This Article Abstract Full Text (PDF) Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Larsson, E. Articles by Larsson, K.-H. Search for Related Content PubMed Articles by Larsson, E. Articles by Larsson, K.-H. Agricola Articles by Larsson, E. Articles by Larsson, K.-H.

Phylogenetic relationships of russuloid basidiomycetes with emphasis on aphyllophoralean taxa

     Botanical Institute, Göteborg University, P.O. Box 461, SE 405 30 Göteborg, Sweden

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CLADE CHARACTERISTICS AND... NEW COMBINATIONS LITERATURE CITED

 

Many homobasidiomycetes are characterized by a combination of gloeocystidia and amyloid basidiospores. They display a great variation in basidioma morphology, including erect and effused forms and gilled and nongilled forms. Earlier studies have shown these taxa to be related, and the group has been named the russuloid clade. Phylogenetic relationships among russuloid basidiomycetes were investigated using sequence data from the nuclear 5.8S, ITS2 and large-subunit rDNA genes. A dataset including 127 ingroup sequences representing 43 genera and ca 120 species were analyzed by maximum-parsimony and neighbor-joining methods. The sampling of taxa had an emphasis on nongilled taxa and two-thirds of the species possessed corticioid basidiomata. Thirteen major well-supported clades were identified within the russuloid clade. All clades except one include corticioid species. Ten characters from basidioma morphology and cultured mycelium were observed and evaluated. Results suggest that gloeocystidia are a synapomorphy for taxa within the russuloid clade while the amyloidity of spores is inconsistent. The ornamentation of spores and type of nuclear behavior seems to be informative characters at genus level. The agaricoid genera Lactarius and Russula are nested in a clade with corticioid species at the basal position. The new combinations Boidinia aculeata, Gloeodontia subasperispora, Gloeocystidiopsis cryptacantha and Megalocystidium wakullum are proposed.

Key words: amyloid spores, corticioid basidiomata, gloeocystidia, Homobasidiomycetes, nuclear rDNA, phylogeny, russuloid clade, sulfobenzaldehyde reaction

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CLADE CHARACTERISTICS AND... NEW COMBINATIONS LITERATURE CITED

 
The classification of macrofungi traditionally has relied almost entirely on macro- and microanatomical features of the basidiome (e.g., Fries 1874, Patouillard 1900, Jülich 1981). However, anatomical simplicity, a scanty fossil record and high phenotypic plasticity make it difficult to separate the traces of ancestry from instances of parallel evolution. As a result, many fungal genera and families that we are familiar with now are shown to be highly artificial. A few examples are Coprinus, the ink-cap genus, (Redhead et al 2001), the coral fungi in Clavariaceae (Pine et al 1999) and Corticiaceae, the family partly at focus in this study (Hibbett and Thorn 2001).

Along with morphological traits, staining reactions also have been widely applied in fungal systematics. Perhaps the best-known example is Melzer's solution, which is used to detect some of the variable polysaccharide components of fungal cell walls. The active substance in Melzer's is iodine. Some polysaccharides react with iodine to produce a bluish-violet (amyloid) coloration, while others turn reddish-brown (dextrinoid). Another chemical test involves sulfuric benzaldehyde, usually in the form of sulfovanilline. When applied to fungal tissue, a dark violet coloration occurs in cells containing certain sesquiterpenes (Gluchoff-Fiasson and Kühner 1982). The reliability of these tests is subject to some uncertainty because color development is influenced by tissue age and condition and also by age and composition of the reagents. Despite these uncertainties, both tests have been used extensively in fungal taxonomy for delimiting species, genera and families (Boidin 1958, Lemke 1964, Eriksson and Ryvarden 1975).

A number of basidiomycetes combine an amyloid reaction of the basidiospore wall with a sulfo-positive reaction (SA+) of thin-walled, tubular or bladder-like cystidia, so-called gloeocystidia. These examples illustrate the variety of fungal forms having such properties: Russula Pers. and Lactarius Pers., known as important ectomycorrhizal partners in forest ecosystems (Gardes and Bruns 1996, Smith and Read 1997) and also widely recognized and collected for consumption; coral fungi in the genus Hericium Pers. growing on stumps and living hardwood trees and highly treasured as medical mushrooms in Asia (Kawagishi et al 1993, 1996); the aggressive parasite Heterobasidion annosum, a polypore causing great economic losses to the forest industry (Stenlid 1986) and Echinodontium tinctorium, the hydnoid Indian paint fungus, also associated with aggressive decay of coniferous trees (Thomas 1958); several thin inconspicuous corticoid species of the genera Gloeocystidiellum Donk, Boidinia Stalpers & Hjortstam and Gloiothele Bres., which mainly live as saprobionts on different kinds of deadwood (Eriksson and Ryvarden 1975).

Donk (1971) was the first to discuss a possible relationship between taxa such as those mentioned above and other groups possessing a system of gloeoplerous hyphae (gloeocystidia) and amyloid basidiospores. His hypothesis was further expanded and developed by Oberwinkler (1977), who also named the entire group the Russulales. Further arguments for recognizing a unique russuloid lineage among the homobasidiomycetes have come from recent molecular phylogenetic studies (Hibbett and Donoghue 1995, Hibbett et al 1997). According to Hibbett and Thorn (2001), the clade includes taxa formerly placed in the families Auriscalpiaceae Maas Geest., Bondarzewiaceae Kotl. & Pouzar, Clavicoronaceae Corner, Corticiaceae Herter sensu lato, Echinodontiaceae Donk, Hericiaceae Donk, Lachnocladiaceae DA Reid, Peniophoraceae Lotsy, Polyporaceae Fr. ex Corda sensu lato, Russulaceae Lotsy, and Stereaceae Pilàt and is estimated to hold about 1000 described species (Hibbett and Thorn 2001 extrapolating from Hawksworth et al 1995).

Taxa with effused, corticioid basidiomata, a smooth hymenophore, and russuloid staining reactions first were placed together in Gloeocystidiellum Donk. The history and scope of the genus was discussed thoroughly by Donk (1956, 1964). Eriksson and Ryvarden (1975) regarded Gloeocystidiellum as unnatural and suggested that the taxa known in Northern Europe could be divided into seven groups. All these groups since have been segregated as genera (Hagström 1977, Jülich 1978, Hallenberg 1980, Hjortstam and Stalpers 1982, Hjortstam 1987b, Boidin et al 1997a), and additional genera based on species from other parts of the world have been added (Jülich 1982, Wu 1995, 1996).

To elucidate evolutionary relationships within the russuloid clade, a dataset based on nuclear rDNA sequence data (5.8S, ITS2 and 26S) was constructed. Taxa were selected with a particular emphasis on corticioid species, but the range covers all major groups of russuloid taxa except gasteroid and secotioid forms (Hibbett and Thorn 2001). We used this dataset to (i) identify major clades among the russuloid fungi, (ii) explore how these clades correlate with previous morphology based classifications and (iii) redefine corticioid genera within the russuloid clade.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CLADE CHARACTERISTICS AND... NEW COMBINATIONS LITERATURE CITED

 
Sampling of taxa – The sampling (Table I) was guided by earlier classifications of Gloeocystidiellum in a wide sense and by the discussions in Donk (1964), Jülich (1981), Stalpers (1996) and Hibbett and Thorn (2001). Nomenclature of the species included follows Nordic Macromycetes vol 3 (Hansen and Knudsen 1997), Hjortstam and Larsson (1995) and Hjortstam (1998), except in those cases where we suggest new taxonomic arrangements. We have adopted the convention for clade names introduced by Moncalvo et al (2002). Clade names are preceded by a slash, are spelled in lower-case letters and never italicized. We aimed to include representatives from all groups with a combination of amyloid basidiospores and gloeocystidia but also species with only the latter characteristic, providing they stain with sulfovanillin. The amyloid reaction is much more widespread and can be found in a number of genera, e.g., in the corticioid Amyloathelia Hjortstam & Ryvarden, Aphanobasidium Jülich, and Melzericium Jülich, in the poroid Anomoporia Pouzar, and in the agaricoid Catathelasma Lovej., Hydropus (Kühn) Singer, Melanoleuca Pat., Mycena (Pers. : Fr.) Roussel, and Panellus P. Karst. No comprehensive sampling was done from groups where basidiospore amyloidity is not combined with typical gloeocystidia.

All sequences first were aligned with an extensive dataset holding more than 600 taxa sampled from all major groups of homobasidiomycetes (data not shown). This dataset is continuously expanded and used as an in-house tool for sequence quality control, approximate phylogenetic placement of new sequences and as a sampling guide. Trees are generated with the neighbor-joining method and the Hasegawa-Kishino-Yano 85 (HKY85) substitution model. Putative russuloid taxa that did not cluster with the russuloid clade or showed ambiguous placement were excluded from the final dataset.

For 21 species two collections were sequenced to verify results but only one sequence was included in the final dataset. In a few cases, two sequences with the same species name are included but then the nucleotide sequences diverge and they might represent different species or originate from different geographical areas.

Basidioradulum radula and Trichaptum abietinum were selected for rooting of trees because several molecular studies suggest the hymenochaetoid clade as sister group to the russuloid clade (Binder and Hibbett 2001, Hibbett et al 1997, 2000, Larsson 2001), although consistently with no support or weak support. Seven additional, more distantly related species from Sistotrema Fr. and the heterobasidiomycete genera Exidia Fr. and Auricularia Bull. were added to the outgroup.

As standard mounting media for microscopic examinations of specimens, 2% KOH, Melzer's reagent and sulfovanillin have been used (Moser 1978).

Molecular techniques – DNA was isolated from herbarium specimens and from cultured mycelia. Mycelia were grown at room temperature for 2 wk in 50 mL MYG liquid media (1% malt extract, 0.4% yeast extract, 1% glucose). Mycelia were harvested and dried between sheets of filter paper and ca 50 mg was placed in microcentrifuge tubes. From herbarium specimens ca 3 x 3 mm of hymenium was used to extract DNA. DNA extractions were carried out using a modified 2% CTAB method (Savolainen et al 1995). Preparations from some of the herbarium specimens were further purified with Gene Clean (Bio 101 Inc.) to exclude PCR inhibitors. The internal transcribed spacer 1 and 2 (ITS 1 and 2) including the 5.8S region of nuclear rDNA, was amplified with primers ITS1F, ITS4B (Gardes and Bruns 1993), ITS1 and ITS4 (White et al 1990). Approximately 1200 bp of the 5' end of the large subunit of the rDNA (nuclear LSU) was amplified with LR0R and LR7 (Vilgalys and Hester 1990).

PCR amplifications were performed in 25 mL reactions using either Taq polymerase (Advanced Biotechnologies) together with reaction buffer IV or Ready To Go^TM PCR beads (Amersham Pharmacia Biotech Inc.). The thermal cycling program included 4 min at 95 C, followed by 30 cycles of 30 s at 95 C, 30 s at 52 C, 60 s at 72 C, and then ended by 8 min at 72 C. Fragments were examined on a 1% SeaKem (FMC) agarose gel, and amplified products were purified with QIAquick spin columns (QIAGEN).

Primers used for sequencing of both strands were ITS3, ITS4 (White et al 1990), LR5, LR21, LR3r (http://WWW.biology.duke.edu/fungi.html), and CTB6 (http://mendel.berkeley.edu/boletus.html). Cycle sequencing was carried out using Thermosequenase flourescent labeled primer cycle sequencing kit with 7-deaza-dGTP (Amersham Pharmacia Biotech Inc.). Twenty-five ng of template DNA and 5 pmol Cy5-labeled primer were used per reaction. Sequences were obtained using an ALFExpress (Pharmacia Biotech Inc.) automated sequencer. Sequences were edited and assembled using computer software ALF manager (Pharmacia Biotech) and Sequencher 3.1 (Gene Codes Corp.). Complete sequences were aligned manually using the data editor PAUP* 4.0 (Swofford 1999). Sequences were submitted to GenBank, and accession numbers are given in Table I.

The LSU sequence of Bondarzewia berkeleyi in this study was taken from GenBank (AF218563). The aligned data matrices are available from the corresponding author upon request.

Phylogenetic analyses – Heuristic searches were performed using PAUP* 4.0b8 (Swofford 1999) on a Power Macintosh computer. All transformations were considered unordered and equally weighted. Variable regions with ambiguous alignment were excluded, and gaps were treated as missing data. Heuristic searches with 1000 random-addition sequence replicates, TBR branch swapping and MAXTREES set to 25 000 with restrictions to save 100 trees in each replicate, were performed.

Neighbor-joining analysis (NJ) was performed on the same dataset using the Hasegawa-Kishino-Yano 85 (HKY85) substitution model.

Ten selected trees, with the best-likelihood score, from the initial heuristic searches were used as starting trees in recurrent heuristic searches, with TBR and NNI branch swapping, under the maximum-likelihood criteria (ML) to search for more optimal trees. ML parameter settings corresponded to HKY85+I model, with the nucleotide substitution rate parameters estimated via ML. Searches using TBR swapping were aborted before completion (after 24 h), due to the large dataset and computationally intensive algorithms. Trees with the best-likelihood scores were saved for comparison.

To compare tree topologies of alternative phylogenetic hypotheses for the dataset, several constrained analyses were conducted. Constrained trees forcing monophyly of /peniophorales and keeping it as a sister clade to the rest of the russuloid clade were constructed using Mac Clade 4.0 (Maddison and Maddison 2000). All nodes within the clades were collapsed. Heuristic searches with 100 random-addition sequence replicate, enforcing constraints and saving only trees compatible with constraint topologies were performed using the same taxa and settings as above. Constrained analyses to test the support for monophyly of the G. porosum-clavuligerum complex were performed in the same way.

Topological differences between the constrained phylogenetic tree hypothesis and unconstrained trees were evaluated with Kishino-Hasegawa (KH) maximum-likelihood ratio test (normal approximation, two-tailed test) and the Shimodaira-Hasegawa (SH) test (using RELL bootstrap) implemented in PAUP*.

To improve the resolution within Stereum and allied genera, 27 sequences from the large dataset were realigned to include more characters from the variable ITS2 region. One sequence of Gloeocystidiellum aspellum was added, and Gloeodontia pyramidata was selected as outgroup. In the parsimony analysis, all transformations were considered unordered and equally weighted. Variable regions with ambiguous alignment were excluded, and gaps were treated as missing data. Heuristic searches with 1000 random-addition sequence replicates, TBR branch swapping and MAXTREES set to auto-increase, were performed.

Relative robustness of clades was estimated by bootstrapping, with the following settings. For the large dataset by using 1000 bootstrap replicates, with five random-addition sequences per replicate, TBR branch swapping and MAXTREES set to 25 000 with restrictions to save 100 trees in each replicate. For the Stereum-restricted dataset by using 1000 bootstrap replicates, with 100 random-addition sequences per replicate, TBR branch swapping and MAXTREES set to auto-increase.

Morphological characters – Morphological features of the basidiome, hymenophore and spore surface, presence of clamps and skeletal hyphae, type of life strategy, nuclear behavior and polarity (if available) have been observed or gathered from literature and compiled in Table II . Data on nuclear behavior and polarity are mainly taken from Boidin and Lanquetin (1984b, 1997) and Boidin (1990). To visualize the distribution of observed characters, Table II shows specimens in the order they occur in the strict consensus tree (Fig. 2).

View larger version (48K): [in this window] [in a new window]   FIG. 2. Strict-consensus tree of 24 772 equally parsimonious trees. Bootstrap values for discussed major clades are given as numeric values. Asterisks indicate other bootstrap values above 80%. Names marked with a dot represent genus type species

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CLADE CHARACTERISTICS AND... NEW COMBINATIONS LITERATURE CITED

View larger version (37K): [in this window] [in a new window]   FIG. 1. Phylogeny of the russuloid lineage based on unweighted parsimony analysis of nuclear rDNA sequence data. One of 24 772 equally parsimonious trees depicted as a phylogram (tree length 2468, CI = 0.3002, RI = 0.7297). Filled dots indicate species that have been combined in Gloeocystidiellum. Bootstrap support is noted only for major clades discussed in the text

The phylogenetic analysis recovered 24 772 equally parsimonious trees of 2468 steps (CI = 0.3002, RI = 0.7297). The trees were recovered from 263 islands, where the islands found in different replicates in fact might belong to the same island. A set of 2630 most-parsimonious (MP) trees (10 from each island) were selected and kept for comparisons and analyses. Figure 1 illustrates one of the MP trees presented as a phylogram to show the number of character state changes per branch. The tree selected was one of the MP trees with the best-likelihood score. Species that have been combined within Gloeocystidiellum are marked with filled dots. Figure 2 illustrates the strict-consensus tree of all 24 772 MP trees.

The bootstrap analysis recovered 13 major supported clades and six species on single branches. Each clade and its corresponding bootstrap value are indicated in the phylogram (Fig. 1) and the strict-consensus tree (Fig. 2). We have chosen names that roughly correspond to the current concepts of orders, families and genera in fungal taxonomy (Hawksworth et al 1995). When competing names are available, we have selected the oldest one. The clades are: /peniophorales, /amylostereaceae, /gloeocystidiellum I, /gloeocystidiellum II, /auriscalpiaceae,/gloeodontia, /aleurocystidiellum, /hericiaceae,/bondarzewiaceae, /albatrellus, /scytinostromella,/russulales and /stereales. Taxa on single branches are Wrightoporia tropicalis, W. lenta, Scytinostromella nannfeldtii, Gloeohypochnicium analogum, Echinodontium ryvardenii and Pseudoxenasma verrucisporum. The bootstrap consensus-tree topology was almost identical to the strict-consensus tree with one main difference, a node uniting /stereales and /gloeodontia in the strict-consensus tree. The NJ analysis recovered the same 13 clades as the parsimony analysis but with a major difference in the basal topology. The NJ tree places /peniophorales as a sister clade to the rest of the ingroup (see Fig. 3). None of the MP trees recovered a similar topology. However, short incomplete (nonconstrained) heuristic searches with the same settings as in the parsimony analysis, occasionally recovered a similar topology as the NJ tree. Such trees were three steps longer (2471 steps) than the most-parsimonious trees. Constrained parsimony analyses simulating the basal topology of the NJ tree (/peniophorales forced as a sister clade to the rest of the russuloid clade) recovered 4700 trees of 2473 steps (five steps longer than the unconstrained trees). The constrained trees were found not to be significantly worse than the unconstrained trees, based on the KH test (P = 0.2982–0.6028) and the SH test (P = 0.754–0.959).

View larger version (35K): [in this window] [in a new window]   FIG. 3. Neighbor-joining distance based phylogenetic analysis of the nuclear rDNA sequence dataset. The NJ tree places /peniophorales as a sister clade to the rest of the ingroup, /eurussuloid

Recurrent heuristic searches under maximum-likelihood criteria with starting trees from the MP analysis recovered several trees with a better-likelihood score than the starting tree. In general the TBR branch swapping recovered better likelihood scores than the NNI branch swapping, despite the incomplete analyses. The result might indicate the existence of more optimal trees not recovered in the heuristic searches.

Phylogenetic analysis of the realigned dataset restricted to /stereales recovered 12 MP trees of 407 steps, based on 118 parsimony-informative characters (CI = 0.5971, RI = 0.6840). Figure 4 presents the strict-consensus tree with bootstrap values indicated above branches. Six clades with bootstrap support above 50% were identified: /gloeocystidiopsis, /stereum, /xylobolus, /chelidonium, /cerussatus and /megalocystidium. Conferticium ochraceum is placed on a single branch between /chelidonium and /cerussatus. Aleurobotrys botryosus and Aleurodiscus amorphus also were placed on single branches at the base of the tree.

View larger version (28K): [in this window] [in a new window]   FIG. 4. Strict-consensus tree of 12 equally parsimonious trees, resulting from parsimony analysis of the realigned dataset of /stereales. Basal nodes with bootstrap values above 50% are indicated. Stippled line indicates uncertain support

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CLADE CHARACTERISTICS AND... NEW COMBINATIONS LITERATURE CITED

 
The phylogenetic analyses identify 13 major clades with strong or fair bootstrap support. These clades are recovered always and appear resistant to changes in taxon sampling, character selection and method chosen for analysis. Basal nodes are poorly resolved and most basal branches collapse to a polytomy in the strict consensus and bootstrap trees.

A basal division of our dataset in /peniophorales and /eurussuloid was recovered in the NJ tree (Fig. 3) while the parsimony analyses generally did not support such a topology. As discussed below there are some morphological trends in /peniophorales that sets this clade apart from the rest of the russuloid lineage and the constrained analysis does not reject the NJ topology. A denser sampling within /peniophorales eventually might show if the NJ topology receives increased support.

All species with a gloeoplerous hyphal system that we tested with our large homobasidiomycete dataset turned out to cluster with the russuloid clade. This gives strong indications that possession of a gloeoplerous hyphal system, often primarily observed as tubular gloeocystidia, is a synapomorphy for the russuloid clade. This assumption, however, must be tested finally with a more inclusive dataset. In most taxa the gloeoplerous system gives a positive reaction with sulfobenzaldehyde, while in others it reacts positive only when fresh but loses this reaction after some period of storage (e.g., Asterostroma). A third group is formed by those taxa that never give a positive reaction with sulfobenzaldehyde. However, sequence analysis and morphological features support the hypothesis that all gloeoplerous systems within the russuloid lineage, whether sulfo-positive or not, are homologous (Table II).

In species having basidiomes with a dense texture dominated by thick-walled hyphae, the gloeoplerous hyphae might be hard to detect, e.g., in Heterobasidion annosum and Bondarzewia montana, which has led to conflicting views presented in the literature (Redhead and Norvell 1993). In both cases sulfo-positive gloeocystidia develop in culture (Gluchoff-Fiasson et al 1983). In Stereum Pers., typical gloeoplerous hyphae are lacking but it seems likely that the lactiferous hyphae present in most Stereum species are homologues.

The amyloid reaction of the spore wall is a second character often used to characterize russuloid species. However, amyloidity is a common phenomenon among homobasidiomycetes and not a synapomorphy for the russuloid clade. For example, Mucronella Fr., possesses amyloid spores and previously was associated with Hericiales based on basidiome morphology but the genus lacks gloeocystidia. In agreement with this, phylogenetic analysis of rDNA sequence data places Mucronella outside the russuloid clade (data not shown). The genus Peniophora Cooke lacks an amyloid reaction entirely while other genera, e.g., Scytinostroma Donk and Albatrellus S.F. Gray, contain both amyloid and non-amyloid species.

Most species in the russuloid clade have basidiospores with an ornamented surface. The nature of this ornamentation has been extensively studied (e.g., Capellano and Keller 1978, Keller 1986, 1997). An outer layer, tectum, forms the ornamentation, and the same layer is responsible for the amyloid reaction. The tectum can be covered by up to three additional layers. Species with smooth basidiospores lack a tectum, and in TEM no difference in spore wall structure between amyloid and non-amyloid spores can be detected (Keller 1997). We invariably have found the presence or absence of an ornamentation to be a generic character within the russuloid clade (but see /hericiaceae below). The fine structure of the spore wall as observed in TEM has been studied in a limited number of species only and its use as a phylogenetically informative character cannot be evaluated.

Corner (1932a, b) introduced hyphal analysis and the concept of hyphal systems in fungal anatomy and taxonomy. He defined skeletal hyphae as thick-walled to subsolid, nonseptate cells that retain ability for growth at the thin-walled tip. They arise from generative hyphae. He also coined the terms monomitic for species with only generative hyphae, dimitic for species with generative and skeletal hyphae, and trimitic for species with generative, skeletal, and binding hyphae. Corner's ideas gained wide acceptance and have greatly influenced basidiomycete systematics in general and generic concepts in polypores in particular. For an overview of the subject, see Pegler (1996) and Hibbett and Thorn (2001).

In the russuloid clade, skeletal hyphae occur in many species and several genera have been introduced for species with a dimitic hyphal system, viz. Scytinostromella Parmasto, Wrightoporia Pouzar and Confertobasidium Jülich. However, among the clades, we identify only /scytinostromella as completely dimitic. Genera such as Scytinostromella and Wrightoporia seem to wither when scrutinized by molecular methods, and Confertobasidium has its closest relative in the monomitic Metulodontia Parmasto. The clade /auriscalpiaceae includes clearly dimitic species such as Gloiodon strigosum and Auriscalpium vulgare and the monomitic Dentipratulum bialovicense. A similar picture is found in /gloeodontia, where dimitic species such as Gloeodontia discolor cluster with the monomitic Gloeocystidiellum subasperisporum. We conclude that skeletal hyphae as presently defined have been overestimated as a taxonomic character at the generic level. Our results indicate that dimitic hyphal systems have evolved many times within the russuloid clade and support the views presented by Hibbett and Thorn (2001).

Recent studies have shown that the agaricoid genera Russula and Lactarius form a monophyletic clade together with their gasteroid and pleurotoid relatives (Miller et al 2001, Calonge and Martin 2000, Henkel et al 2000, Binder and Bresinsky 2002). We show that this clade is nested among corticioid taxa in what we call /russulales. Bootstrap support for /russulales is good (86%) but internal branching is mainly unresolved. A notable exception is the basal node connecting Boidinia furfuracea to the rest of the clade, where bootstrap support is 76%. Boidinia furfuracea is a wood-inhabiting fungus that forms thin, white, strictly resupinate basidiomes on decaying conifer wood. According to Nakasone (1990), it is capable of producing extracellular oxidases and hence causes a white rot. There is no report of mycorrhizal activity connected to Boidinia or Gloeopeniophorella, the second corticioid genus in /russulales. The result suggests that the ancestor to the agaricoid radiation in Russula and Lactarius had a corticioid basidiome and was saprotrophic. This is in accordance with the conclusions in Hibbett et al (2000) that evolution has witnessed several independent shifts in nutritional mode from saprotrophy to mycorrhizal associations. It also is consistent with an evolution from simple to complex fruiting bodies as put forward in Hibbett and Binder (2002).

	CLADE CHARACTERISTICS AND TAXONOMY

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CLADE CHARACTERISTICS AND... NEW COMBINATIONS LITERATURE CITED

 
/peniophorales (bootstrap support 95%) – Our sampling in this clade was too restricted to allow a detailed phylogenetic analysis. Peniophora, Scytinostroma and Vararia P. Karst. together hold close to 150 described species but here are represented by only eight specimens. Within /peniophorales, only two subclades are distinct enough to be recognized, viz. /asterostromataceae and /metulodontia (Fig. 2).

There is no obvious morphological synapomorphy for /peniophorales. The most striking difference from /eurussuloid is the predominance of smooth-walled basidiospores and the tendency toward non-amyloid spore-walls. In addition, species in this clade almost invariably have basidiomes with a smooth hymenophore, while hydnoid, poroid and lamellate basidiomata occur frequently in most of the other clades. Monophyly for the same group of taxa that we here call /peniophorales was detected already in several other studies (Hibbett et al 1997, 2000, Hibbett and Donoghue 2001, Hibbett and Binder 2002).

The genera Asterostroma Massee, Dichostereum Pilàt, Lachnocladium Lév., Scytinostroma and Vararia usually have been referred to a separate family, Lachnocladiaceae (Reid 1965). They are held together morphologically by the thick-walled, dextrinoid hyphae termed astero-, dendro- or dichohyphidia. These hyphidia are functionally equivalent to binding hyphae found in many polypores. It is the dominating hyphal type in many species, imparting them with a more or less tough consistency. Most species in Lachnocladiaceae do not form a closed hymenium. Instead, basidia are initiated deep among the hyphidia and penetrate to the surface just enough to freely shed their spores. This kind of hymenium has been termed a catahymenium as opposed to the more familiar euhymenium with a closed palisade of basidia (Lemke 1964). Basidiomes with a catahymenium are adapted to resist periods of drought and to quickly resume sporulation when conditions become more favorable.

In our analyses Peniophora, Gloiothele and Vesiculomyces Hagström are nested among the traditional members of Lachnocladiaceae. No species in these genera have dichohyphidia or asterohyphidia, but a few species in Peniophora have dendrohyphidia. Since Peniophora is the oldest genus name, we prefer to have that situation reflected in the clade name.

The four species of Scytinostroma in our dataset all occur on separate branches that are distantly separated. They are morphologically quite different. Scytinostroma portentosum has simple-septate hyphae, amyloid subglobose spores and SA+ gloeocystidia, S. odoratum has simple-septate hyphae, non-amyloid ellipsoid spores and SA- gloeocystidia, S. galactinum has nodose-septate hyphae, non-amyloid (except for the suprahilar patch) ellipsoid spores and SA+ gloeocystidia, and S. jacksonii has nodose-septate hyphae, non-amyloid, ovoid spores and SA- gloeocystidia. In addition, the two species representing Vararia do not cluster together and also are separated from the four Scytinostroma branches.

The taxonomic distinction between Scytinostroma and Vararia has been questioned (Boidin and Lanquetin 1987, Boidin et al 1998, Hallenberg 1985, Stalpers 1996). However, there has been general agreement that the two genera are closely related and that they together make up a natural group. Our results strongly suggest that neither skeletal hyphae nor their branching patterns have any predictive power in a phylogenetic context.

Scytinostroma was introduced for species with dextrinoid, sparsely branched, skeletal hyphae, differing from the richly branched dichohyphidia characteristic of Vararia (Donk 1956). It is typified by Scytinostroma portentosum, which in our analyses takes a position close to Gloiothele. We estimate that less than 10 species have characters corresponding to those of the type. However, a thorough molecular investigation of Scytinostroma, Lachnocladium and Vararia is necessary before the circumscription of these genera can be settled.

Five species of Dichostereum (40% of known species including the type D. durum) cluster together and appear to be a monophyletic group. However, in the bootstrap tree the clade collapses to a polytomy. Like Vararia, the genus has dichohyphidia but differs by globose, coarsely ornamented and strongly amyloid spores.

The two Peniophora species sampled cluster together. This large genus with many closely related species is ecologically quite distinct. Species typically are found in exposed situations on dead but still-attached branches. Most species in Peniophora have nodose-septate hyphae, SA+ gloeocystidia, and characteristic thick-walled incrusted cystidia (metuloids). Spores are always smooth and non-amyloid. Basidiomata appear well adapted to desiccation and have a dense, often strongly pigmented hyphal structure.

Hallenberg et al (1996) investigated the phylogenetic relationships in Peniophora. Three species groups were identified, and the genus was confirmed as uniform and well distinguished. In the MP tree, Peniophora is positioned together with Vararia investiens as a sister group to Dichostereum. However, this arrangement has no bootstrap support.

/asterostromataceae (99%) – This clade includes Asterostroma, Scytinostroma portentosum (see above), Gloiothele and Vesiculomyces. The position of Gloiothele and Vesiculomyces in /peniophorales was unexpected. Both genera are characterized by simple septate hyphae, a monomitic hyphal system and globose to subglobose, smooth, amyloid spores. Gloiothele lamellosa, type species of the genus, has a variable hymenophore that can be odontioid, irpicoid or subporoid. Thus it is not surprising that the species appears under additional names as Gloeocystidiellum irpiscescens Boidin and Vesiculomyces epitheloides Boidin & Lanq. (Hjortstam 1987a). The three specimens included in our sampling represent some of the hymenophore variation. Molecular data support the conclusion by Hjortstam (1987a) that these names refer to one and the same species.

Vesiculomyces citrinus was segregated from Gloeocystidiellum because gloeocystidia in basidiomata are SA- (Hagström 1977). Boidin (1958) and Maekawa et al (1982) have reported that SA+ gloeocystidia could be found in cultured mycelium of V. citrinus. However, we could not confirm this observation in our own cultures. The position of V. citrinus in the tree as a sister group to Gloiothele fits well with morphological characters such as simple septate hyphae, narrowly clavate basidia and smooth globose amyloid spores with a prominent apiculus. We suggest that Vesiculomyces be retained as a separate genus.

Asterostroma (type species Corticium apalum Berk & Broome = A. muscicola) appears monophyletic with 71% bootstrap support. In addition, the genus is morphologically well characterized by unique asterohyphidia. This type of hyphidium is short, has a dense branching and develops from thin-walled hyphae. Because long, skeletal-like hyphidia are lacking, Asterostroma has soft, fragile basidiomata and they usually are found in moist, sheltered places. Basidiospores are always globose and either smooth or ornamented. However, the ornamentation does not seem to be homologous to the ornamentation seen in other russuloid species. In the latter case, ornamentation is caused by material deposited in the outer layer of the spore-wall, while in Asterostroma ornamentation seems to consist of lobes formed by several layers of the spore wall. Boidin et al (1997b) recognized 14 species in Asterostroma. Nine of them are very similar morphologically and separated mainly by differences in spore size and ornamentation. Our sampling includes two of these morphotypes, and we found the sequences to be almost identical. Additional sequencing in this species complex should be undertaken.

/metulodontia (100%) – In all analyses there is strong bootstrap support for /metulodontia as the most basal subclade in /peniophorales. Parmasto (1968) introduced Metulodontia to accommodate species with thick-walled incrusted cystidia, so-called metuloids. Eriksson and Ryvarden (1976) showed that the type species, Metulodontia nivea, differs from all other species originally assigned to Metulodontia by having SA+ gloeocystidia, and they regarded Metulodontia as a monotypic genus of uncertain affinity. Here we show that Metulodontia, despite lacking an amyloid reaction of the spores, belong to the russuloid clade.

Jülich (1972) introduced Confertobasidium for athelioid species with brownish basal hyphae and with C. olivaceoalbum as the type. Hjortstam (1987a) showed that the type species had been misunderstood already by Bourdot and that two species—one with SA+ gloeocystidia and skeletal hyphae and one lacking these characters—were mixed into the species concept. The type specimen selected by Jülich has the gloeocystidia and skeletal hyphae. For specimens lacking those elements, the epithet fuscostratus Burt is available, now referred to Leptosporomyces Jülich.

Confertobasidium olivaceoalbum and Metulodontia share an athelioid basidiome construction, small ellipsoid smooth basidiospores and short, narrow SA+ gloeocystidia confined to the hymenium. It is a matter of taste if one wishes to combine the two species in one genus or keep them separate. Metulodontia has inamyloid spores, a monomitic context and two kinds of cystidia, while Confertobasidium has amyloid basidiospores, a dimitic context and only one kind of cystidium.

Ginns and Lefebvre (1993) recently transferred Corticium olivaceoalbum to Scytinostromella. They stressed the presence of gloeocystidia, skeletal hyphae and an athelioid basidiome as the main reasons for the transfer. In our tree, Confertobasidium olivaceoalbum is not placed close to Scytinostromella heterogenea, the generic type of Scytinostromella.

/amylostereaceae (73%) – The clade includes Amylostereum and Artomyces pyxidatus (= Clavicorona pyxidata [Pers. : Fr.] Doty). No obvious morphological characteristics indicate a close relationship between these taxa, and no one before has suggested such a relation. Amylostereum has leathery, more or less reflexed basidiomata with a smooth hymenophore, while Artomyces Jülich has erect, clavarioid basidiomata with a characteristic branching pattern. Connections can be found in micromorphology. They share a constantly clamped hyphal system, a SA+ gloeoplerous system and smooth, amyloid spores. Amylostereum usually is described as lacking gloeocystidia, but Eriksson and Ryvarden (1973) noticed thin-walled gloeocystidia-like organs in A. chailletii. In culture, SA+ gloeocystidia regularly occur (Boidin and Lanquetin 1984a, Nakasone 1990).

In the analyses by Hibbett et al (1997) and Pine et al (1999), Clavicorona pyxidata is nested in a clade with Auriscalpium vulgare and Lentinellus spp. The relation to Lentinellus is supported as well in later studies (Hibbett et al 2000, Hibbett and Donoghue 2001). Our contradictory results prompted us to resequence the culture used to generate our sequence of Artomyces pyxidatus, but the outcome was identical. However, it is possible that the culture actually represents something else. The position of Amylostereum is discussed with /bondarzewiaceae below.

Clavicorona Doty is typified by C. taxophila (Thom) Doty, a small, nonbranched species, quite different from the large, multibranched C. pyxidata. Jülich (1981) questioned the homogeneity of the genus and referred C. pyxidata to a separate genus, Artomyces. Recent systematic and monographic studies in Artomyces and Clavicorona support this arrangement (Lickey 2002).

/gloeocystidiellum I (92%) and II (66%) – Gloeocystidiellum porosum is the generic type for Gloeocystidiellum. In our tree, it clusters with Boidinia granulata and two unidentified specimens. Gloeocystidiellum clavuligerum was first regarded as a synonym of G. porosum but reinstated as a distinct species after a closer investigation of morphology and culture characteristics (Nakasone 1982). Larsson and Hallenberg (2001) found that molecular data support G. porosum and G. clavuligerum as distinct species and that each of them also encompasses other closely related taxa. Our analyses confirm the results of Larsson and Hallenberg (2001). Gloeocystidiellum clavuligerum and G. porosum do not cluster together in any of our analyses irrespective of method, sampling or characters selected. However, constrained trees forcing monophyly of G. porosum/clavuligerum were found not to be significantly worse than the MP trees.

One mitochondrial small-subunit rDNA sequence of Gloeocystidiellum porosum was included in the analyses by Hibbett and Donoghue (2001), where it occurred nested with Laxitextum bicolor and Dentipellis separans (Peck) Donk. The apparent deviation from our results might be explained by the restricted sampling used by Hibbett and Donoghue (2001). In their dataset, G. porosum simply had no closer relative to associate with. It should be noted that Dentipellis separans is a younger synonym of Dentipellis leptodon (Mont) Maas Geest (Ginns 1986). However, most collections in American herbaria named D. separans in fact belong to Dentipellis dissita (Ginns 1986). This is also the correct name for the specimen and the subsequent culture collected by N. Hallenberg and by him determined as D. separans. The culture (FCUG 581) was the sequence source for both D. separans used by Hibbett and Donoghue (2001) and for D. dissita used in this study!

All species in Gloeocystidiellum sensu stricto are strictly resupinate and have a smooth hymenophore. They are monomitic, have nodose-septate hyphae, gloeocystidia that are either SA+ or SA-, ellipsoid ornamented amyloid spores and a heterothallic tetrapolar mating system. Gloeocystidiellum bisporum is a haploid mitosporic derivative from G. clavuligerum, and its status as a distinct species can be questioned.

Our analyses imply that Gloeocystidiellum porosum and G. clavuligerum have a long history as separate taxa, but the generally low resolution at basal nodes prevents definite conclusions about phylogenetic relationships and generic limits. Because morphological characteristics of the two clades overlap almost entirely, we think it would be impractical to introduce a new genus for the taxa around G. clavuligerum.

/auriscalpiaceae (86%) – This clade corresponds to the family Auriscalpiaceae. At its creation, Maas Geesteranus (1963) included the hydnoid genera Auriscalpium Gray and Gloiodon P. Karst. and the lamellate Lentinellus P. Karst. Most of the species have a dimitic hyphal system with clamps and gloeocystidia that give a SA+ reaction. Spores are subglobose to ellipsoid, ornamented, and with a strong amyloid reaction.

Analyses recovered two distinct subclades, with 71% and 86% bootstrap support respectively. The first subclade includes five species with a hydnoid hymenophore, viz. Auriscalpium vulgare, A. villipes, Gloiodon strigosum, G. nigrescens and Dentipratulum bialoviesense. Auriscalpium vulgare is stipitate with a reniform pileus. The abhymenial surface is hairy and dark brown. Auriscalpium villipes has a similar construction, but pilei have only a short stipe or are sessile and more or less broadly attached. Gloiodon species are resupinate or effused-reflexed, and Dentipratulum Domanski finally has a strongly reduced basidiome, consisting of scarcely separate spines connected by a barely visible sterile mycelium. The inclusion of Dentipratulum in /auriscalpiaceae supports earlier results by Boidin et al (1998).

Gloiodon and Auriscalpium are very similar, as already noted by several authors (e.g., Koski-Kotiranta and Niemelä 1988). They are dimitic, and the gloeoplerous hyphae are less obvious than in the other genera in this clade. The only difference is the effused-reflexed basidioma in Gloiodon as opposed to the stipitate basidioma in Auriscalpium. It must be questioned if Gloiodon should be maintained as an independent genus.

The second subclade consists of the lamellate genus Lentinellus and has a bootstrap support of 86%. The genus is characterized by sessile or stipitate, flabelliform basidiomata with serrate lamellae. The pileus is convex to depressed and glabrous or tomentose-villose. Five or six species occurring in Northern Europe, including the type species L. cochleatus, have been sequenced. Lentinellus ursinus and L. castoreus are almost identical and differ only by the distribution of hair on the cap surface and preference for conifer versus deciduous wood. Some authors have accepted them as separate species (Ryman and Holmåsen 1984, Watling and Gregory 1989), while others have treated them as synonyms (Miller and Stewart 1971, Printz 1986, Stalpers 1996). In this study, the glabrous and tomentose variants were shown to have identical nucleotide sequences. However, considering their different ecology, it is still possible that they genetically behave as independent species.

/gloeodontia (78%) – This clade includes Gloeodontia Boidin and Gloeocystidiellum subasperisporum. A relationship between these taxa was shown by Boidin et al (1998) based on ITS sequence data. Nodose septate hyphae, verrucose amyloid spores and SA+ gloeocystidia unite the species. Gloeodontia species have hydnoid to irpicoid basidiomata, more or less thick-walled to dimitic hyphae, encrusted hyphoid cystidia in the hymenium and ellipsoid to subglobose basidiospores. Gloeocystidiellum subasperisporum has a smooth hymenophore, a monomitic hyphal system, no encrusted cystidia, and reniform basidiospores. Gloeodontia columbiensis and G. discolor are heterothallic unifactorial (Boidin and Lanquetin 1984b). Gloeocystidiellum subasperisporum is confirmed as heterothallic, but factorial type is not identified. All three species have a normal nuclear behavior (Boidin and Lanquetin 1984b, Boidin et al 1997a).

Hjortstam and Ryvarden (1988) referred Gloeocystidiellum subasperisporum to Amylosporomyces S.S. Rattan, a genus described as lacking any kind of cystidia (Rattan 1977). We restudied type material of Amylosporomyces echinosporus and found that it has numerous gloeocystidia and spores that are almost identical to those in Gloeocystidiellum subasperisporum. The type collection is not in the best condition and the gloeocystidia do not react with sulfovanillin, which might explain why they have gone unnoticed. We regard Amylosporomyces echinosporus as a synonym of Gloeocystidiellum subasperisporum. In our tree, G. subasperisporum occurs nested in Gloeodontia and we suggest that G. subasperisporum be transferred to Gloeodontia.

/aleurocystidiellum (99%) – Aleurocystidiellum subcruentatum is a dimitic species with discoid basidiomata and large, minutely verrucose, amyloid spores. The skeletal hyphae terminate in the hymenium as cystidia-like, more or less encrusted elements. Gloeocystidia or gloeoplerous hyphae seem to be lacking. These features prompted Lemke (1964) to segregate subcruentatum from Aleurodiscus Rabenh. ex J. Schröt., where the species had been placed by most authors because of the discoid basidiomata and the amyloid basidiospores. Boidin et al (1968) found that Aleurodiscus disciformis exhibited almost identical culture characteristics, although basidiomata differed by being monomitic and by possessing SA+ gloeocystidia. Consequently they intended to make the appropriate combination in Aleurocystidiellum Lemke but failed to include a reference to the basionym. Tellería (1990) later provided a correct combination. Hallenberg and Parmasto (1998) found that molecular data supported a close relationship of the two species. They also studied basidiospores in SEM and showed that the ornamentation in Aleurocystidiellum species is identical and clearly different from Aleurodiscus amorphus, type of Aleurodiscus. Both Aleurocystidiellum species are heterothallic tetrapolar and have a heterocytic nuclear behavior (Boidin et al 1968). They also have a similar ecology; they grow on the bark of living trees. In the tree presented by Wu et al (2001) Aleurocystidiellum is interpreted as belonging to the ingroup but actually is a sister group to the rest of the ingroup. This position is still compatible with our tree, where Aleurocystidiellum has no clear connection to other clades.

/hericiaceae (87%) – This clade includes three genera: Hericium with large coralloid, pileate or substipitate basidiomata supporting a strongly hydnoid hymenophore, Dentipellis Donk with effused-reflexed basidiomata and a hydnoid hymenophore, and Laxitextum Lentz with effused-reflexed basidiomata and a smooth hymenophore. They are all parasitic or saprotrophic and associated with a white rot. Morphological characters uniting the clade are the monomitic hyphal system with clamp connections, SA- gloeocystidia and relatively small subglobose to ellipsoid, slightly thick-walled and amyloid spores. Species seem to have a bifactorial mating system and are either heterothallic or amphithallic, and they have a normal nuclear behavior (Boidin and Lanquetin 1984b, Boidin 1990).

Hericium cirrhatum usually has been referred to a separate genus Creolophus P. Karst. One reason is that basidiospores have been interpreted as smooth. However, when observed in SEM the spore surface appears slightly rugose (Keller 1997). We cannot judge whether this is an artifact caused by the preparation method or a true ornamentation. Since Hericium cirrhatum in our tree occurs nested among Hericium species that all have ornamented spores, we find it most logical to regard Creolophus as a superfluous genus.

The specimen called Dentipellis sp. is resupinate with a smooth hymenophore. It is another striking example that basidiome morphology and hymenophore configuration are mostly of limited value as indicator of relationships.

/bondarzewiaceae (78%) – This clade includes species with resupinate, effused-reflexed or pileate basidiomata and with smooth, poroid or hydnoid hymenophore configuration. All species cause a severe white rot and several species, e.g., Bondarzewia berkeleyi, Echinodontium tinctorium and Heterobasidion annosum, can attack living trees. The hyphal system is characterized as di- or trimitic but, according to Stalpers (1979), the skeletal hyphae are merely sclerified generative hyphae with more distant septa than usual. Donk (1964) did not regard Bondarzewia, Echinodontium, Heterobasidium and Laurilia as related, and he referred them all to different families, viz. Bondarzewiaceae, Echinodontiaceae, Polyporaceae and Stereaceae. Stalpers (1979) suggested Bondarzewiaceae sensu Donk also should include Heterobasidion Bref., Echinodontium and Laurilia Pouzar. His view was further strengthened by the discovery that both Bondarzewia and Heterobasidion have a system of SA+ gloeoplerous hyphae (Gluchoff-Fiasson et al 1983). A summary of opinions on family arrangements can be found in Redhead and Norvell (1993). In the tree presented by Hibbett et al (1997), Bondarzewia, Echinodontium and Heterobasidion belong to the same superclade but not close together. In subsequent studies (Hibbett et al 2000, Hibbett and Donoghue 2001, Binder and Hibbett 2002, Hibbett and Binder 2002), Bondarzewia and Heterobasidion cluster together but always distinctly removed from Echinodontium. Monophyly for Heterobasidion and Bondarzewia also was detected by Bruns et al (1998).

Gross (1964) found the pileate-hydnoid Echinodontium and the resupinate-smooth Laurilia to be congeneric. They share constant presence of clamps, thick-walled encrusted cystidia, and a brown-colored context. Molecular data presented here and by others (Hibbett and Donoghue 2001, Hibbett and Binder 2002) support this view, and we suggest that Laurilia be relegated to synonymy.

Amylostereum and Echinodontium cluster together with high bootstrap support in the analysis by Hibbett et al (2000), Hibbett and Donoghue (2001), and Binder and Hibbett (2002). In our analyses, Amylostereum always is removed from Echinodontium instead forming a moderately supported clade with Artomyces pyxidatus. We have no explanation for this discrepancy; the rigorous four-gene analyses performed by Binder and Hibbett (2002) presents strong arguments for a phylogenetic connection between Amylostereum and Echinodontium. Similar morphological arguments can be found in the numerous conical, apically encrusted cystidia present in both genera.

Echinodontium ryvardenii is a recently described species only found on Juniper on Corsica (Bernicchia and Piga 1998). Despite having a hydnoid hymenophore, a dimitic hyphal system, amyloid, ornamented spores and SA+ gloeocystidia, it is not related to the other Echinodontium species. Its true affinities are unclear.

/albatrellus (100%) – The four species of Albatrellus form a monophyletic group that includes the type A. ovinus. All species in the genus have stipitate basidiomata with a poroid hymenophore. They form ectomycorrhiza and are associated with coniferous trees (Agerer et al 1996, Ryvarden and Gilbertson 1993).

Bruns et al (1998) found that Albatrellus syringae (Parmasto) Pouzar and A. peckianus were placed outside their Russula-containing clade. A paraphyletic origin for Albatrellus later was confirmed by Binder and Hibbett (2002) and Hibbett and Binder (2002).

The position of /albatrellus in the russuloid lineage is unclear, but morphological similarities exist with Hericium and Lentinellus. Both Albatrellus and Hericium are monomitic with nodose septate hyphae and have SA- gloeocystidia. Albatrellus have amyloid hyphae in the context of the basidiomata, a character found also in Hericium and Lentinellus. In Albatrellus the presence of clamp connections and amyloidity of spores seem to be inconsistent characters, as A. subrubescens and A. ovinus lack clamps and only A. confluens and A. subrubescens have amyloid spores.

/scytinostromella (71%) – Scytinostromella heterogenea, type of the genus, and Wrightoporia avellanea make up this small clade. They share a dimitic hyphal system with nodose septate generative hyphae and at least partly dextrinoid skeletal hyphae. They also share small, ellipsoid, finely ornamented basidiospores and a gloeoplerous system. However, support for the clade is not impressive and with only two terminal taxa the result must be regarded with caution.

Scytinostromella was introduced as a genus for resupinate species with Gloeocystidiellum-like features but with a dimitic instead of monomitic hyphal system (Parmasto 1968). Only two species originally were included, the second one being Scytinostromella nannfeldtii (erroneously named subasperisporum by Parmasto). In our tree, S. nannfeldtii is placed on a single branch with no connection to the type species; its generic position remains unclear. We have sequenced Scytinostromella cerina, and this species also occurs on a separate single branch (data not shown).

/russulales (86%) – This clade includes corticioid wood-decaying species, e.g., Gloeopeniophorella convolvens, Gloeocystidiellum aculeatum and Boidinia propinqua and the whole family Russulaceae, with the ectomycorrhizal genera Russula and Lactarius and related gasteroid and pleurotoid forms. Species in this clade have a monomitic hyphal system, spores are globose to subglobose often with a distinct ornamentation and a strong amyloid reaction, and gloeocystidia are sulfo-positive. All species except Boidinia furfuracea have simple septate hyphae. Spores are uninucleate and the secondary mycelium dikaryotic which indicates a normal nuclear behavior. Presumably most species are heterothallic (Boidin et al 1997a). Gloeopeniophorella Rick differs from the other species in the clade by the presence of incrusted cystidia (metuloids).

Boidinia furfuracea is placed at the first node in /russulales as a sister group to remaining taxa and monophyly of species with simple septate hyphae is supported by a bootstrap value of 76%. Boidinia was segregated from Gloeocystidiellum as a monotypic genus (Hjortstam and Stalpers 1982). The authors stressed basidiome structure and spore morphology as the main diagnostic features. Later authors have pointed to the suburniform shape of basidia as an additional characteristic (Jülich 1982, Hjortstam and Ryvarden 1988, Ginns and Freeman 1994). Several species subsequently have been placed in Boidinia, and three of them are included in our tree. Boidinia granulata belongs in /gloeocystidiellum II, and Boidinia macrospora clusters in /stereales. Only Boidinia propinqua belongs to /russulales, but its relation to Boidinia furfuracea could not be resolved. As noted above, Boidinia propinqua differs from B. furfuracea by simple-septate hyphae. Two other species with simple septate hyphae, viz. Gloeocystidiellum aculeatum and Boidinia sp., also have their place in /russulales. Our results indicate that Boidinia still should be treated as a monotypic genus. However, before a more detailed view of the clade can be worked out we will use the concept adopted by Hjortstam and Ryvarden (1988) and further developed by Wu and Buchanan (1998) and accept also species with simple-septate hyphae in Boidinia.

Gloeopeniophorella was introduced with one species G. rubroflava Rick (Rick 1934) but since its introduction has been largely ignored. Recently Boidin et al (1997) restudied the type specimen and concluded that it shares similarities with Gloeocystidiellum convolvens and G. laxum. They also made the appropriate combinations. In our tree Gloeopeniophorella convolvens, G. laxa and one specimen closely related to G. laxa form a subclade in /russulales with weak bootstrap support (66%). The reintroduction of Gloeopeniophorella is a necessary step in the rearrangement of Gloeocystidiellum. However, its relation to the nonclamped species now placed in Boidinia needs further study.

/stereales (97%) – This clade is strongly supported by bootstrap, and the same result was reported by Wu et al (2001). Species in this clade are saprotrophs causing a white rot on various deciduous and coniferous trees. Many species, e.g., Stereum and Aleurodiscus, decay dead but still attached branches and fruit on trunks or dead parts of living trees. Several species have adaptations for resisting periods of dry weather, e.g., a leathery consistency in Stereum or a catahymenial organization of the basidioma in Aleurodiscus (Boidin et al 1985).

There are some striking similarities with /peniophorales: Many species have smooth spores, many species have simple septate hyphae, all species have a smooth hymenophore, and catahymenial organization of the basidiomes is widespread. The restricted dataset of the clade supported six subclades, /gloeocystidiopsis, /xylobolus, /stereum, /chelidonium, /cerussatus and /megalocystidium (Fig. 4). It should be noted that the NJ tree has a rather different topology within /stereales and both /gloeocystidiopsis and /chelidonium collapses.

/gloeocystidiopsis – A weakly supported clade (57%), consisting of two subclades, each with stronger support. In one subclade, the generic type Gloeocystidiopsis flammea groups with Gloeocystidiellum cryptacanthum (99% bootstrap support). The other clade consists of Gloeocystidiopsis heimii and Conferticium ravum (81% bootstrap support). All species are monomitic with simple septate hyphae. Gloeocystidia are long, tubular and always with a SA+ reaction, although it can be weak in G. flammea (Boidin et al 1997a