence on improvement of plant survival and growth enhancement through specific mycorrhizal inoculation necessitates knowledge of the diversity of the colonizing AM fungal species in specific natural ecosystems of a plant species of interest.

Recently, AM fungi have been separated from the polyphyletic Zygomycota and placed in the new phylum Glomeromycota (Schüßler et al., 2001). Identification of AM fungi has relied extensively on the morphology of spores and related structures. But when spores are isolated from the soil and identified, they may not necessarily correlate with the diversity of species colonizing plant roots (Helgason et al., 1999). Also, some species of mycorrhizal fungi may not form spores. When only hyphal structures are present, it is generally possible to identify the respective fungi only to family or genus (Merryweather & Fitter, 1998). In addition, members of some families cannot be detected by standard staining methods (Morton & Redecker, 2001). Applications of PCR-based molecular methods with isolate- or group-specific primers, however, have made possible the identification and analysis of the diversity of AM fungi in planta. Nested PCR procedures with species-specific primers have been used successfully for known fungal species from pot cultures (Abbas et al., 1996; van Tuinen et al., 1998; Chelius & Triplett, 1999; Jacquot et al., 2000; Millner et al., 2001). Studies of field-collected materials also demonstrated the potential of these methods for analysing the taxonomic diversity of AM fungi actively colonizing roots (Clapp et al., 1999; Helgason et al., 1999; Daniell et al., 2001; Kjøller & Rosendahl, 2001; Turnau et al., 2001; Wubet et al., 2003b).

In this study we investigated the diversity of AM fungal species in roots of P. africana collected from two dry Afromontane forests of Ethiopia, using molecular methods. The results provide the first indication of the diversity and species composition of AM fungi in these forests.

Materials and Methods

Study site

The study was carried out in Menagesha and Munessa-Shashemene forests located in the dry Afromontane forest vegetation of Ethiopia, in the north-west and south-west highlands, respectively, at an altitudinal range of 1500–2700 m, with an average annual temperature between 14 and 20°C and annual rainfall between 700 and 1100 mm (Tamrat, 1993; Demel, 1996). Details of the study sites are presented by Wubet et al. (2003a).

Root sampling

In both study sites five single-standing adult plants were randomly selected, and fine roots from each tree were excavated from at least three laterals, starting from the trunk and working out towards the fine roots. Roots were then pooled and washed under running tap water to remove attached soil and debris. Samples were kept at −20°C in 1.5 ml reaction tubes for subsequent molecular analysis.

AM fungal trap culture

Samples from the soil layers where fine roots of P. africana were collected were mixed with autoclaved, medium-sized sand (1 : 4) and poured in 1 l pots. After 24 h incubation with water, Tagetes erecta L. was sown as a trap plant at a density of five seeds per pot. The 5 × 2 pots were randomly placed in a climate chamber with 75% humidity for 12 h at 22°C and an irradiance of 70 µE m⁻² s⁻¹, and 12 h at 18°C in darkness. After seedling germination, pots were watered twice a week to keep 60% of water-holding capacity (WHC). The pot cultures were run for 6 months. During the last month the moisture regime was successively lowered to 30% of WHC to stop plant growth and enhance sporulation of the existing AM fungal species.

Spore isolation and DNA extraction from spores

Spores were isolated from the trap cultures by wet sieving, and grouped based on similar external features such as spore colour, size, shape, visible contents, and shape of the subtending hyphae (Brundrett et al., 1996). Three spore types were identified, and DNA was extracted from single spores of each spore type by crushing a spore in 10 µl PCR buffer with sterile needles, 5 µl of which were then used directly as template for the first PCR amplification.

DNA extraction from roots colonized by AM fungi

Five mycorrhizal root samples (1–2 cm) from each tree were air dried and placed in 1.5 ml reaction tubes together with a tungsten carbide ball (3 mm) and ground (3 min, 13 000 r.p.m.) using a mixer mill (MM 300, Retsch, Haan, Germany). DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's protocol for the isolation of DNA from plant tissues. However, only one elution step with 100 µl of the elution buffer was used.

Nested PCR amplification

The target region for the PCR experiments was the internal transcribed spacer (ITS) region of the nuclear rDNA including the pseudogene coding for the 5.8S ribosomal subunit. The first fungal DNA amplification was performed using the universal primer pair NS5 and ITS4 (White et al., 1990). DNA was amplified in 50 µl reaction volume containing 5 µl 10× PCR reaction buffer, with concentrations of 0.2 mm of each dNTP and 0.5 µm of each primer, including 2.5 units of Taq polymerase and 5 µl of DNA template. The PCRs were done using a Gene-Amp PCR System 2400 (Perkin-Elmer Corporation, Norwark, CT, USA) as described by Redecker (2000): after a hot start at 60°C, an initial denaturation of 3 min at 95°C was followed by five cycles of 30 s at 95°C, 30 s at 52°C and 1.5 min at 72°C. Thereafter, 30 cycles were performed, similar to the first five cycles but with an annealing temperature of 51°C followed by a final extension period of 7 min at 72°C. A 5 µl aliquot of each PCR product was analysed in electrophoresis using a 1.5% agarose gel which was stained with ethidium bromide (0.5 µg ml⁻¹) and photographed under UV light. A 1–3 µl aliquot of the first PCR product was either used directly or diluted (1 : 10–1 : 100 v/v) as template for the nested PCR amplification using the specific primers ACAU1660, ARCH1311, GLOM1310 and LETC1670, each in combination with ITS4, or GIGA5.8R in combination with NS5. Primers and PCR conditions were as reported by Redecker (2000). The nested PCR products were then analysed.

Cloning and sequencing

Nested PCR products were purified following the QIAquick protocol (Qiagen). Cloning was done using the pCR 2.1-TOPO vector system (Invitrogen Corporation, Carlsbad, USA) according to the manufacturer's instructions. Individual transformed colonies (20–30) were picked and used directly in a PCR reaction with the M13 forward and M13 reverse primers. The PCR was carried out in 25 cycles (94°C for 1 min, 55°C for 1 min and 72°C for 1 min). Five of the positive amplified PCR products were then purified as described above, and used in cycle sequencing with the M13 PCR primers as sequencing primers using the ABI PRISM Big Dye Terminator Cycle Sequencing Ready Reaction Kit v.2 (PE Applied Biosystems, Warrington, UK) according to the manufacturer's protocol, except that reaction volumes were reduced by half and the kit was diluted 1 : 1 (v/v) with double distilled water. Electrophoresis and data sampling were performed on an automated sequencer (ABI 373 A Stretch, Applied Biosystems, Foster City, CA, USA). Sequence editing was done using the program sequencher version 3.1.1 (Gene Codes Corporation, Ann Arbor, MI, USA). The sequences have been deposited at the National Center for Biotechnology Information (NCBI) GenBank (http://www.ncbi.nlm.nih.gov) under accession numbers AY236228–AY236336.

Phylogenetic analysis

Sequence similarities were determined using the BLAST sequence similarity search tool (Altschul et al., 1997) provided by GenBank. Sequences were aligned with other published glomeralean sequences using the program ClustalX (Thompson et al., 1997), and further visual alignment was done in Se-Al version 2.03a (Rambaut, 1996). Phylogenetic analysis was carried out to estimate the phylogenetic relationships between the sequences obtained in this study and those of named AM fungi from GenBank. We restricted this first analysis to the 5.8S rDNA region from representative sequences obtained during this study; the corresponding closest matches from GenBank; and sequences representing the seven AM fungal genera. More detailed phylogenetic analyses, using broader samplings for the individual groups detected in the 5.8S analysis, were carried out based on the 5.8S and parts of the ITS2 regions after exclusion of ambiguous positions in the respective alignments.

Phylogenetic relationships were estimated with a Bayesian approach using Markov chain Monte Carlo (MCMC) as implemented in the computer program MrBayes version 3.0 (Huelsenbeck & Ronquist, 2001). In this approach, Monte Carlo Markov chains of trees are constructed which, after stationarity is reached, can be used to approximate the a posteriori probability that groups of taxa are monophyletic given the DNA alignment (the probability that corresponding bipartitions of the species set are present in the true unrooted tree including the given species). The power of this method to reconstruct phylogenetic relationships efficiently has recently been demonstrated by Murphy et al. (2001) for mammalian phylogeny; Kauff & Lutzoni (2002) for ascomycetes; Aanen et al. (2002) and Maier et al. (2003) for basidiomycetes; Wubet et al. (2003b) for glomeromycetes; and also in simulation studies (Alfaro et al., 2003). Four incrementally heated simultaneous Monte Carlo Markov chains were run over 1 000 000 generations using the general time-reversible (GTR) model of DNA substitution, additionally assuming a proportion of invariable sites with gamma-distributed substitution rates of the remaining sites (see Swofford et al., 1996), random starting trees, and default starting parameters of the DNA substitution model. Trees were sampled every 100 generations, giving a total of 10 000 trees. Stationarity of the Markov chains was reached before 1000 trees had been sampled. Thus the first 1000 trees were discarded and the remaining 9000 sampled trees were included in a 50% majority rule consensus tree of each run (also including compatible groups of lower frequencies) to obtain estimates for the a posteriori probabilities. Branch lengths were averaged over the sampled trees. To test the reproducibility of the results, ensuring representative tree sampling (Huelsenbeck et al., 2002), computations were repeated five times using random starting trees and default starting values for the model parameters. One of the respective consensus trees is presented in Figs 1–6.

Figure 1. Phylogenetic relationships of arbuscular mycorrhizal (AM) fungi colonizing Prunus africana (CP), AM fungal spores (SP) and related glomeralean fungi. Bayesian Markov chain Monte Carlo analysis (MCMC) of an alignment of 5.8S rDNA sequences using the general time-reversible model of DNA substitution and assuming a proportion of invariable sites with gamma-distributed substitution rates of the remaining sites, random starting trees, and default starting parameters of the DNA substitution model. Bold numbers on branches are estimates of a posteriori probabilities that the respective groups are monophyletic given the alignment; regular numbers are neighbour-joining bootstrap values from 1000 replications using Kimura two-parameter genetic distances. Branch lengths represent average values over the trees sampled in MCMC analysis, and are scaled in terms of expected numbers of nucleotide substitutions per site. The topology was rooted with Endogone pisiformis. Bold characters designate representatives of AM fungal sequences from this study. GlGr A–C represent groups of the genus Glomus according to Schwarzott et al. (2001).

Figure 2. Phylogenetic relationships of arbuscular mycorrhizal (AM) fungi colonizing Prunus africana and related glomeralean fungi restricted to the Glomus II cluster. Bayesian Markov chain Monte Carlo analysis of an alignment of 279 bp spanning the 5.8S rDNA and a portion of the ITS2 region was carried out as described in Fig. 1. The topology was rooted with Glomus clarum. Bold characters designate AM fungal sequences from this study.

Figure 3. Phylogenetic relationships of arbuscular mycorrhizal (AM) fungi colonizing Prunus africana, AM fungal spores and related glomeralean fungi restricted to the Glomus III cluster. Bayesian Markov chain Monte Carlo analysis of an alignment of 354 bp spanning the 5.8S rDNA and a portion of the ITS2 region was carried out as described in Fig. 1. The topology was rooted with Glomus mosseae and Glomus coronatum. Bold characters designate AM fungal sequences from this study.

Figure 4. Phylogenetic relationships of arbuscular mycorrhizal (AM) fungi colonizing Prunus africana and related glomeralean fungi restricted to the Glomus IV cluster. Bayesian Monte Carlo analysis of an alignment of 289 bp spanning the 5.8S rDNA and a portion of the ITS2 region was carried out as described in Fig. 1. The topology was rooted with Glomus sp. 7.6. Bold characters designate AM fungal sequences from this study.

Figure 5. Phylogenetic relationships of arbuscular mycorrhizal (AM) fungi colonizing Prunus africana, AM fungal spores and related glomeralean fungi restricted to the Glomus V cluster. Bayesian Monte Carlo analysis of an alignment of 332 bp spanning the 5.8S rDNA and a portion of the ITS2 region was carried out as described in Fig. 1. The topology was rooted with Glomus intraradices. Bold characters designate AM fungal sequences from this study.

Figure 6. Phylogenetic relationships of arbuscular mycorrhizal (AM) fungal spores and related glomeralean fungi restricted to the Glomus VI cluster. Bayesian Monte Carlo analysis of an alignment of 353 bp spanning the 5.8S rDNA and a portion of the ITS2 region was carried out as described in Fig. 1. The topology was rooted with Glomus coronatum. Bold characters designate AM fungal sequences from this study.

With all alignments that were used in Bayesian phylogenetic analysis, we also ran neighbour-joining analyses (Saitou & Nei, 1987) using Kimura two-parameter genetic distances (Kimura, 1980) combined with bootstrap analysis (Felsenstein, 1985) from 1000 replicates using the program paup version 4.0b10 (Swofford, 2002).

Taking into account the documented ITS sequence diversity among spores of the same species of AM fungi, and even from single spores (Sanders et al., 1995; Lloyd-Macgilp et al., 1996; Hijri et al., 1999; Lanfranco et al., 1999; Pringle et al., 2000; Jansa et al., 2002), and based on the similarity among sequences from the same AM fungal species available in GenBank (data not shown), we treated sequences with a similarity value of more than 93% of the whole ITS + 5.8S as belonging to the same AM fungal type.

Statistical analysis

Diversity indices of AMF sequence types were computed using the program dindex (version 4.0, http://www.geocities.com/brijeshcm/dindex). We applied a χ² test of independence to infer whether or not the composition of the AM fungal communities differed between the two study sites.

Results

Phylogeny and AM fungal diversity

BLAST search results of ITS sequences obtained from P. africana roots and AM fungal spores isolated from trap cultures indicated that 109 sequences belonged to members of the Glomeromycota, despite the fact that, for the most part, they were only distantly related to the available AM fungal sequences in the NCBI database.

In all Bayesian phylogenetic analyses, we observed a high reproducibility of the results of different runs from the same alignment. There were no discrepancies between phylogenetic hypotheses of the Bayesian analysis on the one hand and neighbour-joining analysis on the other, only the resolution obtained from Bayesian phylogenetic analysis was generally higher than those obtained with neighbour joining. The phylogenetic analysis based on the 5.8S rDNA region from representative sequences obtained during this study; the corresponding closest matches from GenBank; and sequences representing the seven AM fungal genera revealed that our new sequences belong to seven different 5.8S rDNA sequence groups of glomeralean fungi: six belonging to Glomus, grouped into Glomus I–VI, and one belonging to Archaeospora (Fig. 1).

The Archaeospora sequence type (Arch1, with three clone sequences) is distantly related to the sequences of Archaeospora gerdemannii and Archaeospora leptoticha in GenBank, which is also supported by a posterior probability of 100%. The Glomus I group, including one sequence type (Glom1, with four clone sequences), clustered together with Glomus versiforme sequences from GenBank. This group belongs to the family Diversisporaceae (Fig. 1). Comprehensive phylogenetic analyses, based on the 5.8S rDNA and parts of the ITS2 regions of AM fungi within the five Glomus groups that belong to the family Glomeraceae, demonstrated details of their phylogenetic relationships.

The results of the phylogenetic analyses of 5.8S/ITS2 sequences are consistent with our classification of 12 AM fungal sequence types (Glom2 to Glom13) in the Glomus II group (Fig. 2); three AM fungal types (Glom14 to Glom16) in the Glomus III group (Fig. 3); and three AM fungal types (Glom17 to Glom19) in the Glomus IV group (Fig. 4), and show that these sequence types are distinctly different from the related sequences in the database. On the other hand, the respective analyses demonstrate that the AM fungal type Glom20 of the Glomus V group belongs to Glomus etunicatum (Fig. 5), while the AM fungal type Glom21 of the Glomus VI group belongs to the genetically diverse Glomus mosseae complex (Fig. 6).

Of the 22 AM fungal types obtained in this study, 20 belong to the Glomeraceae and one each to the Diversisporaceae and Archaeosporaceae, respectively. No representative AM fungal type was found from the Acaulosporaceae, Paraglomeraceae or Gigasporaceae. For 20 of our AM fungal types, sequence similarity with sequences of identified AM fungi available in GenBank was too low for us to relate them precisely to either morphological or biological species.

Primer specificity

Problems of primer specificity were observed (Table 1). AM fungal types belonging to the Archeosporaceae (Arch1, Fig. 1) and Diversisporaceae (Glom1, Fig. 1) were amplified with the primers LETC1670 and ACAU1660, respectively. A number of AM fungal types belonging to the Glomeraceae were amplified with ACAU1660 and ARCH1311 designed specifically to amplify members of the Acaulosporaceae and Archeosporaceae, respectively.

Table 1.  Arbuscular mycorrhizal (AM) fungal sequence types, specific primers used for the nested PCR amplification, sample tree or AM fungal spore with the respective number of clones, and the study site where the sequences were obtained (ME, Menagesha; MS, Munessa-Shashemene)

AM fungal type	Nested PCR primer	Sample tree or AMF spore (number of clones)	Study site
Arch1	LETC1670	Pa01 (1), Pa03 (2)	ME
Glom1	ACAU1660	Pa07 (4)	MS
Glom2	ACAU1660	Pa08 (4)	MS
	ARCH1311	Pa07 (2), Pa08 (3)	MS
	GLOM1310	Pa01 (3), Pa08 (2)	ME and MS
	LETC1670	Pa07 (3), Pa08 (2)	MS
Glom3	ACAU1660	Pa09 (1)	MS
	LETC1670	Pa06 (1)	MS
Glom4	ARCH1311	Pa06 (1)	MS
	LETC1670	Pa06 (1)	MS
Glom5	ACAU1660	Pa02 (1), Pa06 (1)	ME and MS
	GLOM1310	Pa03 (1), Pa06 (1)	ME and MS
Glom6	GLOM1310	Pa05 (2)	ME
	LETC1670	Pa04 (2)	ME
Glom7	ACAU1660	Pa03 (2)	ME
	LETC1670	Pa05 (1)	ME
Glom8	GLOM1310	Pa03 (2)	ME
Glom9	ACAU1660	Pa03 (2)	ME
	GLOM1310	Pa06 (1)	MS
	LETC1670	Pa05 (1)	ME
Glom10	ARCH1311	Pa04 (1)	ME
Glom11	ACAU1660	Pa07 (2)	MS
	ARCH1311	Pa10 (1)	MS
	LETC1670	Pa10 (1)	MS
Glom12	ACAU1660	Pa09 (1)	MS
	ARCH1311	Pa03 (1)	ME
Glom13	ARCH1311	Pa07 (1)	MS
Glom14	ACAU1660	Pa04 (5), Pa08 (1)	ME and MS
	ARCH1311	Pa05 (4), Pa09 (2)	ME and MS
	LETC1670	Pa03 (3), Pa08 (2)	ME and MS
Glom15	ACAU1660	Pa06 (5), Pa07 (3)	MS
Glom16	ACAU1660	Spore001 (3)	ME and MS
Glom17	ACAU1660	Pa10 (1)	MS
Glom18	ACAU1660	Pa08 (1)	MS
	ARCH1311	Pa07 (1)	MS
Glom19	LETC1670	Pa08 (1)	MS
Glom20	ACAU1660	Pa01 (2), Pa06 (2)	ME and MS
	LETC1670	Pa01 (2), Pa06 (2)	ME and MS
		Spore002 (8)	ME and MS
Glom21	LETC1670	Spore003 (6)	MS

Distribution of AM fungal types

The AM fungal communities detected colonizing roots of P. africana at the two study sites are shown in Table 2. Of the 20 AM fungal types detected from colonized roots of P. africana, five in Menagesha and nine in Munessa-Shashemene, representing 25 and 45% of the total AM fungal types, respectively, were unique to the respective site. The overall fungal diversity and ‘species’ richness were higher in Munessa-Shashemene than Menagesha (Shannon diversity index, H = 2.30, richness, R = 15; and H = 2.14, R = 11, respectively). The composition of the AM fungal communities colonizing P. africana at the two study sites differed significantly (χ² = 49.49, 19 df, P < 0.001), indicating that the colonizing AM fungal community is influenced by site factors.

Table 2.  Number of clones of arbuscular mycorrhizal (AM) fungal types amplified from colonized roots of Prunus africana

	Arch1	Glom																			Total	Mean	SD
		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	17	18	19	20
Sequences were assigned to AM fungal types of Glomus (Glom) and Archaeospora (Arch) based on the similarity of the ITS + 5.85 rDNA sequences. Rows represent the sum from five sample trees within a study site. Study sites are identified as ME, Menagesha; MS, Munessa-Shashemene dry Afromontane forests of Ethiopia. The mean and SD of the number of AMF types detected per sample tree are indicated in the right-hand columns.
ME	3	0	3	0	0	2	4	3	2	3	1	0	1	0	12	0	0	0	0	4	38	3.8	0.71
MS	0	4	16	2	2	2	0	0	0	1	0	4	1	1	5	8	1	2	1	4	54	4.2	2.83
Total	3	4	19	2	2	4	4	3	2	4	1	4	2	1	17	8	1	2	1	8	92	4.0	0.71

AM fungal spores from three species were isolated from trap cultures. Sequences obtained from PCR products from single spores also fall into three AM fungal types: AM fungal types Glom16 and Glom20 were found at both sites, while Glom21 was detected only at the Munessa-Shashemene site. Among the three AM fungal spore sequences, only the Glom20 type was also amplified directly from colonized roots from both sites.

Discussion

Reliability of the molecular data

The combination of BLAST searches and subsequent thorough phylogenetic analysis of the clone sequences obtained in this study with closely related AM fungal sequences in GenBank using the 5.8S rDNA data set from a wide spectrum of organisms (Redecker et al., 1999; data not shown) ensures that both the sequences obtained in this study and the reference sequences taken from NCBI are of glomeralean origin. Thus our analyses should not suffer from distortions caused by sequences from other fungal groups that were mislabelled as glomeralean sequences in public databases, a problem affecting various molecular studies of the Glomeromycota (Clapp et al., 2002).

Phylogenetic position of AM fungal sequences

Based on the phylogenetic hypothesis inferred from the relatively short 5.8S rDNA data set, which is known to provide information to estimate higher-level phylogenetic relationships (Cullings & Vogler, 1998; Redecker et al., 1999), the AM fungal sequences obtained from P. africana roots were assigned to seven glomeralean groups, of which six were Glomus groups (Fig. 1). Our phylogenetic analysis based on 5.8S rDNA (Fig. 1) is consistent with the results of Schwarzott et al. (2001), suggesting that the genus Glomus is not monophyletic, but a heterogeneous assemblage of at least three different groups.

The dominance of Glomeraceae in the dry Afromontane forests is in accordance with our previous report based on morphological characters of the colonizing AM fungi (Wubet et al., 2003a), indicating that this observation is not just caused by bias of the molecular analyses, for example by primers that neglect some nonGlomus groups of glomeromycetes. Predominance of the Glomeraceae has also been reported for other tropical forests (Sieverding, 1991; Onguene & Kuyper, 2001; Husband et al., 2002a,b). Daniell et al. (2001) argued that Glomus types dominate the colonization of arable crops because Glomus species are better adapted to disturbed environments, where their high sporulation rates may enable them to recover more readily. In tropical forests considered to be very stable environments, however, spore-consuming tropical small mammals (Janos et al., 1995; Mangan & Adler, 1999) may contribute to the dispersal of Glomus species and their dominance within roots.

The phylogenetic position of AM fungal types Arch1 and Glom1 in the Archaeosporaceae and Diversisporaceae, respectively, was clarified in the 5.8S rDNA analysis. Subsequent phylogenetic analysis based on the 5.8S/ITS2 region of the distinct Glomus groups belonging to the Glomeraceae, including all the respective clone sequences together with an additional range of reference sequences, allowed elucidation of the phylogenetic relationships between the AM fungal types and sequences from the database. While assignment of Glom20 and Glom21 AM fungal types to G. etunicatum and G. mosseae, respectively, was possible, the remaining AM fungal types could not be assigned to particular morphospecies (Figs 2–5) for a number of possible reasons. First, ITS sequence diversity has been documented among spores of the same species of AM fungi, and even from single spores (Sanders et al., 1995; Lloyd-Macgilp et al., 1996; Hijri et al., 1999; Lanfranco et al., 1999; Pringle et al., 2000; Jansa et al., 2002). Second, most of the AM fungal types obtained in the field are not available in cultures for which sequence data are available. Third, the AM fungal database in GenBank is presently too small to interpret efficiently the inter- and intraspecific genetic variation of these fungi. Proper delimitation of species in the glomeromycetes has still to be established in the future. The new AM fungal types therefore represent either taxa new to science, or taxa with yet unpublished ITS/5.8S sequences. However, we are uncertain whether each of the AM fungal types represents a single morphospecies, or whether some morphospecies include more than one sequence type. Species designation will be possible only after sequencing of identified spores from the respective study sites. Our finding of only three AM fungal spore types using T. erecta as a trap plant suggests the need to use a broader diversity of trap plants, including the host plant, in future studies.

Primer specificity

The specific primers used in this study were reported to amplify the ITS region of rDNA of six clades of AM fungi, with the exception of the G. versiforme clade (Redecker, 2000). We have previously reported problems of specificity concerning the primers ACAU1660 and ARCH1311 (Wubet et al., 2003b). Similar problems of specificity were observed regarding LETC1670, which amplified a sequence type belonging to the Archaeosporaceae (Arch1, Fig. 1) in this study (Table 1). We also obtained AM fungal sequences (Glom1, Fig. 1) belonging to taxa of the G. versiforme clade of Redecker (2000) or the Diversisporaceae of Schwarzott et al. (2001), amplified with the primer ACAU1660 which was designed specifically to amplify members of Acaulosporaceae (Redecker, 2000). These primers are not up to the reported and expected level of specificity when used for field-collected roots.

AM fungal diversity

The AM fungal diversity colonizing roots of P. africana in the dry Afromontane forest ecosystem was found to be high (H = 2.58). Husband et al. (2002b) also reported high AM fungal diversity in the Barro Colorado Island tropical forest (H = 2.33 from two host species), as compared to a temperate grassland (H = 1.71 from two host species; Vandenkoornhuyse et al., 2002); a seminatural woodland (H = 1.44 from five host species; Helgason et al., 1999); or temperate arable fields (H = 1.16 from four host species; Daniel et al., 2001). Although our diversity values based on ITS sequence diversity might not be directly comparable with the diversity value of Husband et al. (2002b) based on SSU RFLP, a high AM fungal diversity is in accordance with the view of Husband et al. (2002a,b), who suggested that below-ground mycorrhizal diversity of tropical forests is high, corresponding to their long-established high plant diversity. However, because of limited success in obtaining amplified PCR products (data not shown), low specificity of the primers used, and low sampling intensity of cloned products, AM fungal richness may have been underestimated in this study. With all these limitations, the richness value of 20 AM fungal types obtained from 50 roots of 10 individuals of a plant species, which accounts for about 14% of the described AM fungal species, supports the view that the overall AM fungal diversity is extremely underestimated (Helgason et al., 2002; Husband et al., 2002a,b; Vandenkoornhuyse et al., 2002). Thus our results corroborate the outlook of other authors (Bever et al., 2001; Helgason et al., 2002; Husband et al., 2002a,b) who have argued that AM fungal diversity is comparable with that of plant communities and suggest that many more AM fungal species will be discovered in the future.

Distribution of AM fungal types

The AM fungal population colonizing P. africana at the two sites is significantly different. This variation between the two sites might be caused by soil factors known to influence AM fungal spore distribution (Cuenca & Meneses, 1996). Smith et al. (2000) and Helgason et al. (2002) also presented experimental evidence on physical and functional selectivity in AM symbiosis in field soils, where a diverse community of AM fungi plays particular roles in association with individual hosts. The results of Bidartondo et al. (2002) also indicate that some AM associations are highly specific, supporting the idea of functional diversity within species of AM fungi. Two of the three AM fungal spores isolated from baiting cultures of rhizosphere soil of P. africana were not amplified directly from colonized roots. This may support the view of Bidartondo et al. (2002) that plant specificity is not caused by a lack of other co-occurring AM fungi.

This first molecular study of AM fungal diversity in colonized roots of P. africana has produced a list of novel AM fungal types new to Ethiopia and to science. We have shown that the mycorrhizal community colonizing roots of P. africana is highly diverse. Isolation of the AM fungal types and screening for a combination of functionally complementary species that may be used in nursery propagation constitutes the future challenge.

Acknowledgements

This work was supported by a grant to T.W. from the Deutscher Akademischer Austauschdienst (DAAD). The authors wish to acknowledge Alemaya University, the Ethiopian Agricultural Research Organization and State Forest Offices for the supply of logistics at the study sites, and Ato Yayehyirad and Ato Tamrat for their assistance during sample collection.