Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
• Glomus intraradices-like fungi are the most intensely studied arbuscular mycorrhizal (AM) fungi. However, there are several AM fungi named as G. intraradices that may not be conspecific. Therefore, the hypothesis was tested that DAOM197198 and similar AM fungi, such as BEG195, correspond to the type of G. intraradices.
• The G. intraradices isotype material, a descendant (INVAM FL208) of the type culture, and a morphologically corresponding AM fungus (MUCL49410) isolated from the type locality were studied and compared with several cultures of DAOM197198 and BEG195.
• Phylogenetic analyses of the partial small subunit (SSU), complete internal transcribed spacer (ITS) and partial large subunit (LSU) nuclear rDNA regions revealed two clades, one including G. intraradices FL208 and MUCL49410, the other containing DAOM197198 and BEG195.
• The two clades were clearly separated by sequence analyses, despite the high intraspecific and intrasporal ITS region sequence divergence of up to > 23%. We conclude that the AM fungi with the identifiers DAOM197198 and BEG195 are not G. intraradices, but fall in a clade that contains the recently described species G. irregulare.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
About 70–90% of land plant species form arbuscular mycorrhiza (Smith & Read, 2008), so it is obvious that the interaction of plants and the obligate symbiotic arbuscular mycorrhizal (AM) fungi of the Glomeromycota (Schüßler et al., 2001) is of major importance for the entire terrestrial ecosystem. In research on AM fungi (hereafter AMF), a fungus named Glomus intraradices is the most frequently used member of the Glomeromycota. To date, > 1200 publications refer to this species, > 130 of which have the name in the title. This wide use resulted from the first AMF established in in vitro root organ culture (ROC) being determined as G. intraradices (Chabot et al., 1992). The descendants of this ROC established in Canada, often referred to as DAOM197198 (or DAOM181602, another voucher number for the same fungus), are extensively used in basic research (e.g. for a genome sequencing project; Martin et al., 2008) and to demonstrate transient genetic transformation (Helber & Requena, 2008). It also is a component of some commercial plant growth-enhancing products (Corkidi et al., 2004; http://www.pro-mixbas.com).
However, a very basic question still remains to be resolved: does the DAOM197198 fungus indeed correspond to Glomus intraradices? It is possible that more than one species may have been identified with this name.
Despite the large body of published work, including recent publications dealing with genetic recombination and anastomoses compatibility between isolates of AMF that are closely related to DAOM197198 (Croll & Sanders, 2009; Croll et al., 2009), the definition of ‘G. intraradices’ is far from clear. The species was described by Schenck & Smith (1982) from a citrus orchard in Florida. The type specimens came from a pot culture established from root fragments of a Citrus sp., a descendant of which was donated to the INVAM collection (http://www.invam.caf.wvu.edu) where it was catalogued as FL208. Since then, many cultures and isolates have been determined to be G. intraradices. Analyses of their rDNA region showed that they belong within the Glomus Group Ab (GlGrAb) (Schwarzott et al., 2001; Jansa et al., 2002b; Börstler et al., 2008). Both DAOM197198 and BEG195 have been identified as G. intraradices, but there appears to be no published work comparing them with the original description, the type material, or FL208. Therefore, these cultures might represent one species, but they may instead also belong to a cohort of related species. This element of doubt stimulated a re-examination of the molecular evidence in relation to the species G. intraradices.
A fungal species is defined by its nomenclatural type, although as such it is a preserved sample and thus not available for study as a living entity. However, being its descendant, the culture G. intraradices FL208 can provide ‘living evidence’ of the true nature of the species. Subculture of FL208, herein termed ‘ex-type’ or ‘type-culture’, are available for such comparative study, so it is possible to investigate whether other AMF in the Glomus Group Ab (Schwarzott et al., 2001) may be other species rather than G. intraradices. Strictly speaking, the Botanical Code defines an ‘ex-type culture’ as being obtained from type material permanently preserved in a metabolically inactive state, but for convenience we extend this here to include cultures such as FL208, derived from the ‘type culture’ through a series of living subcultures.
We compared the G. intraradices isotype with ex-type specimens from pot cultures of FL208. Ex-type material was then compared by partial nuclear small subunit (SSU), internal transcribed spacer (ITS) and partial nuclear large subunit (LSU) rDNA region sequencing with three other cultures: a new isolate from the type locality corresponding morphologically to G. intraradices (now cultured as ROC in the GINCO collection, http://emma.agro.ucl.ac.be/ginco-bel, as MUCL49410), DAOM197198 (from Pont Rouge, Canada); and BEG195 (from Germany). The aim of this research was to determine if these organisms indeed all correspond to G. intraradices.
A secondary aim was to contribute to strategies that might be used for species determination based on DNA sequences. Such identification, termed ‘DNA barcoding’, must be accurate, rapid, cost-effective, culture-independent, universally accessible, and usable by nonexperts (Frézal & Leblois, 2008). For animal DNA barcoding, the mitochondrial cytochrome c oxidase I (COI) gene is widely used. In fungi, COI possesses length variation (0.64–12.3 kb; Seifert et al., 2007) too large to fulfil barcoding requirements. Because the molecular identification of fungi has been based mainly on the ITS rDNA region (Nilsson et al., 2008; http://unite.ut.ee) the ITS region will most likely become the primary barcode for fungi. The Barcoding of Life Database (BOLD, http://www.barcodinglife.org) already supports the storage and analysis of ITS sequences, and we therefore intended also to study, using GlGrAb sequences from public databases, whether the ITS region alone can be used to resolve species in the Glomeromycota.
Materials and Methods
Fungal type material
The isotype of G. intraradices, voucher OSC40255, was borrowed from Oregon State University herbarium (OSC). It consisted of spores and stained roots on dried-out microscope slides, and spores and blue-stained roots preserved in lactophenol. It was examined microscopically by accepted methods (Walker et al., 2007).
Ex-type cultures Pot culture substrate of FL208 was obtained on 23 March 2007 from INVAM (http://www.invam.caf.wvu.edu, where in the ‘accessions details page’ it is described as ‘subculture of the original isotype’). A subculture attempt (Att) number (Att4-36) was assigned to this material on receipt (Att4-0 is the identifier in our database for the original, root fragment, open pot culture of S. Nemec, established in autumn 1974). Replicate ex-type pot cultures were established in Germany (Att4-37, Att4-39–Att4-43), England (Att4-38) and Belgium (Att4-44). Moreover, a ROC (Declerck et al., 1998) was established in Belgium (Att4-45) on transformed carrot (Daucus carota) roots (from one root fragment of Att4-44). From this, one spore cluster on a single hypha was taken to establish Att4-46 (MUCL49413 in GINCO-BEL) on transformed chicory (Chichorium intybus) root (Fontaine et al., 2004). Material from a subsequent chicory ROC (Att4-64) was used for DNA extraction in Germany.
New isolate from the type locality A sample was provided by S. Nemec, collected in Florida, USA, from the type locality of G. intraradices (a citrus plantation between Clermont and Minneola close to Highway 27). From the description of the locality, an approximate latitude and longitude was estimated (28°33′41″N, 81°44′40″W) using Google Earth. A trap culture with roots and soil was established with Plantago lanceolata as host (Att1102-0, 14 October 2001) in disinfested 3 : 1 (v : v) horticultural sand-expanded attapulgite clay (Oil-Dri Corp., Chicago, IL, USA). Voucher samples determined as G. intraradices growing in the culture were made on 9 August 2002 (W4064), 17 March 2003 (W4344), and 3 March 2004 (W4598). From the 2004 sampling, a single spore was germinated on a filter fragment (Brundrett & Juniper, 1995) and successfully used to establish Att1102-7 with P. lanceolata. Spores formed in this culture corresponded broadly to the type material of G. intraradices. They were found abundantly singly in the substrate, in loose clusters of 2 to > 100, attached to fine, hyaline mycelium around roots. The fungus also sporulated heavily in the root cortex. On 6 June 2006, a single spore ROC was established (Att1102-9, MUCL49410) and subsequently subcultured as part of the GINCO collection.
Descendants of DAOM197198 Several cultures corresponding to DAOM197198 (originally collected in Pont Rouge, Canada) were examined. DAOM181602 is an earlier voucher number for a sample from an ancestral pot culture taken in 1981, before the fungus was transferred to the company Premier Tech Ltée. (Québec, Canada). The ROC widely used in AM research (Chabot et al., 1992) was initiated from a pot culture vouchered as DAOM197198 in 1987. Details about the culturing history of this fungus will be provided in a subsequent publication. As is common in many studies we use DAOM197198 as organism identifier, but stress that it is actually a voucher number of the herbarium in Ottawa and thus defines what was present in the culture at the time of sampling. We obtained ROC cultures of this fungus from several sources. (1) Att1192-44 originated from a ROC culture traced back to the laboratory of G. Becard (France), from where it was sent to the laboratory of I. Sanders (Switzerland), then to the laboratory of U. Paszkowski (Switzerland). In 2007 it was sent to the laboratory of M. Parniske (Germany) and from there to our laboratory. (2) Att1192-27 was obtained from the laboratory of P. Bonfante (Italy) in 2007, via the laboratory of P. Lammers (USA), where it was established from material produced by Premier Tech for the genome sequencing project (http://www.jgi.doe.gov/genome-projects/). (3) Att1192-53 was sent to us in 1996 by Y. Piché and spores were stored at −80°C. (4) Att690-23 was obtained via the University of Western Australia by C. Walker in November 2006 and established as a closed pot culture in Munich (March 2007). All this material purportedly stems from the same ROC established in the early 1990s (Chabot et al., 1992).
BEG195 This fungus originally was sampled from an agricultural field with winter cereals near Hannover (Germany). It was cultured at the University of Marburg, passed to the Sainsbury Laboratory (Norwich, UK) and then to the laboratory of M. Parniske (Munich, Germany), and thence, from Att1485-12, to our subculture (Att1485-13) used for DNA extraction.
Glomus proliferum (MUCL41827) This AMF was cultured from banana plantation in Guadeloupe. It was described by Declerck et al., (2000) and is available as ROC from GINCO.
Identifiers used in this publication To distinguish the different cultures and isolates studied here, we refer to them by using the most common descriptors. It should be borne in mind that these correspond either to vouchers (DAOM197198), organisms (INVAM FL208, MUCL49410, BEG195), or to individual subculture attempts (Att). DAOM197198 is used for the Canadian fungus from Pont Rouge (the fungus used in the AMF genome sequencing project), FL208 for the ex-type cultures from Florida, and MUCL49410 (Att1102-12 and descendants) for the new isolate from the type locality. BEG195 (as Att1485-13) was included in the analysis to represent a G. intraradices-like fungus from Europe.
DNA extraction, PCR amplification, cloning and sequencing
DNA extraction and PCR amplification Spores were cleaned and DNA extracted as described in Schwarzott & Schüßler (2001). In a first PCR, an amplicon containing a part of the SSU, the whole ITS1–5.8S–ITS2 region, and a part of the LSU rDNA were amplified with two different primer pairs (Table 1), using the Phusion High-Fidelity PCR Mastermix (Finnzymes, Espoo, Finland). As template 5 l of DNA extract (except for Att1192-27 and G. proliferum, where 2 µl were used) were used in 20-µl final reaction volumes. The final primer concentration was 0.5 µm of each primer. For primers SSUmAf (Krüger et al., 2009) and LR4+2 (ACCAGAGTTTCCTCTGGCT; modified LR4 primer, http://www.aftol.org) the PCR parameters were: 5 min initial denaturation at 99°C; 40 cycles of: 10 s at 99°C, 30 s at 59°C, 1 min at 72°C; final elongation 10 min at 72°C. The cycling parameters for the SSUmAf and LSUmAr (Krüger et al., 2009) PCR mix were identical, except for the annealing (15 s at 60°C) and elongation (45 s at 72°C) parameters.
Table 1. Cultures used in this study
Cloning no. (no. of clones sequenced)
Spore(s) for DNA extract
Information about culture numbers, vouchers, clones, PCR-primers and DNA extraction are shown. ss, Single spore; ms, multispore; rf, root fragment; sc, individual spore-cluster; n, nested PCR.
Att4-41 (rf, pot)/W5413
Att4-38 (rf, pot)/W5166
pHS089 (8), pHS086 (3); pHS080 (2)
SSUmCf/LSUmBr (n); SSU Glom1/NDL22 (n)
Att4-64 (sc, ROC)/W5507
Att1102-12 (ss, ROC)/W5070
Att 1192-44 (rf, ROC)/W5533
Att1192-27 (rf, ROC)/W5495
Att690-23 (rf, pot)/W5499
Att1485-12 (ss, pot)/W5272
Att894-7 (ss, pot)/W3776
Glomus cf. clarum
pHS116 (2), pHS117 (1)
When the first PCR did not result in visible bands after gel electrophoresis of 6 µl of PCR product, a second (nested) PCR was performed. The PCR reactions were initiated as for the first PCR with 0.1 µl, 0.2 µl, 0.5 µl or 1 µl from the first PCR used as template. Either the primer combination SSU-Glom1 (Renker et al., 2003) and NDL22 (van Tuinen et al., 1998) or the AMF specific primers SSUmCf and LSUmBr (Krüger et al., 2009) were used for the nested PCR. The cycling regime for SSUGlom1-NDL22 was: 5 min at 98°C; 30 cycles of: 10 s at 98°C, 30 s at 65°C and 1 min at 72°C; final elongation 10 min at 72°C. For SSUmCf-LSUmBr it was: 5 min at 99°C; 30 cycles of 10 s at 99°C, 15 s at 63°C and 45 s at 72°C; final elongation 10 min at 72°C. Nested PCR amplifications for Att1192-44 (pHS059) were performed with Taq DNA polymerase (Peqlab, Erlangen, Germany) and for Att4-38 (pHS080) with Top-Taq polymerase (Qiagen, Hilden, Germany), using the primer pair SSUGlom1-NDL22. In these cases the PCR program was 5 min at 94°C; 30 cycles of: 30 s at 94°C, 30 s at 58°C and 2 min at 72°C; final elongation of 10 min at 72°C.
Cloning and sequencing The PCR products were cloned with the TOPO TA or the Zero Blunt TOPO PCR Cloning Kit (Invitrogen) according to manufacturer's protocol, except that all components were used as 1/3 volume (except SOC medium for initial bacterial growth, which was used as full volume). The pHS113 clones (G. proliferum) were obtained using the StrataClone Blunt PCR Cloning Kit (Stratagene Agilent Technologies, La Jolla, CA, USA) according to the manual. Clones were analysed using colony PCR and products showing correct fragment size were used for RFLP with MboI, HinfI and RsaI. Selected clones were grown in liquid Terrific Broth media and plasmids isolated with the NucleoSpin Multi-8 Plasmid kit (Macherey & Nagel, Düren, Germany). Alternatively, a ‘quick and easy’ method modified after Ganguly et al. (2005) was used. Sanger sequencing was performed by the LMU Sequencing Service Unit on an ABI capillary sequencer using bigdye v3.1 sequencing chemistry. The new rDNA sequences derived from this study were deposited in the EMBL/GenBank/DDBJ databases with the accession numbers FM865536–FM865617 and FM992377-FM992402.
The 3′ partial SSU rDNA, the ITS region, and the 5′ partial LSU rDNA were either analysed for the sequences derived from this study together with some selected shorter sequences of characterized AMF from the database, or the ITS region alone was used for comparison with public database sequences. Sequences were assembled and proof-read with the program seqassem and aligned with align (both from http://www.sequentix.de). Sequence divergences in per cent (uncorrected pairwise distances) were calculated by using bioedit (Hall, 1999) and based on alignments containing either sequences of G. intraradices-related species or of Ambispora spp., including all positions. The alignment of the maximal common length (representing the SSUGlom1-NDL22 fragment) sequences comprised 1555 sites, 1387 of which could be unambiguously aligned and were used for the analyses.
When sequences derived from the same PCR reaction were identical after excluding ambiguous sites from the alignment, only one was included in phylogenetic analyses and accession numbers of the other sequences were later annotated to the corresponding clade. The analyses of database sequences included identical ones, because it is difficult to interpret whether they came from the same or different spores, PCR reactions or even cultures. Phylogenetic analysis was performed with phylip (Felsenstein, 1989), raxml (Stamatakis & Hoover, 2008), tree-puzzle (Schmidt et al., 2002) and mrbayes (Ronquist & Huelsenbeck, 2003). Consensus trees were constructed after 1000-fold bootstrapped neighbour-joining (NJ, based on Kimura-2-parameter model with phylip 3.8; Felsenstein, 1989) analyses. raxml was set to a maximum likelihood (ML) search for best-scoring tree after 1000 bootstraps and the proportion of invariable sites was estimated by the program. The ML quartet puzzling (MLQP) analyses were performed with tree-puzzle 5.2 (based on GTR model), estimating nucleotide frequencies and gamma distributed heterogeneous rates from the dataset. As an alternative approach, the sequences (SSUGlom1-NDL22 fragment) were aligned automatically using the MAFFT online server (http://align.bmr.kyushu-u.ac.jp/mafft/online/server/), for comparison with the results from the manual alignment. The iterative refinement option of MAFFT was set to FFT-NS-i (Katoh et al., 2002). Phylogenetic analysis was performed by raxml with settings as above.
For the ITS region analyses, public database sequences labelled as G. intraradices and such of closely related species were included. Analyses were based on a manually made alignment. Identical ITS region sequences were excluded and afterwards annotated to the appropriate clade. In total 395 sites could be unambiguously aligned. Phylogenetic analyses were performed with phylip (NJ) and raxml (ML). The two sequence alignments (SSU + ITS + LSU rDNA sequences from this study and ITS region including database sequences) are available from http://www.amf-phylogeny.com.
The isotype material of G. intraradices is in relatively poor condition. It appears heavily parasitized and the spore wall structure was difficult to determine. It was also much darkened in colour because of storage in lactophenol. However, it was possible to see characteristics used by the original authorities to describe the species, along with other details that were not published in the protologue, and to compare them with those of DAOM197198. We do not show the detailed morphological comparisons here, since a detailed re-description including designation of an epitype of G. intraradices is currently in preparation for publication elsewhere.
To characterize the ‘model AMF’ DAOM197198 at the molecular level, we studied several cultures, including some from single-spore isolates. The phylogenetic analyses comprising the partial SSU, entire ITS region, and the partial LSU rDNA sequences show a clear separation into two clades (Fig. 1). The first clade includes the ex-type culture of G. intraradices, FL208 (Att4-38, Att4-41, Att4-64) and the new isolate from the type locality, MUCL49410 (Att1102-12). The second clade contains DAOM197198 (Att690-23, Att1192-27, Att1192-44, Att1192-50) and BEG195 (Att1485-12). Shorter sequences of G. irregulare, a G.intraradices-like species (Błaszkowski et al., 2008), and isolates from Switzerland were included in the analyses and also cluster in the latter clade. When using a fully automated MAFFT alignment as a base for the phylogenetic analyses, the same, distinct clades were resolved. Later, we named the clade containing FL208 and MUCL49410 as the ‘G. intraradices clade’ and the clade containing DAOM197198, BEG195, G. irregulare and the Swiss isolates as the ‘G. irregulare clade’. A separation of these clades also existed when using only the ITS1 region, the partial LSU sequences, or the combined partial SSU + 5.8S + partial LSU (without ITS1 + ITS2) fragment (raxml bootstrap support > 70%, for all three options; data not shown). However, analyses of only the ITS2, or the ITS1 + 5.8S + ITS2 region resulted in a monophyletic grouping for the G. irregulare clade sequences, but the G. intraradices clade appeared paraphyletic. This indicated that the ITS2 region alone carries conflicting phylogenetic signal or too much noise, hindering resolution. Further ITS region analyses including database sequences (see below) also show that this region is not suitable to resolve species in GlGrAb.
The maximal pairwise uncorrected distance values (p-distances) within the G. irregulare clade were 9.3% for the newly obtained SSU + ITS + LSU rDNA sequences. The divergence in the G. intraradices clade was up to 14.1%. An overlap of highest intraspecific p-distances in the G. intraradices clade with the lowest interspecific (relating to the G. irregulare clade) p-distance values was observed for the full-length sequences and ITS region, indicating the lack of a so called ‘barcode gap’ (http://www.barcoding.si.edu/). The variation in the partial LSU sequences only was 7.8% for the G. irregulare clade and 11.8% for the G. intraradices clade.
The most variable ITS region showed p-distances of up to 16.3% for the G. irregulare clade and up to 23.1% for the G. intraradices clade. This enormous ITS region variability was found within one FL208 ex-type culture spore from Att4-41 (Fig. 1) and is the highest ever recorded for an individual AMF spore. Within this spore the maximal p-distance for the SSU + ITS + LSU rDNA sequence is 13.9%.
Analysing the ITS region including database sequences revealed several clades. Most ITS-region sequences designated as G. intraradices fell within the G. irregulare clade, including those of BEG158, BEG195 and the known ITS sequences of the genotypes (II, VI, XII, XV, XVII, XVIII; Croll et al., 2008; Croll & Sanders, 2009) of isolates from a field site in Switzerland (Jansa et al., 2002a; Koch et al., 2004). The FL208 sequences from the public database (Börstler et al., 2008, Sudarshana P. et al., unpublished), as well as KS906 (AF185669-73) cluster together with the FL208 sequences from our analysis. Three new sequences (FM865546 from pHS051-20, FM865599 from pHS099-16 and FM865604 from pHS099-41) that stem from ROC of either a G. intraradices FL208 descendant or the new isolate (MUCL49410) cluster distantly from the main clade (Fig. 2). Sequences from the recently described G. irregulare, which was studied together with a fungus identified as G. intraradices isolated from the same trap culture (Błaszkowski et al., 2008), clearly fall into subclades containing DAOM197198 + BEG195 sequences (Figs 1, 2).
The analyses of database sequences solely from the ITS region indicated that some that were annotated as G. intraradices are resolved to belong to separated, highly supported clades. One of these clades is represented by EY118 + INVAM GR104 (AF185684, AF185686 and AF185651) and another one by INVAM VA110 + INVAM CA502 (AM980854–59) (see also Börstler et al., 2008). However, further VA110 sequences (AF185674-6) group within the G. intraradices FL208 + MUCL49410 clade. Another closely related clade is represented by sequences of Glomus cf. clarum Att894-7 and the G. clarum (CL883A) database sequences AJ243275 and AJ239123. A very distinct cluster of sequences (AJ517450-61; EnvGrA in Fig. 2) annotated as G. intraradices (Renker et al., 2005) is clearly separated from both, the G. intraradices and G. irregulare clades.
The species G. intraradices is defined from its type material and the accompanying protologue. Unfortunately, the latter, while perhaps being adequate at the time it was written, does not describe characteristics in the detail required by recent species descriptions for glomeromycotan organisms. Indeed, defining morphospecies within the clade in which G. intraradices is placed by molecular analysis (GlGrAb, Schwarzott et al., 2001) seems to be difficult. Glomus intraradices produces spores both in the roots and in the substrate. The presence of intraradical spores itself is not a species-specific character but a symplesiomorphy shared with other AMF species. These intraradical spores have considerable variation in size, shape, subtending hyphal characteristics and reaction to Melzer's reagent, and generally a similar wall structure. The extraradical spores are predominantly globose to subglobose. A detailed description of the morphological characteristics of the species is currently in preparation for publication elsewhere, based on a re-examination of the type and study of both ex-type material and a new isolate obtained in pot culture and ROC from the type locality.
Organisms described as G. intraradices encompass more than one species
Our results revealed that many cultures or isolates frequently used in AM research and named G. intraradices very likely do not correspond to that species. In particular, the model fungus DAOM197198 cannot be phylogenetically resolved as G. intraradices (Fig. 1). This is in agreement with recently published studies dealing with mtLSU and nuc5.8S-ITS2 sequence data (Börstler et al., 2008) and analyses of nucITS2 sequences (Jansa et al., 2002b), although these studies did not focus on the species concept of G. intraradices. Based on our analyses, the name G. intraradices should only be applied for the INVAM FL208 descendants (ex-type), MUCL49410 (new isolate from the type locality) and other AMF that share the same phylogenetic and morphological characteristics. Because the recently described species G. irregulare clearly clusters with DAOM197198, although this was not shown in the original publication (Błaszkowski et al., 2008), we use the label ‘G. irregulare clade’ for this group also containing sequences from BEG195 and well investigated isolates from a field site in Switzerland. The published G. irregulare sequences form a weakly supported cluster within that clade (Fig. 1). This is most likely caused by sequencing only one main ITS rDNA variant, because the ITS variability in the published sequences of G. irregulare is only c. 1% and exceptionally low for GlGrAb. This raises questions of how to define species in such a complex clade. The answer to this question will require extensive further detailed analyses.
It can, however, be deduced that the G. irregulare clade (DAOM197198 + BEG195 + G. irregulare + Swiss isolates) and the G. intraradices clade (FL208 + MUCL49410) represent distinct AMF species, because:
• despite the large intraspecific sequence variability, DAOM197198 sequences from AMF cultures that were widely separated in space and time all cluster together in the G.irregulare clade, clearly separated from the G. intraradices clade showing even larger intraspecific (and intrasporal) variability;
• BEG195 (from Europe) sequences are embedded within the DAOM197198 (from North America) sequences, and thus within the G.irregulare clade;
• no isolate gave rise to any sequence (including those from the database) that cluster with members of the other clade, indicating that no rRNA gene flow takes place between the two clades. If the G. irregulare and the G. intraradices clades would be conspecific, we would expect at least some sequences crossing the borders of the phylogenetic clades. This expectation is supported by recently published evidence for genetic recombination between G. irregulare clade members (Croll & Sanders, 2009);
• the G. intraradices and the G. irregulare clades separate clearly from each other and from those representing closely related species (G. proliferum, G. clarum).
Conclusively, the new isolate (MUCL49410) from the type-locality corresponds phylogenetically to the G. intraradices FL208 ex-type culture. This is an indication that FL208 had not been contaminated over the years and should be accepted as indeed corresponding with the type. From these results, the earlier morphological identification of DAOM197198 and BEG195 as G. intraradices, while perhaps satisfying earlier morphological criteria, is shown to be incorrect by our molecular evidence.
Morphological, phylogenetic and biological species concepts
A morphological species concept is historically used for AMF, but a biological concept is preferable if it can be defined. A phylogenetic concept based on nonparalogous molecular markers may be congruent with a biological concept. It might be characterized by using just a single marker but only if such a marker coincides with the species boundaries.
With these concepts in mind, we have investigated some closely related AMF morphospecies in the past, to find out if they can be distinguished by their rDNA sequences. For example, morphological identification was difficult in the family Ambisporaceae in which some species form both acaulosporoid and glomoid spores. Molecular analyses proved that they were well separated from either the Acaulosporaceae or the Glomeraceae and ITS region analyses allowed separation of what were interpreted as different species, improving the morphological concept (Sawaki et al., 1998; Redecker et al., 2000; Walker et al., 2007). Nevertheless, the resolution of the ITS region seems to be limited, and in GlGrAb the situation is complicated. On one hand, isolates in this group were named as G. intraradices (e.g. DAOM197198, BEG195 and isolates from Switzerland) but are phylogenetically clearly separated from the clade that actually includes G. intraradices (FL208). On the other hand, G. irregulare seems morphologically different from both DAOM197198 and BEG195 (Błaszkowski et al., 2008), although from the molecular evidence presented here, they might be interpreted as being conspecific.
How can we explain such a situation? One simple reason could be morphological plasticity making it difficult to distinguish species microscopically in this group of AMF. Glomus irregulare is described from supposedly consistent morphological characteristics, which are not shared by DAOM197198. However, the DAOM197198 descendent Att690-23 is also morphologically different from other DAOM197198 cultures, but supposedly shares the same ancestry. There is, moreover, some preliminary evidence that plasticity may relate to host plant species or culture conditions (Walker, 2008). It is thus possible that morphological differences in this group do not consistently correlate with phylogeny, but we cannot yet draw final conclusions about these aspects.
Possible plasticity may be correlated with the theory of conspecificity of different genotypes of Swiss AMF from one field site belonging to the G. irregulare clade. Some of these showed different growth characteristic phenotypes (Koch et al., 2004) and analyses indicated recombination events, at least for the studied genotypes II (isolate B3) and VI (isolate D2), between isolates (Croll & Sanders, 2009). Anastomosis compatibility experiments using five isolates from that field site indicated that isolates with different genotypes (A4 = XVIII, B3 = II, C2 = XV, C3 = XVII, D1 = VIII) can anastomose and that some progenies of C2 and C3 were genetically recombinant (Croll et al., 2009). It is possible that G. irregulare and DAOM197198-like fungi may be in one anastomosis compatibility group. To answer such questions, AMF must be cultured as isolates.
Regarding a phylogenetic concept for GlGrAb, species cannot reliably be separated by analyses of the ITS region (Fig. 2). For G. intraradices, a phylogenetic signal in the ITS2 region, which was not found in earlier works analysing either 5.8S + ITS2 or ITS2 only (Jansa et al., 2002b; Börstler et al., 2008), hindered phylogenetic resolution. However, when using the full-length (including the 3′ SSU and 5′ LSU rDNA region) fragments, the G. intraradices and G. irregulare clades were clearly separated. Our species concept is also in line with mitochondrial marker analyses (Börstler et al., 2008). We cannot yet conclude whether sequences clustering as more ancestral in a subclade may represent pseudogenes, but if this were to be the case they evolved after speciation because the full-length fragment carries the phylogenetic signal separating the clades. However, it is evident that concerted evolution of the rDNA repeats is extremely relaxed in members of GlGrAb.
Sequence variability and DNA barcoding
For the SSU + ITS + LSU rDNA fragment sequences the phylogenetic analyses resulted in a well-supported tree topology separating G. intraradices and the G.irregulare clade. We obtained the same results with a manual alignment and after fully automated alignment with MAFFT (Fig. 1), showing that a relatively simple, automated phylogenetic approach could resolve these two AMF subclades, interpreted as containing different species. We cannot yet conclude whether the fungi in the G. irregulare clade indeed are all conspecific, but from the molecular evidence this may well be the case.
The current discussion about fungal DNA barcoding (species identification) focuses on the ITS region, because of its historical use for identification of fungi (Nilsson et al., 2008). An important question is: ‘how can AMF species be distinguished and identified by potential DNA barcoding methods?’ This question is directly related to a species concept and the enormous intraspecific rDNA variability in AMF. In their recent publication about suitability of different rDNA regions for fungal DNA barcoding Nilsson et al. (2008) calculated an average glomeromycotan ITS variability of 7.5%. However, most data on Glomeromycota published therein will require thorough reinterpretation because of inaccurate species definition. For example, from 36 ITS sequences used to calculate 7.6% intraspecific ITS variability for G. versiforme, three are from a well-defined culture (BEG47), three are most likely from the same organism (although without identifiers in the public database) and the remaining 30 (> 83%) sequences analysed stem from environmental roots or spores without any reliable species affiliation. There is even evidence that the different G. versiforme sequences encompass distinct species (Gamper et al., 2009). Another dataset is composed of 12 sequences from six different Paraglomus occultum cultures, but because the species concept among Paraglomus is not yet well defined it is unclear whether these cultures indeed are conspecific. For the members of the G. irregulare clade, Nilsson et al. (2008) report 8.7%, intraspecific ITS variability and Jansa et al. (2002b) up to 18% for G. sp. BEG158 (Fig. 2) intrasporal ITS variability. For one G. intraradices FL208 spore we could show > 23% variability. Some further values reported for AMF are 9% for Gigaspora margarita (Lanfranco et al., 1999), 6% for Glomus mosseae (Lloyd-Macgilp et al., 1996), and 13.4% (when calculated based on the alignment of Walker et al., 2007) to 9.8% (median of absolute uncorrected (Hamming) distances calculated after automated pairwise alignments; Nilsson et al., 2008) for Ambispora leptoticha.
Although these values cannot yet be conclusively compared because of the different of clones or variants, general sampling densities, alignments and calculation methods, it appears that G. intraradices-related AMF in the clade GlGrAb show considerable sequence variability within the Glomeromycota. The LSU variability reported here is also higher than in most other studies, especially in the G. intraradices FL208 type culture (11.8%). The uncorrected p-distances for the diversisporacean Glomus aurantium, Glomus eburneum and Diversispora celata are 6.6, 1.4 and 2.5%, respectively, in the partial LSU region sequences analysed (Gamper et al., 2009). Generally, our results show that it will be necessary to include the 5′ LSU region in addition to the ITS region for DNA barcoding of AMF.
How many sequence variants in one species?
Assuming that Saccharomyces, Aspergillus and Neurospora spp. have 45–200 rDNA repeats (Kobayashi, 1998; Simon & Weiss, 2008) and the AMF Scutellospora castanea 75 (Hosny et al., 1999), fewer than 200 rDNA repeats would be expected within a nucleus of the AMF investigated here. It is still debated whether AMF are homokaryotic (Pawlowska & Taylor, 2004) or heterokaryotic (Hijri & Sanders, 2005), but the data presented here cannot resolve this question (see later). There are at least 149 ITS sequence variants from the different cultures and isolates in the G. irregulare clade (Fig. 2). For DAOM197198, we found 23 different ITS variants in the 30 sequences published here. From the public database, five additional sequences were identified for the complete ITS region, and further four covering ITS1 only. This makes 32 variants for DAOM197198, but the total number will be higher. This variability is derived from different cultures, which means that it potentially includes variants derived from recombination in different culturing lineages and may be higher than the number present in one spore. The variability found is not too high to be encoded within one nucleus. An interesting question is whether the very high variability in the GlGrAb rDNA is also reflected in other parts of the genome, which might be a problem when using members of this AMF lineage as genetic model systems.
Other subclades comprising G. intraradices-like AMF species
There are several other species known in the GlGrAb clade, and it was recently indicated that the only published G. proliferum sequence might cluster as a sister lineage to G. intraradices FL208 (Börstler et al., 2008). However, the longer sequences of G. proliferum published here form a clade separated from the G. intraradices and G. irregulare clades, although not with high bootstrap support. Based on the analyses of the ITS region alone further, well-separated clades could be considered as likely to represent distinct species. The cluster EnvGrA, named as ‘G. intraradices Type B’ (Renker et al., 2005), represents a distinct species, being more distant from G. intraradices than from species such as Glomus diaphanum and Glomus sinuosum.
For some of the AMF investigated, phylogenetic relationships cannot be satisfactory interpreted, such as for INVAM VA110. Some VA110 sequenced cluster with FL208, as already indicated by Börstler et al. (2008). These were submitted to the database in 1999 as part of a G. intraradices dataset that obviously included many contaminant sequences, as shown by their phylogenetic placement in different AMF orders. On the INVAM website it is noted that VA110 was derived from a mixed culture containing several species. VA110 is listed there as Glomus sp. on the ‘Accessions Culture Information’ pages and it is likely that sequences appearing in distinct phylogenetic clades actually represent different organisms. In general, sequence data from mixed cultures should be interpreted with caution.
In this study, we showed that G. intraradices (FL208 ex-type culture and MUCL49410 isolated from the type locality) clearly separates from the AMF in the G.irregulare-clade (DAOM197198 + BEG195 + Swiss isolates). Further, even more distant clades (e.g. EnvGrA) that were annotated as G. intraradices represent different, possibly undescribed AMF species. The model fungus used in AM research, DAOM197198, does not represent G. intraradices and is closely related to or perhaps even conspecific with G. irregulare.
We thank R. Halse, the curator of OSC for the loan of the type material, S. Nemec for providing the sample from the type locality, D. Redecker for discussions during the TRACEAM meeting held in January 2008, and J. Morton at INVAM for the provision of FL208. We thank M. and C. Krüger for critical reading and providing some unpublished data, and the GINCO for subculturing the FL208 and Att1102-9 descendants. Thanks to all who supplied samples. Work in Germany and the grant for H.S. were funded by the Marie Curie Early Stage Research Training Fellowship of the European Community's Sixth framework Programme (MEST-CT-2005-021016) and A.S. was funded by the German Research Foundation (DFG).