Fungal Species from Rhododendron sp.: Discosia rhododendricola sp.nov, Neopestalotiopsis rhododendricola sp.nov and Diaporthe nobilis as a New Host Record.

In the present study, we report two new asexual fungal species (i.e., Discosia rhododendricola, Neopestalotiopsis rhododendricola (Sporocadaceae) and a new host for a previously described species (i.e., Diaporthe nobilis; Diaporthaceae). All species were isolated from Rhododendron spp. in Kunming, Yunnan Province, China. All taxa are described based on morphology, and phylogenetic relationships were inferred using a multigenic approach (LSU, ITS, RPB2, TEF1 and TUB2). The phylogenetic analyses indicated that D. rhododendronicola sp. nov. is phylogenetically related to D. muscicola, and N. rhododendricola sp. nov is related to N. sonnaratae. Diaporthe nobilis is reported herein as a new host record from Rhododendron sp. for China, and its phylogeny is depicted based on ITS, TEF1 and TUB2 sequence data.


Introduction
Rhododendron, a genus of shrub and small to large trees belonging to Ericaceae, is an indicator of health for forest areas [1], commonly found in low-quality acidic soil and sterile conditions. This plant is mainly distributed in India and southeastern Asia, extending from the northwest Himalayas (Arunachal Pradesh) to Bhutan, eastern Tibet, Nepal, north Myanmar, Sikkim, and west central China [2]. Rhododendron flowers are used as food, to produce fermented wine, and to make herbal tea due to their distinctive flavor and color [3,4]. Fungi colonizing Rhododendron include Alternaria alternata, Aspergillus brasiliensis, Chrysomyxa dietelii, C. succinea [5], Diaporthe nobilis [6], Epicoccum nigrum, Mucor hiemalis, Pestalotiopsis sydowiana and Trichoderma koningii [7]. However, given the economic importance of this plant, it is imperative to assess the fungal species associated with it.

Phylogenetic Analyses
The sequence alignment and phylogenetic analyses were performed as outlined by Dissanayake et al. [43] and Chaiwan et al. [42,44,45]. Phylogenetic analyses were performed using a combined Discosia dataset of ITS, LSU, RPB2, TEF1 and TUB2 sequence data and a combined Neopestalotiopsis and Diaporthe dataset of ITS, TEF1 and TUB2 sequence data. Taxa used in the analyses were obtained through recent publications [16,28,46]. The phylogenetic analyses were carried out using maximum parsimony (MP), maximum likelihood (ML) and Bayesian posterior probabilities (BYPP). PAUP v4.0b10 was used to conduct the parsimony analysis to obtain the phylogenetic trees [47]. Trees were inferred using the heuristic search option with 1000 random sequence additions. Maxtrees were set to 1000, branches of zero length were collapsed and all multiple parsimonious trees were saved. Descriptive tree statistics for parsimony-tree length (TL), consistency index (CI), retention index (RI), relative consistency index (RC) and homoplasy index (HI)-were calculated for trees generated following the Kishino-Hasegawa test (KHT) criteria [48], which was performed in order to determine whether trees were significantly different. Maximum-parsimony bootstrap values equal or greater than 60% are given as the second set of numbers above the nodes.
Maximum likelihood analysis was performed by using RAxML-HPC2, New Orleans, LA on XSEDE (8.2.8) [45,[48][49][50]. The search strategy was set to rapid bootstrapping and the analysis was carried out using the GTRGAMMAI model of nucleotide substitution. Maximum likelihood bootstrap values equal to or greater than 60% are given as the first set of numbers above the nodes.
Bayesian inference (BI) analysis was conducted with MrBayes v. 3.1.2 to evaluate the posterior probabilities (BYPP) using Markov chain Monte Carlo sampling [51]. Two parallel runs were conducted using the default settings, but with the following adjustments: six simultaneous Markov chains were run for 2,000,000 generations and trees were sampled every 200 generations. The distribution of log-likelihood scores were examined to determine stationary phase for each search and to decide if extra runs were required to achieve convergence, using the program Tracer 1.4 [52]. The first 10% of generated trees were discarded and the remaining 90% of trees were used to calculate posterior probabilities (PP) of the majority rule consensus tree. The phylogenetic trees were viewed in FigTree v. 1.4 [53] and edited using Microsoft Office Power Point 2007 and Adobe Photoshop CS6 Extended [42].

Genealogical Concordance Phylogenetic Species Recognition (GCPSR) Analysis
The related species were analyzed using the Genealogical Concordance Phylogenetic Species Recognition model. The pairwise homoplasy index (PHI) [54] is a model test based on the fact that multiple gene phylogenies will be concordant between species and discordant due to recombination and mutations within a species. The data were analyzed by the pairwise homoplasy index (PHI) test [54]. The test was performed in SplitsTree4 [55,56] as described by Quaedvlieg [57] to determine the recombination level within phylogenetically closely related species using a five-locus concatenated dataset to determine the recombination level within phylogenetically closely related species. If the PHI is below the 0.05 threshold (Φw < 0.05), it indicates that there is significant recombination in the dataset. This means that related species in a group and recombination level are not different. If the PHI is above the 0.05 threshold (Φw > 0.05), it indicates that it is not significant, which means the related species in a group level are different. The new species and its closely related species were analyzed using this model. The relationships between closely related species were visualized by constructing a split graph, using both the LogDet transformation and splits decomposition options.

Discosia, Habitat and Known Distribution Checklist Associated with Rhododendron sp.
An updated checklist of Discosia based on the SMML database (https://nt.ars-grin. gov/fungaldatabases/) (accessed on 10 June 2022) is provided [58]. Those species for which molecular data are available are indicated. The distribution information regarding the type or original descriptions available and the locality from which Discosia have been recorded on Rhododendron spp. is provided, including all the specimens encountered during this study.

Phylogenetic Analyses
The combined sequence alignments of Discosia comprised 54 taxa (Table 1), with Immersidiscosia eucalypti MFLU16-1372 and NBRC 104195 as the outgroup taxa. The dataset comprised 4364 characters including alignment gaps (LSU, ITS, RPB2, TEF1 and TUB2 sequence data). The MP analysis for the combined dataset had 430 parsimony-informative, 3522 constant, and 412 parsimony-uninformative characters, and yielded a single most parsimonious tree (TL = 1353, CI = 0.777, RI = 0.764, RC = 0.594; HI = 0.223). The RAxML analysis of the combined dataset yielded a best scoring tree with a final ML optimization likelihood value of −22,013.917605. The matrix had 840 distinct alignment patterns, with 66% undetermined characters or gaps. Bayesian posterior probabilities from Bayesian inference analysis were assessed with a final average standard deviation of split frequencies = 0.009983. The phylogenetic tree in this study showed that our strain (Discosia rhododendricola KUN-HKAS 123205 and MFLU20-0486) is related to D. muscicola with high support value in the phylogenbetic tree ( Figure 1). Sequence alignments are deposited in TreeBASE. The combined sequence alignments of Neopestalotiopsis comprised 89 taxa (Table 2), with Monochaetia monochaeta CBS115004 and M. ilexae CBS101009 as the outgroup taxa. The dataset comprised 2634 characters including alignment gaps (ITS, TUB2 and TEF1 sequence data). The MP analysis for the combined dataset had 631 parsimony-informative, 1524 constant, and 479 parsimony-uninformative characters, and yielded a single most parsimonious tree (TL = 2304, CI = 0.679, RI = 0.813, RC = 0.552; HI = 0.321). The RAxML analysis of the combined dataset yielded a best scoring tree with a final ML optimization likelihood value of −24,500.881631. The matrix had 1268 distinct alignment patterns, with 35.77% undetermined characters or gaps. Bayesian posterior probabilities from Bayesian inference analysis were assessed with a standard deviation of split frequencies = 0.024223. The phylogenetic tree in this study showed that N. rhododendricola KUN-HKAS 123204 and MFLU20-0046 belonged to a separate clade, phylogenetically related to N. sonneratae, N. coffeae-arabicae and N. thailandica with 88% MP support ( Figure 2). Sequence alignments are deposited in TreeBASE.
Culture characteristics: Colonies grown on PDA were filamentous, raised, filiform margin, reached 4-5 cm in 5 days at 25 • C, brown to black, mycelium superficial, branched, septate, white mycelium with aerial on the surface, and produced black spore mass.
Culture characteristics: Colonies grew on PDA, filamentous, flattened, dense and felty, reaching 5-6 cm in 14 days at 25 • C, white to brown on the surface, mycelium superficial, branched, and septate.
The new taxon, D. rhododendricola, was phylogenetically related to D. muscicola, described by Nicot-Toulouse Morelet (1968), and isolated from Cephalozia bicuspidate (Cephaloziaceae) in France. However, no morphological data are available for comparison [94]. Discosia rhododendricola sp. nov. was isolated from Rhododendron sp. and its morphology was compared. The ascomata and conidia of D. rhododendricola were larger than those of D. artocreas, whereas the sizes of conidiophores, conidiogenous cells and apical appendage were similar.
Discosia rhododendricola is similar to D. macrozamiae (CPC 32109) [62], but the phylogenetic tree showed that our species was closely related to D. muscicola CBS 109.48. However, for D. muscicola CBS 109.48, only rDNA sequence data were available (Figure 1). It should be pointed out that when the ITS DNA sequences of Discosia muscicola were subjected to a blast search, the closest hits were Aspergillus species similar to A. avenaceus. Our novel species have DNA sequence data from three regions (LSU, ITS, and RPB2), but we can only compare the LSU region for D. muscicola CBS 109.48, as there are no sequence data of the protein coding gene available for comparison. Based on the previous study of Wijayawardene et al. [16], 34 genera are recognized in Sporocadaceae. In this study, we introduce Discosia rhododendricola as a new species based on phylogenetic analyses and the pairwise homoplasy index.
Discosia species share similar morphological characters, but most characters are not meaningful in species delineation. In this study, our new species constitutes a different branching pattern in our phylogeny and appears distinct from extant species. A relationship among species based on similar conidial characters does not necessarily correlate with our phylogenetic relationships, and this indicates that morphology has little significance for reliable species identification.
Herein, we introduce a new species, Neopestalotiopsis rhododendricola KUN-HKAS 123204, within the Neopestalotiopsis genus that was separated from the other Neopestalotiopsis clade based on morphological and molecular phylogenetic analyses ( Figure 2). Neopestalotiopsis are characterized by their conidia with versicolor median cells, by indistinct conidiogenous cells [15] and the ITS, TUB2 and TEF1 sequences. The newly described species is phylogenetically related to the group of N. sonneratae, N. coffeae-arabicae and N. thailandica in the phylogenetic tree (Figure 2), and the relationship is not strongly supported. Our new species was found on a Rhododendron sp. plant host from China, while N. sonneratae was reported on leaf spots on Sonneronata alba L. [21] and Neopestalotiopsis thailandica was reported on leaf spots of Rhizophora mucronata Lam. Both strains have been reported in Thailand [21], and N. coffeae-arabicae was found on leaves of Coffea arabica in China [75].
Diaporthe have been reported on Rhododendron spp. from Europe (Latvia) [6]. The strain (KUN-HKAS 123203) was isolated from Asia (China), indicating that the species is distributed in different geographical locations on the host; however, there is a need for more collections of microfungi associated with Rhododendron, targeting a wide variety of geographical locations. A checklist for Discosia species associated with Rhododendron is also provided herein.

Discosia Species Associated with Rhododendron sp.: Habitat, Known Distribution and Checklist
The above information is based on the USDA Systematic Mycology and Microbiology Laboratory (SMML) database [58], relevant literature, data from this study while current names and fungal classifications used are according to Index Fungorum (2022) [14], and an outline of Ascomycota [16]. Species confirmed with DNA sequence data are marked with an asterisk.