Additions to the taxonomy of Lagarobasidium and Xylodon (Hymenochaetales, Basidiomycota)

Abstract Lagarobasidium is a small genus of wood-decaying basidiomycetes in the order Hymenochaetales. Molecular phylogenetic analyses have either supported Lagarobasidium as a distinct taxon or indicated that it should be subsumed under Xylodon, a genus that covers the majority of species formerly placed in Hyphodontia. We used sequences from the ITS and nuclear LSU regions to infer the phylogenetic position of the type species L.detriticum. Analyses confirm Lagarobasidium as a synonym of Xylodon. Molecular and morphological information show that the traditional concept of L.detriticum covers at least two species, Xylodondetriticus from Europe and X.pruinosus with known distribution in Europe and North America. Three species currently placed in Lagarobasidium are transferred to Xylodon, viz. X.magnificus, X.pumilius and X.rickii. Three new Xylodon species are described and illustrated, X.ussuriensis and X.crystalliger from East Asia and X.attenuatus from the Pacific Northwest America. The identity of X.nongravis, described from Sri Lanka, is discussed.

Controversies over the taxonomic position of Peniophora detritica emerged during the last decades. In modern morphology-based systems, it was first attributed to Hyphodontia J. Erikss., mainly due to hyphal characters and the shape of basidia (Eriksson 1958, Langer 1994. A second solution was introduced by Eriksson and Ryvarden (1976) who stressed the shape of cystidia and the thick-walled cyanophilous basidiospores and placed the species in Hypochnicium. The third option and the one chosen by Jülich (1974), was to place P. detritica in a genus of its own (Jülich 1974, 1979, Hjortstam and Ryvarden 2009. Larsson et al. (2006) used the nrLSU and 5.8S genes for a phylogenetic analysis of Hymenochaetales and recovered Peniophora detritica nested in a fairly well-supported clade that also included several species usually classified in Hyphodontia. This result supported the original opinion on relationships introduced by Eriksson (1958) but also showed that Hyphodontia sensu Eriksson was polyphyletic. The clade with Peniophora detritica. recovered by Larsson et al. (2006), was later identified as Xylodon, type species X. quercinus, a genus that now covers the majority of species earlier referred to Hyphodontia (Hjortstam and Ryvarden 2009). On the other hand, Dueñas et al. (2009) studied sequences from the ITS region and concluded that molecular information supported recognition of the separate genus Lagarobasidium. These same ITS sequences have been used by several subsequent researchers, who therefore maintained Lagarobasidium separate from Hyphodontia sensu lato (Yurchenko and Wu 2014, Riebesehl et al. 2015, Chen et al. 2016, Kan et al. 2017, Yurchenko et al. 2017, Chen et al. 2018).
In the present study, we revise the Lagarobasidium detriticum complex based on morphological and molecular methods. We propose to consider Lagarobasidium as a later synonym of Xylodon and to restore Odontia pruinosa as an independent species. In addition, we describe three new Xylodon species and make five new combinations.

Morphological methods
Type material and specimens from herbaria H, S, O, GB, BPI, TAAM and BAFC were studied. Herbarium abbreviations are given according to Index Herbariorum (Thiers).
Microscopic methods are described in Miettinen et al. (2006). All measurements were made in Cotton Blue (CB, Merck 1275) with phase contrast illumination (1250×). The following abbreviations are used in microscopic descriptions: L -mean spore length; Wmean spore width; Q -mean L/W ratio; n -number of spores (hyphae, basidia) measured per number of specimens. We excluded 5% of measurements from each end of the range representing variation of basidiospores and cystidia. Excluded extreme values are given in parentheses when they differ substantially from the lower or higher 95% percentile.

Phylogenetic analyses
DNA sequences were edited in Geneious (Biomatters Ltd, Auckland, New Zealand) or in Sequencher 5.2.4 (Gene Codes Co., Ann Arbor, MI, USA) and deposited in Gen-Bank (Table 1). We compiled two sequence datasets. The first one contains full ITS sequences from 83 specimens. The second dataset includes ITS and nLSU sequences from 24 specimens and is a subset of the taxa in the ITS-only dataset. In both datasets, Hastodontia hastata (Litsch.) Hjortstam & Ryvarden (Hymenochaetales) was included as outgroup (Larsson et al. 2006). We generated 13 ITS and 6 nLSU sequences for this study; other sequences used in the analyses were downloaded from GenBank (Benson et al. 2018) or UNITE (Kõljalg et al. 2013) (Table 1). Alignments were calculated through MAFFT 7.407 online server (https://mafft.cbrc.jp/alignment/server/) using the L-INS-I strategy (Katoh et al. 2017) and then manually adjusted. The alignments are deposited in TreeBASE (http://purl.org/phylo/treebase/phylows/study/TB2:S23057).
We inferred phylogenetic trees with maximum likelihood (ML), maximum parsimony (MP) and Bayesian Inference (BI) but provide only the last one since all trees show congruity of the phylogenetic signal. Substitution models were determined with the aid of TOPALi 2.5 (Milne et al. 2008) based on Bayesian information criterion  (BIC). GTR + G (nst = 6, rates = gamma) were the best-fit models for the whole ITS region in the ITS dataset as well as in the ITS + nrLSU dataset. SYM + G (nst = 6, rates = gamma, statefreqpr = fixed(equal)) was the best-fit model for the nrLSU region in the ITS + nrLSU dataset. The suggested models were implemented in the Bayesian phylogenetic analyses. We performed Bayesian inference with MrBayes 3.2 (Ronquist et al. 2012). In the analyses, three parallel runs with four chains each, temp = 0.2, were run for 3 million generations. All chains converged to <0.01 average standard deviation of split frequencies. A burn-in of 25% was used in the final analyses. Maximum-likelihood (ML) analysis was performed in RAxML 7.2.8 (Stamatakis 2006) implemented in Geneious. Following models suggested by TOPALi 2.5, we preferred to use the GTR model with gamma correction (GTRGAMMA) in ML analysis for both datasets. The bootstrapping was performed using the 'Rapid bootstrapping' algorithm with the number of bootstrap replicates set as 1000.
Maximum parsimony (MP) analysis was performed using MEGA 7 (Kumar et al. 2016). We used the Subtree-Pruning-Regrafting (SPR) algorithm using all sites. The number of bootstrap replicates was set as 1000.

Results
For both datasets, the Bayesian inference returned trees with two main clades (Figures 1,  2); the largest clade is well-supported and corresponds to Xylodon (pp 1.0), while the other clade is unsupported and includes Lyomyces, the Hyphodontia crustosa group, H. pruni and Rogersella griseliniae (pp 0.89). Basal relationships within Xylodon are not resolved. Peniophora detritica and its allied species are nested within Xylodon and form a well-supported subclade together with X. borealis and X. brevisetus (Figures 1, 2). Maximum likelihood and maximum parsimony returned similar topologies and relevant support values from these analyses are indicated on nodes in Figures 1, 2.
In the ITS-only tree, three terminal branches represent new species that are described below. Xylodon attenuatus occurs as a sister taxon to X. rimosissimus; X. crystalliger forms a subclade with X. astrocystidiatus, X. paradoxus and X. heterocystidiatus; and X. ussuriensis is the sister taxon to X. detriticus and X. pruinosus (Figure 1).
The results allow us to introduce new species and new combinations as follows.
Distribution and ecology. North-western USA (Washington), on angiosperm and gymnosperm wood (fallen decorticated logs).
Remarks. Eriksson and Ryvarden (1976) selected Bourdot 7226 (in herb. S) as lectotype. They also treated Hyphodontia nikolajevae and Odontia pruinosa as synonyms. However, the type specimens of H. nikolajevae and O. pruinosa reveal small differences from the type material and other collections of X. detriticus studied by us. The main  features of X. detriticus versus the two other taxa are narrower basidiospores (must be observed in cotton blue) and longer, narrower cystidia having no distinct intercalary inflation (Tables 2, 3, Figures 5, 6). Eriksson and Ryvarden (1976) attributed the differences in cystidia morphology between Bourdot's specimen and types of H. nikolajevae and O. pruinosa to different stages of basidiocarp development. Our investigation indicates that the differences are genetic and species specific. Differences in basidiospore size and shape are detectable in CB but not in KOH, which could explain why they have gone unnoticed in earlier studies. Hjortstam and Ryvarden (2009) added Hyphodontia magnacystidiata to the synonymy of X. detriticus. This species is, as far as we know, only known from the type, collected on dead wood of Populus in New York, USA (Lindsey and Gilbertson 1977). It has an odontioid basidiocarp and its cystidia are similar to those of X. pruinosus (Table 3, Figures 6, 8). On the other hand, the basidiospore size is very close to X. detriticus (Table 2). In the absence of sequenced material, it is not possible to decide whether this is an independent species or not. Considering that the single specimen was growing on wood and that X. detriticus is not yet found in North America, we prefer to keep H. magnacystidiata as a synonym of X. pruinosus (see below).
Remarks. The type specimen of Hyphodontia nikolajevae Parmasto reveals no essential differences from the type and other collections of X. pruinosus studied by us. On average, Xylodon pruinosus has wider basidiospores than X. detriticus (Table 2). For a detailed description and illustration, see Greslebin and Rajchenberg (2000). The presence of both hymenial, capitate cystidia and enclosed, tubular to moniliform cystidia with homogenous contents strongly stained by cotton blue, make this species morphologically reminiscent of Xylodon brevisetus and X. tuberculatus. X. pumilius differs from both by a smooth hymenium and thick-walled basidiospores. For a description, see Hjortstam and Ryvarden (1982). Gorjón (2012) could not verify the presence of large capitate cystidia, similar to those present in X. magnifica and included in the original description by Hjortstam and Ryvarden (1982). We restudied the isotype in herbarium O and can confirm that these large cystidia do exist, which supports a possible position of this species close to X. detriticus and X. pruinosus.  Etymology. Ussuriensis (lat., adj.) -from the river Ussuri in Russian Far East and adjacent China.
Distribution and ecology. East Asia (Russian Far East -Primorie), on decayed angiosperm wood; seemingly not rare in secondary oak-dominated forest.
Remarks. The distinctly thick-walled tubular cystidia of X. ussuriensis make it different from other Lagarobasidium-like species treated here. Subhymenial astrocystidia found in X. ussuriensis are also present in some specimens of X. detriticus although they are apparently rare in the latter species.

Discussion
Our study confirms the results from Larsson et al. (2006) and Larsson (2007) that Peniophora detritica clusters with Xylodon quercinus, the type species of Xylodon. Here we also show that Peniophora pruinosa, the type of Lagarobasidium, belongs in Xylodon and is a sister species to X. detriticus. This contradicts the results published by Dueñas et al. (2009) who came to the conclusion that Lagarobasidium was a genus separate from Hyphodontia sensu lato. As support for that result, they published ITS sequences of L. detriticum and the new species L. calongei (GenBank FM876211 and FM876212,respectively). However, at least the sequence of L. detriticum (FM876211) seems to be based on a misidentification or contamination during the laboratory process. This sequence is 100% identical to several sequences of Hyphoderma roseocremeum, a species belonging in Polyporales (e.g. UNITE database UDB031922).
Blasting FM876212 against public sequence databases does not return any reliable results, which, if the sequence is correct, suggests that the species does not belong in Xylodon. Remaining species referred to Lagarobasidium and not already discussed include L. cymosum (D.P. Rogers & H.S. Jacks.) Jülich and L. subdetriticum (S.S. Rattan) J. Kaur & Dhingra. The former has been placed in Hypochnicium because of the thick-walled basidiospores but numerous subulate cystidia makes it a deviating element in that genus. Only access to sequence information can disclose its relationships. Lagarobasidium subdetriticum was originally described in Hyphodontia and should be retained in that genus also when the genus is taken in a restricted sense (Hjortstam and Ryvarden 2009).
For the phylogenetic analyses of Hyphodontia sensu lato, only nuclear ribosomal genes have so far been applied. All published results confirm that Hyphodontia sensu lato is polyphyletic and that most species can be referred to one of three clusters, viz Hyphodontia sensu stricto, the Kneiffiella cluster and the Xylodon cluster (including Lyomyces). Within these clusters the relationships are not well resolved when the ribosomal genes are the sole source for genetic information. On such detailed level, analyses become highly sensitive to sampling and outgroup choice. It is clear that both a wider sampling and more markers must be included in analyses in order to establish a stable genus level classification for all species that have been referred to Hyphodontia in a wide sense.