Digitodesmium polybrachiatum sp. nov., a new species of Dictyosporiaceae from Brazil

Digitodesmium is a genus of saprobic fungi, generally associated with decaying wood in freshwater habitats or in the soil. Fungi in this genus produce cheiroid, euseptate conidia on sporodochia. During an exam of a necrotic robusta coffee stem sent from Nova Venécia, state of Espírito Santo, to the Plant Clinic at the Universidade Federal de Viçosa (Brazil), for disease diagnosis, a fungus, recognized as having the typical features of Digitodesmium, was observed. The fungus was isolated in pure culture and DNA was extracted. Sequences of the partial 18S ribosomal RNA gene, large subunit of the nrDNA, internal transcribed spacer, and translation elongation factor 1-α were generated. The combination of results of the phylogenetic analysis with the exam of the morphology led to the conclusion that the fungus from the dead coffee stems represents a monophyletic distinct lineage within Digitodesmium and an undescribed species for the genus. The concatenate tree also revealed that Digitodesmium is divided in two distinct clades. The novel species can be differentiated morphologically from other species of Digitodesmium by the size of the conidia, the number of arms, and the presence of appendages. The new species Digitodesmium polybrachiatum is hence proposed herein. A comparative table of conidial morphology for the species in the genus is also included.


Introduction
Dictyosporiaceae was introduced by Boonmee (2016) to accommodate a group of fungi belonging to the Dothideomycetes that are saprobes on decaying wood and plant debris in terrestrial and freshwater habitats typically having cheiroid, digitate, palmate, and/or dictyosporous conidia. Dictyosporium, the type genus of the family, has been reported as saprobic on dead or decaying wood worldwide (Goh and Hyde 1996;Ho et al. 2002;Pinnoi et al. 2006;Pinruan et al. 2007). Corda (1836) established the genus with D. elegans as the type species. A phylogenetic analysis based on ITS sequence data has shown that the family Dictyosporiaceae comprises 44 distinct lineages that correspond to ten genera (Boonmee et al. 2016). More recently, three new genera were added to this family (Li et al. 2017;Liu et al. 2017;Iturrieta-González et al. 2018).
Digitodesmium was proposed in 1981 to accommodate the species D. elegans, isolated from rotten wood (Taxus baccata) in the UK (Kirk 1981). After that, six more species were described within this genus, namely, D. recurvum recorded from freshwater habitats in Hong Kong, China (Ho et al. 1999); D. bambusicola on bamboo culms submerged in river from the Philippines (Cai et al. 2002); D. heptasporum found on wood submerged in forest stream, from Yunnan, China (Cai et al. 2003); D. intermedium and D. macrosporum, obtained respectively from plant debris and from a soil sample, both collected in Spain (Silvera-Simón et al. 2010); and D. chiangmaiense isolated from dead wood in Thailand (Hyde et al. 2019).

Isolation
Samples of the stem, taken from diseased robusta coffee plants (Coffea canephora), were collected at a commercial plantation at Nova Venécia (state of Espírito Santo, Brazil). Numerous plants in that plantation were presenting a combination of bark flaking on stems, wilt, and dieback of plants. This disease has been the cause of increasing worries for coffee growers of northern Espírito Santo and southern Bahia. Controversy surrounds the etiology of this disease with suspicions ranging from the Fusarium Wilt reported in Brazil (Belan et al. 2018) to the Coffee Bark Disease and Coffee Wilt Disease reported only on the African continent (Siddiqi and Corbett, 1965;Geiser et al. 2005). An agronomist based at Nova Venécia forwarded us the samples composed of bare-rooted adult plants (part of stems with root system). The stem presented bark flaking. While analyzing the sample in search of the possible causal agent of the disease, it was noticed that in parts of well advanced necrotic tissued colonies of an unusual conidial fungus were present. These appeared to have no relation with the disease, but were examined in detail, nonetheless. A dried culture of the isolate was deposited in the local herbarium/fungarium (Herbarium VIC).
Conidia were transferred to the center of a potato dextrose-agar (PDA) plate supplemented with 0.1 g/L streptomycin sulfate and maintained in a controlled temperature room at 25 °C under a 12-h daily light/12-h dark regime (light provided by two white and one near-UV lamps placed 35 cm above the plates) with a sterile fine pointed needle.
These were spread over the surface of the medium with a sterile loop, and, after a 12-h incubation, individual germinated single conidia were transferred to test tubes containing potato carrot-agar (PCA). Long-term preservation was performed on silica gel and also at -80° C in cryogenic microtubes containing a 10% glycerol solution as described in Dhingra and Sinclair (1995). Two representative cultures were selected and deposited in the local culture collection-Coleção Octávio de Almeida Drumond of the University Federal of Viçosa (COAD).

Morphological characterization
Fungal structures formed on sporulating colonies in vegetable broth-agar (VBA), as described in Pereira et al. (2003), were transferred to slides and mounted in drops of lactoglycerol. Observations of fungal structures were made under an Olympus BX53 light microscope adapted with differential interference contrast lighting and fitted with a digital image capture system (Olympus Q-Color 3 ™). Biometric data was obtained from the measurement of at least 30 representative fungal structures.
Colony description was based on the observation of fungal colonies on malt extract-agar (MEA) and VBA (Pereira et al. 2003), after 40 days under a daily 12 h light regime at 25 °C. Color terminology followed Rayner (1970).

Molecular characterization and multilocus phylogenetic analysis
Genomic DNA was extracted from each of the isolates grown in potato-dextrose (PD)-liquid medium-in the dark for one week. Mycelium of each isolate was dried on sterile filter paper for 2 days and transferred to a sterile plastic tube containing zirconium spheres and placed in a grinder (L-Beader-3, Loccus Biotecnologia). After 20 s of grinding, the resulting suspension was drained into a sterile plastic tube and used for DNA extraction. This was performed with the Wizard Genomic DNA Purification Kit following the manufacturer's protocol.
Our preliminary observations of the fungus morphology suggested that the fungus belonged to Digitodesmium or a related genus. Robust phylogenies have been published for Pleosporales, and particularly for the Dictyosporiaceae demonstrating that rDNA sequences (LSU and SSU) as well as TEF1 and ITS sequences are informative for delimiting families, genera, and species of fungi in that group (Tanaka et al. 2015;Boonmee et al. 2016;Yang et al. 2018). Therefore, target regions of the partial 18S ribosomal RNA gene (SSU), large subunit of the nrDNA (LSU), internal transcribed spacer (ITS), and translation elongation factor 1-α (TEF1) were amplified using fungal specific primers NS1 and NS4 for partial SSU rDNA (White et al. 1990), LROR and LR5 for partial LSU rDNA (Vilgalys and Hester 1990), ITS4 and ITS5 for ITS region (White et al. 1990), and EF1-983 and EF1-2218R for TEF1 region (Rehner 2001). PCR products were analyzed on GelRed ™ (Biotium Inc., Hayward, CA, E.U.A.) and visualized under UV light to verify the size and purity of amplificons. The PCR products were sequenced by Macrogen Inc., South Korea (http:// www. macro gen. com). The nucleotide sequences were edited with software SeqAssem ver. 07/2008 (Hepperle 2004).
The consensus sequences were compared with others deposited in the GenBank database using the MegaB-LAST program. Sequences from GenBank were aligned using MUSCLE (Edgar 2004) and built in MEGA X 10.1 software (Kumar et al. 2018). All of the ambiguously aligned regions within the dataset were excluded from the analyses. Gaps (insertions/deletions) were treated as missing data.
Bayesian inference (BI) analyses employing a Markov Chain Monte Carlo method were performed with all sequences, first with each locus separately and then with the concatenated sequences. The alignments consisted of 22 parsimony-informative positions/1024 bp for SSU, 104/1315 bp for LSU, 252/638 for ITS, and 200/987 bp for TEF1. Before launching the BI, the best nucleotide substitution models were determined for each gene with MrModelTest 2.3 (Posada and Buckley 2004). Once the likelihood scores were calculated, the models were selected according to the Akaike Information Criterion (AIC). The GTR + I + G model of evolution was used for SSU and LSU regions, SYM + I + G was used for ITS, and GTR + G was used for TEF1. One concatenated tree with the four regions was generated with Sequence Matrix (Vaidya et al. 2011) and estimated on the CIPRES web portal using MrBayes on XSEDE 3.2.6 (Miller et al. 2011).
Additionally, a maximum likelihood (ML) tree was generated with the nearest-neighbor-interchange (NNI) ML heuristic method and the Tamura-Nei substitution model as tree inference options, using CIPRES web portal. The chain stabilities of the phylogenetic tree were assessed by using the bootstrap re-sampling strategy with 1000 bootstrap test replicates. The resulting tree topologies using the two methods (ML and BI) were then compared, and the phylogram layout was edited with CorelDRAW Graphics Suite 2017.

Phylogeny
The alignment to construct phylogenetic trees included 63 strains (Table 1) representative of GenBank, representing the family Dictyosporiaceae and two isolates of Periconia igniaria were used as outgroup taxon. The combined matrix consisted of 3964 characters including alignment gaps (SSU, 1024;LSU, 1315;ITS, 638;and TEF1, 987). The trees obtained with ML and BI had an equivalent topology. The phylogenetic analyses inferred from the combined dataset (Fig. 1) indicated that the two strains of the fungus COAD 3174 and COAD 3175 clustered together with 100% (ML) and 1.0 (BI) support. This clade formed a distinct lineage within the genus Digitodesmium, forming a sister clade of the species D. chiangmaiense. The genus Digitodesmium is clearly divided into two distinct lineages highly supported: the first (100% ML and 1.0 BI support) including D. bambusicola CBS 110,279, Digitodesmium sp. TBRC 10,037, and Digitodesmium sp. TBRC 10,038 and the second (99% ML and 1.0 BI support) including D. chiangmaiense KUN-HKAS 102,163 and the two strains obtained in this study.
Notes: The isolates obtained in this study had a distinct morphology from the other species described in Digitodesmium (Table 2). Digitodesmium polybrachiatum differs from D. macrospora, D. intermedium, and D. heptasporum by having smaller and narrower conidia (35-54 × 15-19 μm vs. 130-145 × 19-26 μm; 39-76 × 25-35 μm; and 50-75 × 32.5-70 μm, respectively). Digitodesmium bambusicola, D. chiangmaiense, and D. elegans, despite having conidia with similar dimensions to the newly proposed species, have few arms in their conidia as compared to D. polybrachiatum. In addition, in the phylogenetic tree, the isolates of D. bambusicola and D. chiangmaiense were in separate clades to that of D. polybrachiatum. Other characteristics that also helped distinguishing D. polybrachiatum from other species in the genus are the occasional presence of isolate globoid appendages on its conidia, which are either hyaline and thin-walled or pale brown and thicker-walled, and the absence of conspicuous septal pores. Appendices are only known for D. bambusicola, and inconspicuous septal pores are only found in D. elegans.

Discussion
In the present study, the new species Digitodesmium polybrachiatum was described and recognized as distinct based on the combination of a multilocus phylogenetic analysis using SSU, LSU rDNA, ITS, and TEF1 sequences-which indicated it to belong to a novel monophyletic lineage of the genus and a morphological study that indicated it to be morphologically different from other species in the same genus. However, the combined phylogenetic tree showed that there is a taxonomic inconsistency in this genus. The sequences available for previously described Digitodesmium are grouped into two different highly supported clades. The significant phylogenetic distance between these clades strongly suggests that D. bambusicola belongs to a different genus from D. chiangmaiense and D. polybrachiatum. Nevertheless, in order to fully clarify this situation and elucidate which of these two clades represents Digitodesmium sensu stricto, it is necessary to compare the available sequences with those of the type for the genus-D. elegans IMI 238430e (Kirk, 1981). Unfortunately, there seem to be no pure cultures of this fungus available for study, and there are no sequences of this species available in databases. Therefore, we decided not to propose any nomenclatural changes for Digitodesmium at this stage and to wait for D. elegans to Sequences obtained in this study are highlighted in italics. Ex-type strains are indicated in " T " after the collection number Until the present work, members of Digitodesmium had been reported only from Europe and Asia (Kirk 1981;Ho et al. 1999;Cai et al. 2002Cai et al. , 2003Silvera-Simón et al. 2010;Hyde et al. 2019). This is the first time that a species of Digitodesmium is reported from the Americas.
Author contribution TFN conducted the isolation of strains, DNA extractions, PCR amplifications, phylogenetic analyses and wrote the manuscript. BWF prepared the morphological characterization and participated in writing of the manuscript. RWB is the research leader. He corrected the text and guided throughout the development of the study.
Funding This study received financial support from the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).

Data availability
The datasets generated and analyzed during the current study are available either in GenBank at NCBI (National Center for Biotechnology Information), as indicated in the text, or available from the corresponding author.

Conflict of interest
The authors declare no competing interests.