Multi-gene phylogenetic evidence suggests Dictyoarthrinium belongs in Didymosphaeriaceae (Pleosporales, Dothideomycetes) and Dictyoarthrinium musae sp. nov. on Musa from Thailand

Abstract Dead leaves of Musa sp. (banana) were collected in northern Thailand during an investigation of saprobic fungi. Preliminary morphological observations revealed that three specimens belong to Dictyoarthrinium. Phylogenetic analyses of combined SSU, LSU, ITS and tef1-α sequence data revealed that Dictyoarthrinium forms a clade in Didymosphaeriaceae (Massarineae, Pleosporales, Dothideomycetes) sister to Spegazzinia. Based on contrasting morphological features with the extant taxa of Dictyoarthrinium, coupled with the multigene analyses, Dictyoarthrinium musae sp. nov. is introduced herein. Our study provides the first detailed molecular investigation for Dictyoarthrinium and supports its placement in Didymosphaeriaceae (Massarineae, Pleosporales, Dothideomycetes). Previously, Dictyoarthrinium was classified in Apiosporaceae (Xylariales, Sordariomycetes).

Dictyoarthrinium was introduced by Hughes (1952) with D. quadratum as the type species. Dictyoarthrinium africanum was simultaneously introduced. Damon (1953) reexamined the type material, descriptions and illustrations of Tetracoccosporium sacchari (Johnston and Stevenson 1917) and mentioned that T. sacchari was congeneric with Dictyoarthrinium quadratum. Therefore, Damon (1953) combined T. sacchari as Dictyoarthrinium sacchari. Damon (1953) also named D. quadratum as the heterotypic synonym of D. sacchari. Rao and Rao (1964) introduced D. lilliputeum and D. microsporum, while Kobayasi et al. (1971) introduced D. rabaulense as novel taxa to the genus. Somrithipol (2007) introduced D. synnematicum and currently seven epithets of Dictyoarthrinium are listed in Index Fungorum (2020). All Dictyoarthrinium species were introduced, based only on morphological data. Vu et al. (2019) sequenced D. sacchari (CBS 529.73) and submitted LSU data to GenBank as the only valid molecular record for the genus.
Dictyoarthrinium is characterised by basauxic conidiogenous cell development (Hughes 1952;Damon 1953;Matsushima 1971). Basauxic development is demonstrated by conidiogenous cells in which elongation occurs at a basal growing point after formation of a single, terminal blastic conidium at its apex (Cole 1976). Conidiophores of Dictyoarthrinium are minutely verruculose, subhyaline and transversely septate (Ellis 1971). Usually, the septa are dark brown and appear as thick stripes on the conidiophore. Conidiophore mother cells are often hyaline or pale brown and cup-shaped (Hughes 1952) or subspherical (Ellis 1971). The length of conidiophores varies within the genus, but in some species, the dimensions are more or less similar. Conidia of Dictyoarthrinium arise from the conidiophore at terminal or lateral parts. Conidiogenesis is monoblastic or polyblastic and integrated (Ellis 1971). Conidia are simple, solitary, dematiaceous and often four-celled. Some taxa (e.g. D. africanum) have 16-celled conidia (Hughes 1952). The surface of conidia is verruculose and most species have warts on the surface. However, the conidia of D. rabaulense are densely echinulate with long spines (Kobayasi 1971). The conidia vary in shape from square to spherical, subspherical or oblong. Most conidia appear flattened on one side. As a specific feature, only D. synnematicum possesses synnemata with filaments (Somrithipol 2007). Stroma, setae and hyphopodia have not been observed in Dictyoarthrinium.
Many Dictyoarthrinium species are saprobes that colonise dead plant materials, although D. rabaulense was recorded even from soil and air (Kobayasi et al. 1971;Ellis 1976). Most Dictyoarthrinium species occur on monocotyledonous plants. The genus is widely distributed across the tropics, mainly in terrestrial environments (Ellis 1971;1976). The sexual morph of Dictyoarthrinium is unknown. Hosts, substrates and geographical distributions of extant Dictyoarthrinium species are listed in Table 1.
A study was undertaken to determine the saprobic fungi associated with Musa sp. (banana) in Thailand, during the dry season. Three hyphomycetous taxa that morpho-

Sample collection, morphological studies and isolation
Dead leaves of Musa sp. were collected from Thailand during the dry season (December to August) of 2018 and 2019. Specimens were transferred to the laboratory in cardboard boxes. Samples were examined with a Motic SMZ 168 Series microscope. Powder-like masses of fungal conidia were mounted in water for microscopic studies and photomicrography. The specimens were examined using a Nikon ECLIPSE 80i compound microscope and photographed with a Canon 550D digital camera fitted to the microscope. Measurements were made with the Tarosoft (R) Image Frame Work programme and images used for figures were processed with Adobe Photoshop CS3 Extended v. 10.0 software (Adobe Systems, USA). Single spore isolation was carried out following the method described in Chomnunti et al. (2014). Germinated spores were individually transferred to potato dextrose agar (PDA) plates and incubated at 25 °C in daylight. Colony characteristics were observed and measured after 3 weeks at 25 °C. Herbarium specimens were deposited in the Mae Fah Luang University (MFLU) Herbarium, Chiang Rai, Thailand. Living cultures were deposited in the Culture Collection of Mae Fah Luang University (MFLUCC). Faces of fungi numbers (Jayasiri et al. 2015) and MycoBank numbers (http://www.MycoBank. org) were obtained for the respective taxa.
Polymerase chain reactions (PCR) were conducted according to the following protocol. The total volume of the PCR reaction was 25 μl and consisted of 12.5 μl of 2 × Power Taq PCR MasterMix (a premix and ready to use solution, including 0.1 Units/ μlTaq DNA Polymerase, 500 μm dNTP Mixture each (dATP, dCTP, dGTP, dTTP), 20 mM Tris-HCl pH 8.3, 100 mMKCl, 3 mM MgCl 2 , stabiliser and enhancer), 1 μl of each primer (10 pM), 2 μl genomic DNA extract and 8.5 μl double distilled water (ddH 2 O). The reaction was conducted by running for 40 cycles. The annealing temperature was 56 °C for ITS and LSU, 57.2 °C for tef1-α and 55 °C for SSU and initially 95 °C for 3 min, denaturation at 95 °C for 30 seconds, annealing for 1 min, elongation at 72 °C for 30 seconds and final extension at 72 °C for 10 min for all gene regions. PCR amplification was confirmed on 1% agarose electrophoresis gels stained with ethidium bromide. The amplified PCR fragments were sent to a commercial sequencing provider (TsingKe Biological Technology Co., Beijing, China). The nucleotide sequence data acquired were deposited in GenBank.

Sequence alignment
Sequences obtained in this study were subjected to BLAST search in GenBank (https://blast.ncbi.nlm.nih.gov/Blast.cgi). BLAST search results and initial morphological studies supported that our isolates belong to Didymosphaeriaceae. Other sequences used in the analyses were obtained from GenBank based on recently published papers (Tanaka et al. 2015;Jayasiri et al. 2019) (Table 2) and BLAST search results. The single gene alignments were done by MAFFT v. 7.036 (http://mafft.cbrc.jp/alignment/ server/large.html; Katoh et al. 2019) using the default settings and later refined, where necessary, using BioEdit v. 7.0.5.2 (Hall 1999). Table 2. Selected taxa with their corresponding GenBank accession numbers in the family Didymosphaeriaceae that are used in the phylogenetic analyses. Type strains are indicated as superscript T and newlygenerated strains are indicated in bold.

Phylogenetic analyses
Maximum Likelihood (ML) trees were generated using the RAxML-HPC2 on XSEDE (8.2.8) (Stamatakis et al. 2008;Stamatakis 2014) in the CIPRES Science Gateway platform (Miller et al. 2010) using GTR+I+G model of evolution. Bootstrap supports were obtained by running 1000 pseudo-replicates. Maximum Likelihood bootstrap values (ML) ≥ 60% are given above each node of the phylogenetic tree in blue (Fig. 1). Bayesian analysis was conducted with MrBayes v. 3.1.2 (Huelsenbeck and Ronquist 2001) to evaluate posterior probabilities (PP) (Rannala and Yang 1996;Zhaxybayeva and Gogarten 2002) by Markov Chain Monte Carlo sampling (BMCMC). Two parallel runs were conducted, using the default settings, but with the following adjustments: four simultaneous Markov chains were run for 2,000,000 generations, trees were sampled every 100 th generation and 20,001 trees were obtained. The first 4,000 trees, representing the burn-in phase of the analyses, were discarded. The remaining 16,001 trees were used for calculating PP in the majority rule consensus tree. Branches with Bayesian posterior probabilities (BYPP) ≥ 0.95 are indicated above each node of the phylogenetic tree (Fig. 1). Phylogenetic trees were visualised with the FigTree v1.4.0 programme (Rambaut 2011).
Culture characteristics. Conidia germinating on PDA within 18 hrs. Colonies on PDA reaching a diameter of 55 mm after 14 days at 25 °C, raised, moderately dense, entire margined, brownish-grey at maturity; reverse white to greyish-white. Notes. Based on BLAST search results of SSU, LSU, ITS and tef1-α sequence data, our strain (MFLUCC 20-0107) showed high similarity to the taxa in GenBank as follows (SSU = 99.26% to Paraconiothyrium brasiliense (isolate GF1), LSU = 96.14% to Alloconiothyrium aptrooti (CBS 981.95), ITS = 93.00% to Kalmusia italica . In the multigene phylogeny, MFLUCC 20-0107 groups with Dictyoarthrinium sacchari, sister to D. musae with strong statistical support (ML = 100%, BYPP = 1.00). Our strain shares similar morphological features with D. sacchari (Subramanium 1952;Ellis 1971) and did not differ significantly. There are slight differences in conidial dimensions and the length of conidiophores of our collection and other D. sacchari collections by previous studies. Conidial dimensions and the length of conidiophores may differ due to diverse environmental effects and host associations. LSU sequence data of D. sacchari (CBS 529.73) are identical with our strain (MFLUCC 20-0107). Unfortunately, ITS, SSU and tef1-α sequence data of CBS 529.73 are not Conidiophores and conidia (e, with distinct mother cell) g, h mature conidiophores with four-celled terminal conidium i conidiophore with two celled terminal conidium j developmental stages of conidia on conidiophore k colony on PDA after 21 days l-q conidia. Scale bars: a = 1000 μm (a); 20 μm (b, j); 50 μm (c-i); 5 μm (l-q).
available in GenBank to compare with our strain. LSU data of Dictyoarthrinium musae have 2.24% of base pair difference with D. sacchari . Dictyoarthrinium sacchari was reported on Musa sp. from Thailand in Lumyong et al. (2003) without morpho-molecular justifications. In this study, we document D. sacchari with detailed morphological illustrations, description, herbarium material and a living culture coupled with DNA sequence data (SSU, LSU, ITS) for a better taxonomic resolution.  650 (f, g, h, j, k). Redrawn from Rao and Rao (1964), Ellis (1971), Kobayasi et al. (1971) and Somrithipol (2007).

Discussion
Both Dictyoarthrinium and Spegazzinia are characterised by basauxic conidiophores (Hughes 1952;Ellis 1971;Tanaka et al. 2015). Spegazzinia often has stellate (α) and disc-shaped (β) conidia (Ellis 1971;Tanaka et al. 2015). The conidia of Dictyoarthrinium (except D. africanum) share some similar characteristics with disc-shaped, β conidia of Spegazzinia. Both conidia are brown, 4-celled and constricted at the septa. Conidia of Dictyoarthrinium have characteristic hyaline or brown warts. Rarely, some taxa of Spegazzinia, for example, S. deightonii, also bear blunt ended spines. Most disc-shaped conidia of Spegazzinia are not warted. In addition, stellate conidia of Spegazzinia are always 4-5-celled and spinulose (Ellis 1971;Tanaka et al. 2015). There are contrasting morphological features of the basauxic conidiophores of both genera. The conidiophores of Dictyoarthrinium are hyaline to subhyaline with septa that appear as dark brown or light brown stripes throughout the conidiophore. The conidiophores (in stellate conidia) of Spegazzinia are more elongated, narrow, aseptate and dematiaceous.
Dictyoarthrinium quadratum (type of Dictyoarthrinium) is the heterotypic synonym of D. sacchari. Dictyoarthrinium quadratum has a terminal mature conidium with one to two cells. As described in Hughes (1952), these 2-celled conidia remain on the conidiophore, even when other conidia fall off. This feature is absent in D. musae. The terminal conidium of D. musae always ends up with four cells. The conidia of D. quadratum are obliquely upwardly directed, whereas the conidia of D. musae are obliquely downwardly directed (Fig. 2). The conidiophores of D. quadratum are erect and straight while D. musae has more curved conidiophores.
Dictyoarthrinium africanum differs significantly from D. musae by having 16-celled conidia. The conidia of D. rabaulense are completely black and densely echinulate with spines sometimes up to 4 μm long (Ellis 1976). However, D. musae has brown warts on the surface of conidia, while D. lilliputeum has hyaline warts. Dictyoarthrinium microsporum has longer conidiophores (250 μm) than D. musae. Morphological features of Dictyoarthrinium species are illustrated in Fig. 4. A key to the species of Dictyoarthrinium is provided below.
Key to the species of Dictyoarthrinium To date, the taxonomy and phylogeny of most genera that have basauxic conidiogenesis (Hughes 1952) have been resolved with their correct taxonomic placements. Dictyoarthrinium and Endocalyx represented the sole unresolved genera. We transferred Dictyoarthrinium to Didymosphaeriaceae based on morphological and molecular evidence. This study uses multigene sequence data of SSU, LSU, ITS and tef1-α for the first time to confirm the taxonomic placement of Dictyoarthrinium in Didymosphaeriaceae.