Diversispora celata sp. nov: molecular ecology and phylotaxonomy of an inconspicuous arbuscular mycorrhizal fungus


  • Hannes A. Gamper,

    1. Netherlands Institute of Ecology (NIOO-KNAW) – Centre for Terrestrial Ecology, Department of Terrestrial Microbial Ecology, Boterhoeksestraat 48, PO Box 40, 6666 ZG Heteren, The Netherlands;
    Search for more papers by this author
  • Christopher Walker,

    1. Royal Botanic Garden Edinburgh, 20A Inverleith Row, Edinburgh EH3 5LR, UK;
    Search for more papers by this author
  • Arthur Schüßler

    1. Genetics, Department of Biology I, Ludwig-Maximilian-University Munich, Großhadernerstrasse 4, 82152 Planegg-Martinsried, Munich, Germany
    Search for more papers by this author

Author for correspondence:
Hannes Gamper
Tel: +31 26 4791205
Fax: +31 26 4723227
Email: h.gamper@nioo.knaw.nl


  • • The increasing numbers of taxonomically unassigned phylotypes reported in molecular ecological studies contrast with the few formally described arbuscular mycorrhizal fungi (AMF; Glomeromycota). Here, a species new to science with Glomus-like spores is phylogenetically, morphologically and ecologically characterized.
  • • From single spore isolates of a previously recognized member of the Diversisporaceae from Swiss agricultural grassland, 17 new nuclear internal transcribed spacer (ITS), large subunit (LSU) and small subunit (SSU) ribosomal RNA (rRNA) gene sequences were determined and compared with 14 newly generated sequences of two close relatives and public database sequences, including environmental sequences, of known geographic origin.
  • • SSU ribosomal DNA (rDNA) sequence signatures and phylogenies based on ITS, LSU and SSU rDNA sequences show that the fungus belongs to the genus Diversispora. It is described as Diversispora celata sp. nov. Comparison with environmental sequences in the public domain confirmed its molecular genetic distinctiveness and revealed a cross-continental distribution of close relatives.
  • • The value of combining morphology and phylogeny to characterize AMF was reinforced by the morphological similarity to other species and the inconspicuous nature of D. celata spores and mycorrhizas. Inclusion of all three nuclear rDNA regions in species descriptions will facilitate species determination from environmental phylotypes.


The recognition of species, whether named or unnamed, is fundamental to understanding how diversity affects ecosystem processes (Tilman et al., 2001; Zak et al., 2003). However, for arbuscular mycorrhizal fungi (AMF) of the phylum Glomeromycota (Schüßler et al., 2001b), the detection and discrimination of species are difficult and many are inadequately defined. Natural communities usually harbour AMF from several phylogenetic clades. Progress in ascribing physiological and ecological traits to particular AMF taxa or lineages (Hart & Reader, 2002; Munkvold et al., 2004; Maherali & Klironomos, 2007) largely depends on their molecular characterization and the availability of pure reference cultures. Once traits have been ascribed to particular species, or phylotypes, it may be possible to predict the influence of AMF on plant growth in the field and to understand ecosystem processes (van der Heijden & Scheublin, 2007). Molecular characterization, therefore, forms a foundation for ecological enquiries, which also require modern species descriptions to be integrated into current ecological knowledge.

Species and phylotypes from the Glomeromycota are often widely distributed geographically, probably as a result of their long biogeographical history (Pirozynski & Malloch, 1975; Redecker et al., 2000). However, the substantial environmental phylotypic richness (Öpik et al., 2006, 2008; Vandenkoornhuyse et al., 2007, Kottke et al., 2008) is much greater than the approximately 200 formally described AMF species (http://www.amf-phylogeny.com). Thus, the vast majority of existing species are not formally named. Moreover, the majority of described species have been neither sequenced nor established as cultured mycorrhizal symbionts, thus precluding further characterization.

The suitability of molecular genetic markers for comprehensive ecological community sampling can be compromised either by inadequate taxon resolution, for example as a result of DNA sequence conservation, or by failure to detect lineages, for example as a result of selectivity of PCR primers (Schüßler et al., 2001a; Walker et al., 2007). To address such problems, more reliable and comprehensive sequence databases are needed that are based on accurately identified organisms (Schüßler et al., 2003; Vilgalys, 2003; Bidartondo et al., 2008; Nilsson et al., 2008).

Characterizations and descriptions of AMF species are by their nature limited by the state of knowledge and the availability of reference material when they were made, and species identification solely on morphological grounds may be impossible, for example in the Diversisporales and Archaeosporales (Walker & Schüßler, 2004; Walker et al., 2007). Provision of sequence information in species descriptions is therefore essential for correct classification and may inform new species descriptions and taxonomic revisions. For molecular ecological investigations of phylotaxonomic richness and application of AMF in agriculture and horticulture, species recognition and identification must be simple and reliable. Methods used in field ecology should be applicable to all life stages, including mycelia in the soil and roots, and are often based on nuclear ribosomal DNA (rDNA) markers (Redecker et al., 2003).

The Diversisporales contain four AMF families, the Acaulosporaceae, Gigasporaceae, Entrophosporacea and Diversisporaceae. The first two are reasonably well circumscribed, whereas the Entrophosporaceae were emended to contain only two species (Sieverding & Oehl, 2006) without verification of their placing in the Diversisporales. The Diversisporaceae encompass species with very distinct spore morphologies, including those with ‘Glomus-like’ (glomoid) spores that can be identified only by a combination of molecular and morphological characters (Walker et al., 2004a,b; Walker & Schüßler, 2004).

Here we report the characterization of such a Glomus-like fungus, which was isolated from a fertile grassland in Switzerland where it is a common root symbiont (Gamper & Leuchtmann, 2007). Its molecular characterization had already shown it to belong in the Diversisporaceae, but our detailed morphological studies and new sequence data for the nuclear internal transcribed spacer (ITS) region and large subunit (LSU) rRNA gene show that the AMF is a species new to science. Phylogenetic comparisons with environmental rDNA sequences from public databases were made to study the geographic distribution of its known relatives. The isolates used in this study, including the type culture, are available to the research community in international culture collections.

Materials and Methods

Isolation and pure culturing

Trap cultures were established with soil from permanent monoculture plots of Trifolium repens cv. Milkanova in the Swiss long-term Free-Air CO2 Enrichment (FACE) experiment in Eschikon, 20 km north-east of Zurich (47°27′N, 8°41′E, 550 m asl). After 5 months, Plantago lanceolata L. seedlings were inoculated with single spores of the different morphotypes found, resulting in four pure isolates, two from plots subjected to elevated atmospheric pCO2 field conditions (60 Pa; isolates FACE 234 and FACE 272) and two from plots subjected to ambient atmospheric pCO2 field conditions (FACE 83 and FACE 410). FACE 83 was trapped with Allium porrum L. cv. Dubouchet and Glycine max (L.) Merrill as host plants, and the others with Leontodon autumnalis L. and Dipsacus fullonum L. as host plants. Single spore isolate FACE 234, specimens of which were chosen as the TYPUS of D. celata sp. nov., was registered and released to the scientific research community under the accession number BEG231 by depositing it in the International Bank for the Glomeromycota (BEG; http://www.kent.ac.uk/bio/beg/). The other isolates were also registered and are available under the following numbers: BEG230 (FACE 83), BEG232 (FACE 272) and BEG233 (FACE 410). All four isolates were also sent to the International Culture Collection of Arbuscular Mycorrhizal Fungi (INVAM; http://invam.caf.wvu.edu/) in the USA. See Supporting Information Methods S1 and S2 for further details about the site of origin, culturing conditions, culture histories, and other cultures analysed.

DNA extraction, PCR amplification, cloning and sequencing

Genomic DNA templates for the isolates listed above and for Glomus aurantium J. Blaszk., V. Blanke, C. Renker & F. Buscot and Glomus eburneum L.J. Kenn., J.C. Stutz & J.B. Morton (INVAM AZ420A) were prepared from single and multiple spores following the protocols of Schwarzott & Schüßler (2001), Jansa et al. (2003) and Gamper & Leuchtmann (2007). Fragments of more variable domains of the nuclear LSU rRNA and SSU rRNA genes were amplified after Gamper & Leuchtmann (2007) and Helgason et al. (1998), respectively. Near full-length SSU rRNA gene fragments were generated following Schwarzott & Schüßler (2001), and the nuclear ITS region was amplified as in Walker et al. (2007). All PCR amplicons were cloned for bidirectional plasmid sequencing on Prism®310, 3100, 3730xl ABI automated capillary sequencers (Applied Biosystems, Forsters City, CA, USA). Details are provided in Supporting Information Methods S3.

Phylogenetic sequence analyses

All three nuclear rDNA sequence partitions were phylogenetically analysed using the maximum likelihood (ML) algorithms of tree-puzzle (version 5.2; Schmidt et al., 2002) and phyml (version 2.4.4; Guindon & Gascuel, 2003), Bayesian inference (BI) in MrBayes (version 3.1.2; Ronquist & Huelsenbeck, 2003), and maximum parsimony (MP) and neighbour joining (NJ) tree searches in paup (version 4b10; Swofford, 2003). Separate best-fit sequence evolutionary models for the different data sets were determined in ModelTest (version 3.7; Posada & Crandall, 1998) by hierarchical likelihood ratio testing and branch support evaluated using bootstrap analyses.

To set the results in a broader ecological and biogeographical context, an additional multiple sequence alignment for the SSU rRNA gene, mainly with environmental sequences, and an alignment with ITS region sequences of members of the Diversisporaceae were created. The environmental sequences represent close matches to the new cultures in BLASTn searches in the public databases. The SSU rDNA data set covers reference sequences of the Diversisporales and all near full-length sequences of the Diversisporaceae. The LSU rDNA and ITS region data sets incorporate all Diversisporaceae sequences. All phylogenetic trees were rooted with Gigasporaceae sequences, except for the SSU rRNA gene tree of the Diversisporales for which members of Glomus groups A and B (Schwarzott et al., 2001) were chosen as an outgroup. The new ITS, LSU and SSU sequences were deposited in the public databases under the accession numbers listed in Supporting Information Table S1. The alignments (SSU rDNA, LSU rDNA and ITS region rDNA) are available at http://www.amf-phylogeny.com. Further details are given in Supporting Information Methods S4.

Morphological analyses

Specimens were examined morphologically using the techniques published in Walker et al. (1993), and the formal description followed the rules of the current Botanical Code (http://ibot.sav.sk/icbn/main.htm). Full details of the analyses are provided in Supporting Information Notes S1.


Four single spore isolates (FACE 83, 234, 272 and 410) of an AMF producing spores morphologically indistinguishable from each other were established. This AMF corresponds to a common root symbiont at the site of origin that had previously been detected with a taxon-specific LSU rDNA PCR amplification assay (Gamper & Leuchtmann, 2007). Molecular analysis of the SSU rRNA gene and ITS rDNA region showed the isolates to be indistinguishable from each other and different from any existing species characterized. Similarly, the morphological examination showed that no other described species shared its characteristics. The conclusion was drawn that the isolates represented a new species in the Diversisporaceae and a formal species description is presented to comply with the rules laid out in the Botanical Code. More comprehensive details are provided in Supporting Information Notes S1–S3.

Taxonomy of Diversispora celata sp. nov.

Diversispora celata C. Walker, Gamper and Schuessler sp. nov. (Fig. 1).

Figure 1.

Diversispora celata sp. nov.: morphological characteristics and comparison with Diversispora spurca. (a) Spores in water, showing overall shape and colour variation; spores singly (b) and in a loose cluster (c) in polyvinyl alcohol–lactic acid–glycerol (PVLG) (b, c) showing variation in size and shape; (d–g) spores in detail, showing variation in size and shape, and the persistent subtending hypha; (g) immediate reaction of spore when immersed in PVLG with Melzer's reagent – the yellow staining fades later; (h, i) detail showing some variation in the shape of the subtending hypha and occlusion by a slightly distal septum; (j) crushed spore with a funnel-shaped base similar to that found in Glomus mosseae; (k) plasmolysis in PVLG giving the erroneous impression of a thick ‘amorphous’ wall; (l) spore wall structure of a young spore, showing a thin, colourless outer component attached to the thicker, coloured main structural component; (m, n) comparison of wall structures of D. celata (m) and D. spurca (n) with three wall components arrowed (1, 2 and 3); (o, p) hyphal bridging or wound healing in D. celata (o) and D. spurca (p); (q, r) Plantago lanceolataD. celata mycorrhizas stained relatively weakly in acidified ink, showing indistinct arbuscules and hyphae (q) and an appressorium with an underlying hyphal coil (r).

Sporae juventute hyalinae, tum eburneae vel carneo-eburneae, maturitate ochraceae, globosae, subglobosae, ovoideae, obovoideae vel interdum irregulares, 5–160 × 48–192 µm. Hypha subtentens recta, recurvata vel infundibuliformis, septo occlusa. Parietis sporarum partibus tribus in turmis duabus. Pars 1 evanescens, hyalina, arcte adhaerens a partem 2 laminarem, luteum. Pars 3 flexibilibus, fortasse post obturamentum sporae efficerens.

Spores formed singly and in loose clusters in the soil; colourless when young, becoming ivory (2) to pinkish cream (4), and eventually ochre (9) with age; globose to subglobose to ellipsoid, ovoid or obovoid, occasionally irregular; 53–160 × 48–192 µm; with a straight to recurved, parallel-sided to funnel-shaped subtending hypha, occluded by a septum formed from the laminated structural wall component. Spore wall structure of three components in two groups. Component 1 evanescent, thin (< 1 µm), colourless, often with fine soil particles incorporated in a matrix of variable thickness, tightly adherent to laminated, yellow component 2 that is 2–5 µm thick. Wall component 3 flexible, much less than 1 µm thick, possibly developing after spore occlusion. Placed in the genus Diversispora (Diversisporaceae) by its SSU sequence signatures and distinguished from other members of the genus by its nuclear ITS region and LSU rDNA sequences.

Collections examined  All from pot cultures with P. lanceolata established with spores from soil-trap pot cultures. Switzerland, field station of the Swiss Federal Institute of Technology (ETH) Eschikon-Lindau ZH, near Zurich (47°27′N, 8°41′E). Isolate FACE 234: single spore isolate taken 20 Jul. 2005, H. Gamper (Att 1278-2, W4718 TYPUS E, Isotype MSB); loc. cit., 27 Jul. 2005, H. Gamper (Att 1278-2, W4719) (E); loc. cit. from a subculture of Att 1278-2, H. Gamper 2 & 3 Jun. 2008) (Att 1278-3, W5461, W5462); loc. cit. from a subculture of Att 1278-2, H. Gamper 2 & 15 Feb. 2008) (Att 1278-5, W5401).

Etymology  From the Latin, celator, a concealer, referring to the small size and pale colour of the spores, which make them difficult to see (concealed) in substrate extractions.

Distribution and habitat  Isolated from an agricultural grassland in Switzerland and cultivated in pot culture with P. lanceolata, Glycine max, Allium porrum, Leontodon autumnalis and Dipsacus fullonum. Forming arbuscular mycorrhizas; vesicles not observed.

See Supporting Information Notes S1 and S2 for details of the morphology of D. celata sp. nov. and comparisons with other members of the Glomeromycota.

Molecular phylogeny and distinctiveness of D. celata

Analyses of near full-length nuclear SSU rRNA and partial LSU rRNA gene sequences revealed that D. celata is phylogenetically affiliated with other members of the genus Diversispora (Figs 2, 3). In the SSU rRNA gene tree, relatively high sequence conservation precludes resolution at the species level. The partial LSU rDNA sequences appear to resolve the species level (Fig. 3), although this may only be because of the presently poorer taxonomic coverage for this rRNA gene partition. Together, the SSU and LSU rRNA gene trees showed D. celata to be closely related to Diversispora spurca, Glomus versiforme, G. aurantium and G. eburneum. The phylogenetic signal from the ITS rDNA region (ITS1, 5.8S rRNA gene, and ITS2) resolves D. celata as most closely related to a phylotype known from environmental studies in Africa (Namibia) and North America (Arizona, USA) (Fig. 4). Diversispora celata, G. eburneum and the environmental phylotypes from Namibia and Arizona fall within a sister clade to that formed by G. versiforme, G. aurantium, and an ‘uncultured G. versiforme’ from Europe (Turingia, Germany); the latter possibly representing another undescribed species. The recently reported ‘Glomus fulvum clade’ (Redecker et al., 2007), comprising Glomus fulvum, G. pulvinatum and Glomus megalocarpum, forms a more distantly related lineage within the Diversisporaceae (Fig. 4), probably at the level of genus as suggested by Redecker et al. (2007). New sequences of all three rDNA regions for G. eburneum showed it to be the most closely related named species to D. celata. We cannot make such detailed comparison with the generic type species D. spurca, as its phylogenetic placement can presently only be inferred from SSU rDNA sequences.

Figure 2.

Phylogram of cultured and/or voucher-based members of the Diversisporales derived from nuclear small subunit (SSU) rRNA gene sequences, indicating the family affiliation of Diversispora celata (grey box). The tree topology and molecular distances (nucleotide substitutions per site) were derived from maximum likelihood (ML) inferences based on 1692 aligned and 275 parsimony-informative sites, using TreePuzzle. Representatives of the Glomeraceae were chosen to root the tree. Support values from quartet puzzling, Bayesian, ML, maximum parsimony (MP) and neighbour-joining distance analyses are shown. Consistent branch support of > 80% among all five analysis methods is indicated by thick branches. Branch support values are only given if at least three analysis methods yielded ≥ 50%; MP tree length 1054, consistency index (CI) = 0.503. If available, culture, attempt, voucher numbers, and geographic origin are shown. Genus abbreviations: D, Diversispora; Gi, Gigaspora; G, Glomus; P, Pacispora; S, Scutellospora. Geographic abbreviations: AZ, Arizona; CA, California; ENG, England; FL, Florida; HE, Hessen; OR, Oregon; QC, Quebec; RJ, Rio de Janeiro; SCT, Scotland; SU, Suomi; SZ, Szczecin; TA, Tres Arrovos; WV, West Virginia. Public sequence database accession numbers are shown in parentheses; consensus A, X86687 + Y17651 + AJ132666 + AJ276088; consensus B, AJ301860-63 + AJ276076 + Y17644; consensus C, AJ276077 + AJ276078 + Y17649+Y17650; consensus D, AF038590 + U31997; consensus E, AJ871274 + AJ871275; consensus F, AB041344 + AB041345; consensus G, AJ871270–AJ871272; consensus H, AJ306443 + AJ306445 + AJ306446; consensus I, AJ276092 + AJ276093; consensus J, AJ619940–AJ619943; consensus K, AJ619944–AJ619947; consensus L, AJ619948–AJ619951; consensus M, AJ619952–AJ619955.

Figure 3.

Phylogram derived from nuclear large subunit (LSU) rRNA gene sequences, indicating the family affiliation of Diversispora celata (grey box) with respect to cultured and/or voucher-based members of the Diversisporales. The tree topology and molecular distances were derived from maximum likelihood (ML) inferences based on 652 aligned and 274 parsimony-informative sites, using TreePuzzle. Representatives of the Glomeraceae were chosen to root the tree. Maximum parsimony (MP) tree length 967, consistency index (CI) = 0.504. Annotations of support values, genus abbreviations and sequence accession numbers are as in Fig. 2. Geographic abbreviations: AZ, Arizona; IA, Iowa; ENG, England; FL, Florida; NH, New Hampshire; QC, Quebec; SCT, Scotland; SUM, Sumatra; UT, Utah; WB, West Bohemia.

Figure 4.

Phylogram derived from sequences of the nuclear ribosomal internal transcribed spacer (ITS) region, indicating the genus affiliation of Diversispora celata (grey box) and the distribution of its close relatives in the Diversisporaceae over the globe. Cultured and/or voucher-based members of the Diversisporaceae are shown in bold. The tree topology and molecular distances (nucleotide substitutions per site) were derived from maximum likelihood (ML) inferences based on 417 aligned and 206 parsimony-informative sites, using TreePuzzle. Representatives of the Gigasporaceae were chosen to root the tree (distance shortened by 50% for better display). Maximum parsimony (MP) tree length 581, consistency index (CI) = 0.410. Annotations of support values, genus abbreviations and sequence accession numbers are as in Fig. 2. Geographic abbreviations: AZ, Arizona; FL, Florida; TH, Thuringia. Consensus A: AF185690 + AF185691 + AF185695. Note: for environmental sequences and unidentified cultures the most likely genus affiliation is given and not the taxonomic affiliations mentioned in public databases.

Although not as reliable as phylogenetic analyses, pairwise distances are an alternative approach to the evaluation of the distinctiveness of species. The ranges of sequence identity, as calculated for intra- and interspecific comparisons of the NS31-AM1 SSU rDNA fragment, ITS, and LR1-FLR2 LSU rDNA region (Supporting Information Table S1), and evidence for shared rDNA sequence variants among closely related species within the genus Diversispora are reported as Supporting Information Notes S3.

Geographical distribution of Diversisporaceae phylotypes

Similar environmental ITS and SSU rDNA sequences from public databases with different geographical origins indicate a worldwide distribution of species in the Diversisporaceae. Figure 5 shows the phylogenetic relationship of D. celata to environmental SSU rDNA sequences from Australia, Botswana, the USA, Estonia, Ethiopia, Germany, Italy, Micronesia, the Netherlands, Panama, Portugal, South Korea, Sweden, and the UK. The sequences of D. celata cluster together with environmental sequences from Africa, America, Asia and Europe, and with those of the generic type species D. spurca, G. eburneum (AZ420A), Otospora bareai, and G. versiforme (BEG47) (Fig. 5). Members of the Diversisporaceae occur in both natural and disturbed habitats such as arable land and polluted sites. The ‘G. fulvum clade’ (Redecker et al., 2007) comprises environmental sequences from Portugal and South Korea, which demonstrates that its members are not restricted to the tropics.

Figure 5.

Phylogram derived from nuclear small subunit (SSU) rRNA gene sequences, indicating the phylogenetic relationship of Diversispora celata (grey box) with environmental sequences from public databases and their geographic origin. Cultured and/or voucher-based members of the Diversisporaceae are shown in bold. The tree topology and molecular distances (nucleotide substitutions per site) were derived from maximum likelihood inferences based on 1743 aligned and 134 parsimony-informative sites, using TreePuzzle. Representatives of the Gigasporaceae were chosen to root the tree. Maximum parsimony (MP) tree length 329, consistency index (CI) = 0.763. Annotations of support values, genus abbreviations and sequence accession numbers are as in Fig. 2. Geographic abbreviations: AZ, Arizona; CA, California; SCT, Scotland; TA, Tres Arrovos. Consensus A, AJ301860-63 + AJ276076 + Y17644; consensus B, X86687 + Y17651 + AJ132666 + AJ276088; consensus C, AJ276077 + AJ276078 + Y17649 + Y17650; consensus D, AF038590 + U31997; concatenated E: AM400229 + AM905318 (reported as Otospora bareai, but see Discussion). Note: for environmental sequences and unidentified cultures the most likely genus affiliation is given, if possible, and not the taxonomic affiliations mentioned in public databases.

In the ITS tree (Fig. 4), AMF from two INVAM cultures, AZ237 from the USA and NB101 from Namibia, are most closely related to the Swiss D. celata. The most closely related described species is from the USA (G. eburneum AZ420A). Glomus aurantium from maritime dunes in Israel, G. versiforme (BEG47) from potted plants in an Oregon conservatory, and many environmental sequences from Thuringia in Germany are closely related but split into several additional clusters.

In contrast to the SSU and ITS region rDNA sequences, environmental LSU rDNA sequences are underrepresented in the public databases and do not allow interpretation of the wider geographical distribution of Diversisporaceae. There is only one group of 11 short sequences (AF396808–AF396810 and AF396819–AF396826, apparently wrongly ascribed to Gigaspora) in the public databases. These were obtained by nested PCR from field-collected root material sampled within 17 km of the type locality for D. celata with purportedly Gigaspora-specific primers (Jansa et al., 2003). Two of those sequences (AF396819 and AF396820) group with G. versiforme BEG47, but the remaining sequences cannot reliably be affiliated to any taxon, perhaps because of their shortness and a lack of sufficiently closely related phylotypes. One further environmental short LSU rDNA sequence (AB206237, from Mount Fuji in Japan; Wu et al., 2007) possibly comes from another member of the Diversisporaceae (not shown).


This discovery, isolation, and comprehensive morphological and molecular characterization of D. celata, as a species new to science, originated from a molecular ecological initiative to detect AMF lineages directly in root samples from the field (Gamper & Leuchtmann, 2007) and to establish AMF single spore cultures. Four pure cultures producing spores morphologically indistinguishable from each other were established (see Supporting Information Notes S1 and S2 for detailed morphological comparisons). Close correspondence among partial LSU rRNA gene sequences from the distinct isolates (Gamper & Leuchtmann, 2007) was linked with morphology to preliminarily characterize the organism.

For a more detailed analysis, additional LSU rDNA sequences, along with SSU and ITS rDNA region sequences, were generated, including some from closely related fungi (Supporting Information Table S1). The phylogenetic analyses (Figs 2–5) resolved D. celata in a monophyletic clade that also included the generic type species D. spurca (Figs 2, 3). This ‘Diversispora clade’ is clearly distinct from a sister clade comprising G. fulvum, G. pulvinatum and G. megalocarpum (Fig. 4) with divergent SSU sequence signatures (Walker & Schüßler, 2004; Redecker et al., 2007).

Recently, our placement of D. celata within the genus Diversispora had to be reinterpreted when O. bareai was described (Palenzuela et al., 2008). Two relatively short nuclear SSU rDNA sequences ascribed to Otospora cluster within the ‘Diversispora clade’ (Fig. 5). This clade was interpreted here and in previous molecular studies as the genus level and therefore Otospora could render Diversispora paraphyletic. However, the sequences of Otospora completely match those signatures formally described in the protologue of Diversispora (Walker & Schüßler, 2004) as being specific for the genus, which challenges the erection of the new genus Otospora. Moreover, the spore morphology of O. bareai is acaulosporoid, unlike the glomoid spores of all other AMF within the Diversisporaceae. Such morphological diversity is not impossible, as both acaulosporoid and glomoid spores were demonstrated to form in the genus Ambispora (Walker et al., 2007). However, as a parsimonious explanation it is also possible that the short sequences reported as those of O. bareai could have been derived from xenobiotic DNA (Schüßler et al., 2003) of a Diversispora sp. present in the mixed trap pot culture. Further verification of the systematic position of Otospora is needed, as it cannot, in our opinion, yet be reliably placed in a phylogenetic framework.

rDNA-based species identification of D. celata

The SSU rDNA analyses placed the new AMF species in the genus Diversispora, defined by the clade containing D. spurca. From our analyses, the LSU and ITS region rDNA sequences probably provide resolution down to the species level. MP, NJ, tree puzzling, Bayesian, and ML analyses yielded similar SSU and LSU tree topologies and branch support values, except for some terminal branches and the usual higher posterior probability values from Bayesian analyses, which, for the highly variable ITS region, showed some deviations in tree branching. However, problems with Bayesian analyses of rather short, rapidly evolving sequences are known and make these topologies less reliable than those from ML inferences.

Presently, molecular characterization of taxa in Glomeromycota is best achieved using the rDNA region. However, as for Ambispora (Walker et al., 2007), the SSU rDNA does not resolve species in Diversispora. It is important to note that most previous molecular ecological studies were concerned with a taxonomic rank higher than that of species, because SSU rDNA was analysed. For species-level identification, rapidly evolving rDNA regions such as the ITS region will inform future molecular ecological studies, once the phylogenetic ‘backbone’ is sufficiently robust. Therefore, although pronounced intraspecific variation may impose some limitations, for species characterization it is desirable also to report the more variable ITS region and partial LSU rDNA sequences in future species descriptions. Such data would be essential for comparison of the findings of molecular ecological studies. In future, short, distinct nucleotide sequences, representing taxon-specific DNA signatures or ‘microcodes’ (Summerbell et al., 2005; Landis & Gargas, 2007), may help in the endeavour to ascribe species to environmental sequence types.

Living cultures for species descriptions and future studies

Diversispora celata was characterized by examining isolates obtained from independent soil-trap cultures, allowing analysis, to some degree, of intraspecific variation. Some previous AMF species descriptions have examined variation within single spore isolates (Walker & Vestberg, 1998; Declerck et al., 2000) and among more than one single spore isolate (Walker et al., 2007). Similar characterization of new species would be facilitated by the existence of living pure cultures which also would permit future investigations on morphological, developmental and physiological variation, and also allow predictions of responses to environmental influences (van der Heijden & Scheublin, 2007). The present species description is comprehensive and combines morphology and molecular phylogeny, but it is just the first step towards understanding the ecological traits of D. celata. It is desirable that the ‘whole fungus’ (Kendrick, 1979; Walker, 1985) be considered when describing AMF to achieve biologically and ecologically meaningful species characterizations. Such holistic species descriptions will also consider other aspects of the biology of AMF, such as anastomosis compatibility and symbiotic functional traits.

Molecular ecology and global distribution of Diversisporaceae

Molecular investigations are indispensable for the study of the ecology of AMF, with their cryptic lifestyles. The majority of AMF species will remain uncultured for the near future, and therefore molecular detection and identification are particularly important. Uncovering AMF molecular diversity in geographically distant locations will provide important information about global taxon distribution.

The SSU and ITS rDNA sequences of D. celata, when compared with environmental sequences from public databases, confirmed the molecular distinctiveness of the new species and revealed a global distribution of close relatives. For the Diversisporaceae, the geographical annotations of database entries suggest worldwide occurrence (Fig. 5) and evidence for undescribed species (Fig. 4). The two undescribed AMF in INVAM cultures AZ237 (USA) and NB101 (Namibia) were closest to D. celata and clusters of environmental sequences from Germany may represent further Diversispora species. In contrast to the apparently better taxonomic resolution of the ITS rDNA data set, poor geographical coverage impaired inferences about the global distribution of the Diversisporaceae, because most studies employing the ITS region as a marker were for relatively restricted areas of central Europe. For LSU rDNA the low number of publicly available sequences prevented analysis of the geographical distribution. The only environmental Diversisporaceae sequences originate from near the type locality of D. celata in Switzerland and from Japan.

Evidence from several molecular ecological studies may be severely biased by the PCR primer combinations used. This may be exemplified by the supposedly AMF-specific LSU rDNA primer pair FLR3/FLR4 (Gollotte et al., 2004), which shows poor binding site matching for members of the Diversisporaceae. In particular, the reverse primer FLR4 shows several mismatches at its 3′ end to all known Diversisporaceae sequences, which will therefore be overlooked in studies in which these primers are used. Results from studies using nonphylogenetically inclusive primers must therefore be interpreted carefully. For example, D. celata was commonly found in Trifolium repens root samples from Switzerland when species-specific PCR primers were used (Gamper & Leuchtmann, 2007) but would not have been detected if FLR4 had been used. Similar problems exist for other AMF lineages when using the SSU rDNA primer AM1 (Helgason et al., 1998).

Future species detection and descriptions in Glomeromycota

Our preliminary analysis of the geographical distribution of members of the Diversisporaceae underlines the importance of increasing the number of well-annotated reference sequences from identified AMF for future molecular ecological and biogeographical examination. Where doubt exists concerning species determination, database entries should indicate this. A change of policy by the public sequence databases (Bidartondo et al., 2008) would allow improvements in the taxonomic annotation of older sequence accessions.

To enable natural classification (Woese et al., 1990), taxonomy should ideally be informed by phylogenetic analyses, which would also help to avoid misclassifications in the wrong genera. The Botanical Code, however, only requires the designation of a dead type specimen along with a validly published description, although living cultures are encouraged in recommendation 8B.1, which reads ‘Whenever practicable a living culture should be prepared from the holotype material of the name of a newly described taxon of fungi or algae and deposited in at least two institutional culture or genetic resource collections’. This recommendation, although important for follow-up studies, is not always followed. Consequently many named species exist without cultures available to allow further study of their biology, genetics and ecology. If living cultures cannot be provided (many AMF appear to be recalcitrant), inclusion of comprehensive molecular evidence as part of species descriptions is essential for future revisions within a natural classification system, or for comparison with molecular ecological data.

This work continues the trend to integrate morphological and molecular evidence for AMF characterization and species descriptions, and proposes that, as far as possible, a ‘whole fungus’ approach should be encouraged by use of type isolates and their descendant cultures made generally available to the scientific community for future investigations. Particular care should be taken to ensure that public database entries are accurately named, where possible, and clearly labelled as ‘unidentified’ if not well characterized.


We are grateful to J. Blaszkowski, who generously provided spores of G. aurantium, and J. M. Trappe for the Latin diagnosis. HG received funding through research grants from the ETH, the BBSRC and the NIG during the time this study was carried out. AS received funding from the DFG, and work carried out in Germany was also funded by the DFG.