Taxonomy and phylogeny of Dichostereum (Russulales), with descriptions of three new species from southern China

Abstract Nine species of Dichostereum were subjected to phylogenetic analyses, based on a combined dataset of ITS1-5.8S-ITS2-nrLSU-tef1 sequences. The morphology of specimens collected from China and Australia were studied. Three species, D.austrosinense, D.boidinii and D.eburneum, collected from southern China, are described and illustrated as new to science, based on the morphological and molecular evidence. Dichostereumaustrosinense is characterised by the relatively large gloeocystidia (80–130 × 8–15 µm) and basidiospores (7.3–8 µm in diam.) with large warts and crests. Dichostereumboidinii is distinguished by its thick basidiomata and relatively small basidiospores (5.5–6.5 µm in diam.) with large warts and crests. Dichostereumeburneum is unique in having pale basidiomata growing on bark of living Castanopsis, abundant crystals in the context and basidiospores with dense and large ornamentations. A key to the 5 species of Dichostereum in China is given.

Dichostereum was once treated as a subgenus of Vararia P. Karst. (Peniophoraceae, Russulales) by some mycologists since the two genera are very similar in morphology except that the latter has smooth basidiospores (Boidin 1967;Parmasto 1970;Lanquetin 1973). Boidin and Lanquetin (1977) emended the description of Dichostereum and retained it as a separate genus. Later, Boidin and Lanquetin (1980) monographed the genus and provided a key to its 11 species based on evidence of morphology, distribution and intercompatibility tests of cultures. Based on limited sampling, their results showed that D. effuscatum (Cooke & Ellis) Boidin & Lanq. and D. granulosum (Pers.) Boidin & Lanq. were widely distributed, while the other species seemed to be rather endemic (Boidin and Lanquetin 1980). Few studies on the genus have been carried out since then and many regions including East Asia need further collecting and study (Boidin et al. 1987, Boidin andMichel 1998).
Previously, two species, Dichostereum boreale (Pouzar) Ginns & M.N.L. Lefebvre (= D. granulosum) and D. pallescens (Schwein.) Boidin & Lanq., were reported in temperate China (Dai 2011). The species diversity of the genus in China is still unclear. In the present paper, we provide a morphological and phylogenetic study of the genus based on specimens mostly collected from southern China. This is part of an ongoing study of the corticioid fungi of the Russulales in China.

Morphological studies
Voucher specimens were deposited in the herbaria of Beijing Forestry University, Beijing, China (BJFC) and in the Centre for Forest Mycology Research, U.S. Forest Service, Madison, USA (CFMR). Freehand sections were made from dried basidiomata and mounted in 2% (w/v) potassium hydroxide (KOH), 1% phloxine (w/v) or Melzer's reagent. Microscopic examinations were carried out with a Nikon Eclipse 80i microscope (Nikon Corporation, Japan) at magnifications up to 1000×. Drawings were made with the aid of a drawing tube. All measurements were carried out with sections mounted in Melzer's reagent. Ornamentations were excluded from the measurements of basidiospores. Scanning electron micrographs (SEM) were taken with a JEOL JSM-6700F microscope (JEOL, Japan). Dried specimens were mounted directly in gold and platinum and examined and photographed at 10.0 kV. Colour names and codes follow Kornerup and Wanscher (1978). Herbarium code designations are from Index Herbariorum (Thiers, continuously updated).
For both Maximum Likelihood (ML) and Bayesian Inference (BI), a partitioned analysis was performed with the following five partitions: ITS1, 5.8S, ITS2, nrLSU and tef1. The ML analysis was performed using RAxML v.8.2.10 (Stamatakis 2014) with the bootstrap values (ML-BS) obtained from 1,000 replicates and the GTRGAMMA model of nucleotide evolution. The maximum parsimony (MP) analysis was performed using PAUP* 4.0a162 (Swofford 2003). Trees were generated using 100 replicates of random stepwise addition of sequence and tree-bisection reconnection (TBR) branch-swapping algorithm with all characters given equal weight. Branch supports (MP-BS) for all parsimony analyses were estimated by performing 1,000 bootstrap replicates with a heuristic search of 10 random-addition replicates for each bootstrap replicate. The BI was performed using MrBayes 3.2.6 (Ronquist et al. 2012). The best-fit substitution model for each partitioned locus was estimated separately with jModeltest v.2.17 (Darriba et al. 2012). Four Markov chains were run for 6,000,000 generations until the split deviation frequency value was lower than 0.01. The convergence of the runs was checked using Tracer v.1.7 (Rambaut et al. 2018). Trees were sampled every 100 th generation. The first quarter of the trees, which represented the burn-in phase of the analyses, was discarded and the remaining trees were used to calculate Bayesian posterior probabilities (BPP) in the majority rule consensus tree. All trees were visualised in FigTree 1.4.2 (Rambaut 2014).

Phylogenetic inference
The ITS-nrLSU-tef1 sequence dataset contained 37 ITS, 38 nrLSU and 18 tef1 sequences from 40 samples representing 26 ingroup taxa and the outgroup (Table 1). Twenty ITS, 18 nrLSU and 18 tef1 sequences were generated for this study. The dataset had an aligned length of 2239 characters, of which 1596 were constant, 163 variable characters were parsimony-uninformative and 480 were parsimony-informative. MP analysis yielded six most parsimonious trees. jModelTest suggested TIM2ef+G, K80+G, TPM1uf+G, TIM2+I+G and TrN+I+G to be the best-fit models of nucleotide evolution for ITS1, 5.8S, ITS2, nrLSU and tef1 markers, respectively, for the Bayesian analysis. The average standard deviation of split frequencies of BI was 0.004704 at the end of the run. MP and BI analyses resulted in an almost identical tree topology compared to the ML analysis. Only the ML tree is shown in Fig. 1 with ML and MP bootstrap values ≥50% and Bayesian posterior probabilities ≥0.95 labelled along the branches.
Remarks. Dichostereum eburneum is characterised by the pale basidiomata on bark of living tree, the presence of abundant crystals in context and basidiospores with dense and large ornamentations. Ecologically and macroscopically, D. eburneum resembles Dendrothele Höhn. & Litsch., but the microscopic features are largely different (Nakasone and Burdsall 2011). Dichostereum kenyense Boidin & Lanq. is similar to D. eburneum by sharing the large ornamentations of basidiospores, but differs in having wider span of dichohyphae, slightly larger basidiospores (7-8 µm) and a distribution in Africa (Boidin and Lanquetin 1980).

Discussion
To date, 14 species of Dichostereum have been described worldwide including the three new species in the present paper (Boidin and Lanquetin 1980). Amongst them, 5 species, D. brevisporum (S.S. Rattan) Boidin & Lanq. from India, D. kenyense, D. orientale and D. ramulosum (Boidin & Lanq.) Boidin & Lanq. from Africa and D. peniophoroides from Caribbean regions, were not included in the present analyses. In order to resolve the infra-generic phylogenetic relationships of Dichostereum, samples of these species and any additional undescribed taxa should be included. The family Peniophoraceae sensu Larsson (2007) formed a strongly supported clade in Russulales and included about 15 genera (Larsson and Larsson 2003;Miller et al. 2006;Larsson 2007;Leal-Dutra et al. 2018). Except for the coralloid Lachnocladium Lév. and the insect symbiont Entomocorticium H.S. Whitney, Bandoni & Oberw., all the other genera in the family are corticioid fungi, such as Asterostroma Massee, Peniophora Cooke, Scytinostroma Donk and Vararia. However, recent molecular and morphological studies showed that two species of Parapterulicium Corner with coralloid basidiomata belong to Peniophoraceae in the Russulales rather than Pterulaceae of the Agaricales (Leal-Dutra et al. 2018). In the phylogenetic tree, the type species, Parapterulicium subarbusculum Corner formed a distinct lineage, while P. octopodites Corner is closely related to Scytinostroma galactinum (Fr.) Donk and its relatives. More studies on the taxonomy and phylogeny of Peniophoraceae are needed, since some large genera such as Scytinostroma and Vararia are still polyphyletic and many species are undescribed.