Multigene phylogenetics of Polycephalomyces (Ophiocordycipitaceae, Hypocreales), with two new species from Thailand.

Polycephalomyces (Ophiocordycipitaceae) species are found in subtropical regions and are parasitic or hyperparasitic on insects. Two new species, P. aurantiacus and P. marginaliradians, parasitic on Ophiocordyceps barnesii and larva of Cossidae respectively, are introduced in this paper. Morphological comparison with extant species and DNA based phylogenies from analyses of a multigene dataset support the establishment of the new taxa. Polycephalomyces aurantiacus, exhibiting a hyperparasitic lifestyle on Ophiocordyceps barnesii, differs from other species in producing orange conidia in mass and have longer β-phialides in culture. Polycephalomyces marginaliradians differs from other Ophiocordyceps species by producing single stromata with a stipe, smaller perithecia and branched α-phialides and catenate α-conidia and is parasitic on Cossidae. A combined nrSSU, nrLSU, ITS, tef-1a, rpb1 and rpb2 sequence data was analysed phylogenetically including Ophiocordyceps and Polycephalomyces taxa. The new species described herein are clearly distinct from other species in Polycephalomyces. We provide a key to the species of Polycephalomyces and discuss relevant interspecies relationships.


Results
Molecular phylogeny. Table S1 comprises  The genus Polycephalomyces currently includes 15 species and only 11 species have available DNA sequence data in GenBank (Table S1), excluding the new taxa described in this study. Our multigene phylogenetic analyses herein reveal that our new taxa constitute a strongly supported monophyletic subclade and nested in between other Polycephalomyces species (Fig. 1). In particular it is noted that Polycephalomyces aurantiacus and P. marginaliradians share a close phylogenetic affinity to P. nipponicus and P. kanzashianus (Fig. 1).

Description of
Notes: We collected Ophiocordyceps barnesii in this study which was colonized by an orange hyperparasite which we introduce below as Polycephalomyces aurantiacus. This species may be important in future industrial production of Cordyceps species, which are increasingly being produced because of their medicinal properties and biopesticides potential 25,26 . The specimen was deposited in MFLU Herbarium (MFLU 17-1393).

Discussion
Studies based on morphology and DNA sequence analyses have provided insights into the phylogeny of Polycephalomyces to resolve generic delimitation. Species of this genus are commonly known to exhibit a parasitic mode of life on insects and other fungi 2,5,12 . Our fungal diversity studies on entomophagous fungi have led to the discovery of two species, new to science, which we accommodate in Polycephalomyces. Molecular data also reveals that our new genus belongs to the family Ophiocordycipitaceae as circumscribed by Matočec et al. 12 . Species which exist in their sexual state display morphs such as fertile, capitulate, globose, tuberiform to pulvinate stromata and immersed, elongated pyriform perithecia while the asexual morphs occur as branched or unbranched synnemata, ending up with clavate to spherically flared, hymeniform aggregations of conidiophores, and produce large masses of conidia united in collective globular mucus 12 . To date, six species, including P. ramosus 16,17 , P. sinensis 6,18 , and P. agaricus 9 are considered as parasites of entomogenous fungi, while six species are recorded as entomogenous 6,7,9,11 . Some species such as P. lianzhouensis and P. yunnanensis colonise both entomogenous fungi and insects 7,8 . Because of their economic importance, species of this genus have been the subject for various research. The most recent new species introduced is P. yunnanensis and multigene phylogeny reveals a close relationship to P. formosus, P. ramosopulvinatus and P. sinensis based on 5-loci (nrSSU, nrLSU, tef-1α, rpb1 and rpb2) phylogenetic analyses 8 . Our taxonomic investigations herein reveal two new species of Polycephalomyces, P. aurantiacus and P. marginaliradians. Our morphological examination suggests that both species fit clearly within the generic concept of Polycephalomyces and both species produce two types of conidia. However, they exhibit different mode of life and there are sufficent morphological differences that can justify their segregation into two species. These two new species similar to P. agaricus, P. formosus, P. ponerae, P. sinensis, P. ramosus, and P. yunnanensis have produce two types of conidia, while P. ditmarii, P. lianzhouensis, P. paludosus and P. tomentosus have only one type of conidia. Polycephalomyces aurantiacus and P. marginaliradians have two types of phialides, while P. formosus and P. sinensis have only one type of phialide. Polycephalomyces ponerae also differs from P. aurantiacus and P. marginaliradians by producing Akanthomyces-like β-phialides and parasitic on ant (Ponera Latreille). Polycephalomyces agaricus differs from P. aurantiacus and P. marginaliradians by producting agaric shaped synnemata and parasitic on Ophiocordyceps sp. Polycephalomyces yunnanensis is distinct from P. aurantiacus and P. marginaliradians as it produces longer α-conidia fusiform, catenate or clump together β-conidia and parasitic on O. nutans. Polycephalomyces ramosus differs from P. aurantiacus and P. marginaliradians in having longer synnemata, shorter β-phialides and parasitic on Hirsutella guignardii. Polycephalomyces aurantiacus is distinct from P. marginaliradians as it is parasitic on the fungus Ophiocordyceps barnesii and produces longer synnemata, shorter β-phialides and catenate α-conidia, wheras P. marginaliradians is parasitic on insect and produces shorter synnemata, longer β-phialides and catenate β-conidia (further morphological differences are outlined in Table S2).
Phylogeny based on our concatenated dataset recovered also support that our two new species belong to Polycephalomyces and are distinct from each other (Fig. 1). A close relationship is observed between the two species, but both constitute independent and strongly supported monophyletic subclades indicative of two phylogenetically distinct speces. To further compare our two species, we delved in pairwise nucleotide sequence comparison and noted sufficient differences to justify them as independent taxa 27 . ITS pairwise nucleotide sequence comparison between P. aurantiacus and P. marginaliradians revealed striking differences in 15 base pairs that justifies that both are different from each other and hence can be considered as two distinct species. There are also 6, 21, 4, 15 and 13 differences in the nrSSU, nrLSU, tef-1α, rpb1, and rpb2 DNA sequence data respectively. Two species not considered in our phylogenetic sampling are P. ditmarii and P. paludosus due to the unavailability of sequence data. However these two are different from our new species with respect to one type of conidia occurring on its natural substrate and under cultural conditions 28,29 . The hosts from which our new species have been recovered are also different. Polycephalomyces ponerae was not considered in our phylogenetic sampling as the DNA (ITS) sequence is too short, ambiguous and did not align well with other species. However, Polycephalomyces ponerae is morphologically different from our new species with respect to Akanthomyces-like β-phialides and parasitic on ant (Ponera Latreille). Further morphological differences among species are detailed in Tables S2 and S3. Our multigene phylogeny derived herein also provides robust and well-resolved intergeneric relationships between Polycephalomyces and Ophiocordyceps. Members of both genera are clearly distinct from each other and we managed to successfully identify and sequence Ophiocordyceps barnesii, the host from which P. aurantiacus was isolated. Further interspecies taxonomic relationships are also elucidated in our molecular phylogeny. All Polycephalomyces species currently analysed constitute a strongly supported monophyletic lineage (Fig. 1), which corroborates with previous taxonomic schemes 5,12 . In particular, a robust relationship in observed between P. onorei and P. agaricus sharing P. yunnanensis as sister taxa. These three species are also markedly different in terms of morphological characters. Polycephalomyces yunnanensis is clearly distinct from P. onorei and P. agaricus in terms of being parasitic on Ophiocordyceps nutans (Pat.) G.H. Sung, J.M. Sung, Hywel-Jones & Spatafora, longer synnemata, cylindrical to subulate α-phialides and subglobose or ellipsoidal α-conidia 8,10 . Polycephalomyces ramosopulvinatus is closely related to P. lianzhouensis, but each species is positioned in different well-supported subclades and hence merit species status. Polycephalomyces ramosopulvinatus is also different from P. lianzhouensis with respect to being parasitic on nymph of Cicada and characterised by a long stipe and pseudo-immersed, pyriform perithecia. While phylogeny resolves our new species into well-segregated subclades, we note that relationships of P. formosus, P. tomentosus, P. ramosus and P. sinensis are still obscure and the concatenated dataset used herein did not provide adequate species resolution. A similar phylogenetic scenario is observed for P. nipponicus and P. kazanshianus. Whether these species are conspecific warrants further taxonomic investigations. The latter two species do share some morphological resemblances to P. marginaliradians especially with respect to the yellow cylindrical stipe with capitate lateral fertile part (known from their sexual morph). However, P. marginaliradians differs in having a capitate stromata with stipe, smaller perithecia and parasitic on Cossidae, while P. nipponicus and P. kanzashianus have polycephalous stromata and parasitic on Cicadidae. Meanwhile, P. onorei and P. ramosopulvinatus are distinct from P. marginaliradians by producing bigger perithecia and parasitic on caterpillar (Arctinae) and nymph of Cicada respectively. Phylogenies retrieved herein also support them as separate taxonomic entities.

Materials and Methods
Collection, isolation, and morphology study. Four fresh specimens were collected from southern Thailand (Prachuap Khiri Khan Province), and two from northern Thailand in the soil. The specimens were noted and photographed in the field and transported to the laboratory individually in plastic boxes and stored at 4 °C until examined. Strains were isolated from single spore isolation from both stomata and synnemata following the protocol described in Chomnunti et al. 30 .Cultures were incubated at 18 °C for 14-25 days on potato extract agar (PDA) as outlined by Vijaykrishna et al. 31 . Herbarium material is deposited at MFLU herbarium and HKAS herbarium and Facesoffungi numbers and Index Fungorum numbers are provided as in Jayasiri et al. 32 (Table S4); ITS4/ITS5 for internal transcribed spacer gene region (ITS) 34 , NS1/NS4 for partial small subunit ribosomal RNA gene region (SSU) 34 , LROR/LR5 for partial large subunit rDNA gene region (LSU) 35 . 983 F/2218 R for partial translation elongation factor 1-alpha gene region (TEF-1α) 36 , CRPB1A/RPB1Cr for partial RNA polymerase II largest subunit gene region (RPB1) 37 . RPB2-5F/RPB2-5R for partial RNA polymerase II second largest subunit gene region (RPB2) 37 . PCR amplifications were conducted as outlined by Jeewon et al. 38,39 and PCR products were sequenced by GenScript Biotechnology Co., Nanjing, China.
ML trees were estimated by using the software RAxML 7.2.8 Black Box 42,43 in the CIPRES Science Gateway platform 44 . MrModeltest v.2.3 45 was used to determine the best-fit model of evolution for Bayesian analyses. MrBayes v.3.1.2 46 was used to evaluate posterior probabilities (PP) 47,48 by Markov Chain Monte Carlo sampling (BMCMC). Six simultaneous Markov chains were run for 2,000,000 generations and trees were sampled every 100th generation and 20,000 trees were obtained. The first 20% of trees were discarded, which representing the burn-in phase of the analyses, while the remaining trees were used for calculation posterior probabilities in the majority rule consensus tree (critical values for the topological convergence diagnostic is 0.01). Pylogenetic trees were also constructed based on parsimony analyses as detailed by Cai et al. 49 and Jeewon et al. 50 figured in FigTree v1.4.0 program 52 . Bayesian Posterior Probabilities (BYPP) equal to or great than 0.95 were given 53-63 below each node (Fig. 1).