Mycologia, 90(5), 1998, pp. 854-869. © 1998 by The New York Botanical Garden, Bronx, NY 10458-5126 The genus Oidiodendron: species delimitation and phylogenetic relationships based on nuclear ribosomal DNA analysis Sarah Hambleton that this group has diverged significantly from the Department of Biological Sciences, University of two other genera in the family. Relationships in the Alberta, Edmonton, Alberta, Canada T6G 2E9 monophyletic group were correlated with anamorph state produced; three spp., and B. stria- Keith N. Egger Myxotrichum tosporus, all with distinct or reported Oidiodendron Natural Resourcesand Environmental Studies, states, nested with while University of Northern British Columbia, Prince anamorphic species, species George,British Columbia, Canada V2N 4Z9 of Myxotrichum with other anamorph states were ex- cluded. In addition, sequence divergence measures Randolph S. Currahl between the meiotic and mitotic species clustered in Department of Biological Sciences, University of two monophyletic clades were found to be compa- Canada T6G 2E9 Alberta, Edmonton, Alberta, rable to the intraspecific levels for Oidiodendronspp. These results support the initial hypothesis that Oi- diodendron is closely related phylogenetically to the Abstract: Nuclear ribosomal DNA sequences (ITS genus Myxotrichumand suggest that the generic con- region) of fifteen species in the hyphomycete genus cept of Byssoascus needs reexamination. Oidiodendron and ten species from the 4 genera in Key Words: ascomycetes, ericoid mycorrhiza, mo- the Myxotrichaceae, Byssoascus, Gymnostellatospora, lecular systematics, Myxotrichaceae, Onygenales Myxotrichum, and Pseudogymnoascus, were analysed to: (i) reveal the levels of intra- versus interspecific sequence variation within the genus Oidiodendron, INTRODUCTION clarify species delimitation and examine the useful- ness of some morphological characters used for iden- Species of Oidiodendron Robak (1932) (Hyphomy- tification; (ii) assess the possible conspecificity of doc- cetes) are commonly recovered from humus or de- umented ericoid mycorrhizal strains of Oidiodendron; caying wood and bark. The genus is readily identified and (iii) test the hypothesis based on morphological by the distinctive arborescent conidiogenous appa- inference that the genus Oidiodendron belongs with ratus. Erect, dematiaceous conidiophores terminate the genus Myxotrichumin the Myxotrichaceae (Ony- in a complex branching head of fertile hyphae that genales). Comparison of molecular and morpholog- segment basipetally into arthrospores. The mono- ical data for multiple strains of 0. griseum, 0. tenuis- graph of Barron (1962) provides a thorough discus- simum and 0. maius revealed that conidiophore sion of conidiogenesis with descriptions of nine spe- length and the production of a diffusing pigment are cies. The most comprehensive key to the genus in- not reliable key characters for the genus. Several his- cludes 13 species (Domsch et al., 1980) and a further torically important ericoid mycorrhizal strains, doc- six have been described in the Index of Fungi, Inter- umented as 0. griseum, were reidentified as 0. maius. national Mycological Institute, Vol. 1-6, 1940-1996. Parsimony analyses of 23 Oidiodendronstrains showed The genus is putatively placed within the Myxotri- that three highly supported monophyletic groups, chaceae (Onygenales) since the only teleomorphic each one consisting of a pair of species, are resolved species known to have an Oidiodendron state (Myxo- within the genus. A low level of sequence divergence trichum arcticum, M. cancellatum, M. setosum, and between the species in these pairs suggests conspe- Byssoascus striatosporus)belong to this family (Hawks- cificity for each pair. Other interspecific relationships worth et al., 1995, p. 320). were not well-supported by bootstrap values. Parsi- Though some Oidiodendronspecies have additional mony analysis of a second dataset composed of mi- distinctive features, the primary characters used for totic and meiotic taxa showed that Oidiodendron, My- identification are conidium size, shape and ornamen- xotrichum and Byssoascus form a well-supported tation as well as conidiophore length and cultural monophyletic group within the Myxotrichaceae, and morphology. All these vary even on one culture me- dium and the ranges can overlap among species (Bar- Accepted for publication May 21, 1998. ron, 1962). Such variability in morphology leads to 1 Corresponding author, email: [email protected] confusion about species delimitation and uncertainty 854 HAMBLETON ET AL.: PHYLOGENETIC RELATIONSHIPS OF OlDIODENDRON 855 as to which characteristics should be emphasized for simum and 0. maius) using sequence analysis of the identification purposes. internal transcribed spacer regions of the nuclear Three culturally similar species with intergrading rRNA gene (ITS-1 and ITS-2) as a basis for species features are Oidiodendron maius, 0. griseum, and 0. delimitation. The data were then used to assess the tenuissimum, the first two of which have been docu- stability of some key morphological characters for mented as ericoid mycorrhizal fungi and isolated identification purposes, notably conidiophore length from field-collected roots. In our recent study of the and pigment production. Secondly, morphological root endophytes of the Ericaceae in Alberta, Canada, assessment and RFLP analysis were used to examine a large number of Oidiodendron isolates was recov- whether five historically important ericoid mycorrhi- ered (Hambleton and Currah, 1997). Most were zal Oidiodendronstrains are conspecific. The third ob- identified as 0. maius, possessing the very long co- jective was to test the hypothesis that species of Oi- nidiophores and undulating branches of fertile hy- diodendronrepresent taxa within the Myxotrichaceae. phae that are distinctive for the species. Some strains were intermediate between 0. maius and 0. griseum MATERIALSAND METHODS based on conidiophore characteristics, though the conidia more closely matched those of 0. maius. Molecular.-Fungal strains chosen for this study (TA- Restriction fragment length polymorphism (RFLP) BLE I) included 15 species of Oidiodendron (14 de- analysis of PCR-amplified ITS regions was used to scribed species and one new species revealed by the identify strains of 0. maius that were morphologically analysis), seven species of Myxotrichum, and a single atypical (Hambleton and Currah, 1997). The ubiq- strain for each of Byssoascus striatosporus, Gymnostel- uitous presence of 0. maius in the roots of Ericaceae latospora japonica and Pseudogymnoascus roseus. All of boreal habitats, and the difficulties with identifi- four genera in the Myxotrichaceae were represented. cations based on morphology, led to speculation that The only well-recognized species in the family not historical records identifying 0. griseum as an ericoid included were M. ochraceumBerkeley & Broome, G. mycorrhizal fungus may have been confounded by frigida Uchiyama, Kamiya & Udagawa, and three spe- ambiguous features. cies for which there are no cultures available, M. Pigment production on agar media is another aeruginosum Montagne, M. bicolor(Ehrenberg) Fries, character that may not be as informative as once and M. berkeleyiApinis. Ex-type and authentic strains thought. The production of a dark brown diffusing were used where possible as species reference stan- pigment cited in the original description of 0. fus- dards for both the molecular and morphological cum (Robak, 1932, p. 251) (=0. tenuissimum; analyses. Hughes, 1958), which was used as a key character for Fungi were grown on thin plates of E-strain agar an intraspecific cultural grouping in the key of (Egger and Fortin, 1990) for 3 wk in the dark at Domsch et al. (1980), was also reported for 0. gri- room temperature. Eight plugs of mycelium, 0.5 cm seum by Tokumasu (1973) and Hambleton and Cur- diam with a minimum of agar attached, were used rah (1997). for DNA extraction following the chloroform/isopro- For one of the ericoid endophytic isolates of Ham- panol protocol of Gardes and Bruns (1993) with mi- bleton and Currah, and for a strain examined by Tok- nor modifications (Hambleton and Currah, 1997). umasu as well, the identification was more problem- Primers ITS1 and ITS4 (White et al., 1990) were atic because the upper range of the conidiophore used to amplify a portion of the nuclear ribosomal lengths was greater than described in the original de- RNA genes (rDNA) including both ITS spacer scription of 0. griseum (Melin and Nannfeldt, 1934, regions and the 5.8S subunit. PCR reaction volumes p. 440). Barron (1962) had already revised the co- of 100 ,iL contained PCR buffer, 25 mM MgCl2, 50 nidiophore length range for 0. griseum upwards with !1M each of dGTP, dATP, dTTP, and dCTP (Boehrin- the result that it overlapped with the ranges of both ger Mannheim Biochemica), 0.4 pJMof each primer 0. tenuissimum and 0. maius. Unfortunately, in this and 2 units of Taq DNA polymerase. One FLLof di- case, RFLP analysis was unable to differentiate 0. gri- luted DNA was used as the template for each reaction seum from 0. tenuissimum (Hambleton, unpub. except for a DNA-free, negative control. PCR ampli- data). Further, RFLP analysis of the genus as a whole fications were carried out on a DNA Thermal Cycler suggested that for some pairs of species, morpholog- (Gene E. Techne Ltd., Princeton, New York) using ical variation might only be significant at the subspe- the following cycle parameters: 94 C denaturation for cies