Insight into the Taxonomic Resolution of Apiospora : Introducing Novel Species and Records from Bamboo in China and Thailand

: Taxonomic studies of bambusicolous fungi in China and Thailand have resulted in the collection of three fascinating saprobic coelomycetes strains. Morphology coupled with combined gene analysis of ITS, LSU, TUB2 , and TEF1-α DNA sequence data showed that they belong to the genus Apiospora , family Apiosporaceae . A new species from Thailand, Apiospora mukdahanensis , and new records of A. locuta-pollinis from China are herein described. In addition, based on both morphological data coupled with phylogenetics and nomenclatural analyses, A. mori is proposed as a new combination. Maximum likelihood, maximum parsimony and Bayesian analyses were performed to clarify the phylogenetic afﬁnities of the species obtained in this study. Newly obtained strains are compared with morphologically-and phylogenetically-related taxa. The comprehensive descriptions, illustrations, and updated phylogeny are provided and discussed for intra-and intergeneric relationships within Apiospora species.

Apiospora is classified by asexual morph characteristics that produce basauxic conidiophores and unicellular globose to obovoid conidia, usually rounded in face view and lenticular in side view, with a longitudinal germ slit [2,7,10,11,32].The sexual morph is characterized as having multi-locular perithecial stromata, clavate to broadly cylindrical asci and hyaline ascospores surrounded by a thick gelatinous sheath [2,7,8,11].Apiospora was previously known as the sexual morph of the genus Arthrinium [33,34].According to the International Code of Nomenclature for Algae, Fungi, and Plants (ICN) [35], Apiospora was a synonym of Arthrinium due to the early introduction of Arthinium and is more commonly used in the literature [3].Crous and Groenewald [3] and Wang et al. [4] provided the upgraded phylogenetic trees of Arthrinium species (=Apiospora) using combined ITS, TEF1-α, and TUB2 sequence data with additional strains (collected from various hosts, substrates, and locations) and indicated that Arthrinium seems to be a species complex which needs further taxonomic revision and epitypification.Multi-gene phylogeny of ITS, LSU, TEF1-α, and TUB2 sequences conducted by Pintos et al. [5] revealed that Arthrinium caricicola, the type species, and other species of Arthrinium mostly found in Carex sp.formed independent lineages unrelated to other species of Arthrinium, and reported that Apiospora occurred on other hosts.However, the taxonomic placement of both genera was uncertain until Pintos and Alvarado [2] resolved this issue and presented Arthrinium and Apiospora as well-supported distinct clades suggesting they are separate genera.
The morphological identification of Apiospora species is challenging because most species share similar morphological characteristics (e.g., conidia).In addition, their morphological features can vary depending on incubation periods and different substrates [3].Thus, morphological characteristics integrated with molecular phylogeny have been widely accepted to distinguish Apiospora species [3][4][5][6][7][8][9]12,17,[36][37][38][39].Presently, 117 epithets are recognized in Apiospora [40], comprising 76 Arthrinium species, which were synonymized under Apiospora [2,12,25].The taxonomic position of other taxa, which lack sequencing information and comprehensive morphological descriptions, remain uncertain and require further study.In this study, we isolated apiospora-like taxa from bamboo in China and Thailand.The morphological characteristics and molecular analyses of ITS, LSU, TUB2, and TEF1-α were applied to determine a new species, Apiospora mukdahanensis, and one new record of A. locuta-pollinis.Furthermore, Arthrinium mori is also transferred to Apiospora as a new combination on the basis of phylogenetic evidence.The host association, geographical distribution, and species diversity of Apiospora are also discussed.

Sample Collection, Fungal Isolation and Morphological Examination
Dead and decaying bamboo specimens were collected during a series of field trips conducted in China and Thailand from the year 2019-2021.The specimens were packed in zip-lock plastic bags prior for further study.Fungal colonies on the host substrate were observed using a stereo microscope (Nikon SMZ800N, Tokyo, Japan).The micromorphological characteristics were documented and photographed with a compound microscope (Nikon Eclipse Ni U, Tokyo, Japan) equipped with Nikon DS-Ri2 camera.The measurements of fungal structures (i.e., conidiomata, conidiophore mother cells, conidiophores, conidiogenous cells, and conidia) were made using the Tarosoft (R) Image Frame Work program.Images used for figures were combined and edited using Adobe Photoshop CS6 Extended version 10.0 software (Adobe Systems, San Jose, CA, USA).Single-spore isolation was conducted to isolate the fungus as detailed in Senanayake et al. [41].The germinating conidia were inoculated on potato dextrose agar (PDA) and incubated at 28 • C for two weeks.Culture characteristics were observed and described after four weeks.Axenic cultures were kept in 2 mL sterilized screw-cap tubes containing PDA for short-term storage and duplicated in 10% glycerol for long-term storage.The type specimen and living

Phylogenetic Analyses
The sequences generated by this study were supplemented with the related taxa resulting from the nucleotide blast search in GenBank (www.ncbi.nlm.nih.gov/blast/,accessed on 1 September 2022) and recent publications [2,12,25,39,48,49] (Table 1).The matrix of consensus sequences was aligned using MAFFT v. 7.475 on the web portal (http: //mafft.cbrc.jp/alignment/server/index.html)[50] and then the ambiguous sites were manually trimmed using BioEdit 7.1.3.0 [51].Phylogenetic trees based on the concatenated ITS, LSU, TUB2, and TEF1-α sequence data (analysis 1) and ITS, LSU, and TEF1-α sequence data (analysis 2) were inferred to clarify the phylogenetic relationships of Apiospora species using maximum likelihood (ML) and Bayesian inference (BI) analyses.In order to clarify the phylogenetic placements of new strains and related strains in A. locuta-pollinis/marii clade, ML, maximum parsimony (MP), and BI were analyzed based on the concatenated ITS, LSU, TUB2, and TEF1-α sequence data (analysis 3).Phylogenetic trees of these combined gene datasets were further compared to check the congruence of the tree topologies.
ML analyses were implemented using RAxML-HPC2 (v.8.2.12) on the CIPRES web portal (http://www.phylo.org/portal2/)[52].The GTRGAMMAI model of nucleotide substitution with 1000 rapid bootstrap replicates was used.BI analyses were performed using MrBayes v.3.2.7a via the CIPRES web portal (http://www.phylo.org/portal2/).The optimal substitution model of nucleotide evolution was determined using MrModeltest v. 2.3 [53].In the first and second analyses, GTR + I + G was chosen as the best-fit model for the ITS, LSU, and TEF1-α datasets, and HKY + I + G for the TUB2 dataset.For the third analysis, the best-fit model for the ITS, LSU, TUB2, and TEF1-α datasets were as follows: SYM + I, GTR, GTR + G, and K80 + I.Ten million Markov chain Monte Carlo sampling (MCMC) generations were run with six simultaneous Markov chains to calculate Bayesian posterior probabilities [54][55][56].Trees were sampled every 100th generation.When the average standard deviation of split frequencies was constantly below 0.01, the runs were automatically stopped and the first 25% of the generated trees were discarded.The remaining trees were used to evaluate the posterior probabilities (PP) of the majority rule consensus tree.MP analyses were conducted with the heuristic search option, as implemented in PAUP v. 4.0b10 [57].Clade stability was determined using a bootstrap analysis with 1000 replicates, random sequence additions with maxtrees set to 1000 [58].The MP tree was described for descriptive tree statistics including Tree Length (TL), Consistency Index (CI), Retention Index (RI), Relative Consistency Index (RC), and Homoplasy Index (HI) under different optimality criteria.Phylogenetic trees were viewed in FigTree v1.4.0 [59] and formatted using Adobe Photoshop CS6 software (Adobe Systems, San Jose, CA, USA).

Host and Geographical Distribution of Apiospora Species
To investigate geographical distribution and host-substratum diversity of the Apiospora species, the data were summarized based on the USDA fungal database (https://nt.ars-grin.gov/fungaldatabases/, accessed on 1 September 2022), academic literature, and this study).
Morphological characteristics of our new strains (KUNCC 22-12408 and KUNCC 22-12409) were compared with the type strain of Apiospora locuta-pollinis (LC 11683) (Table 2).Both new strains are similar to the type strain of A. locuta-pollinis in having globose to subglobose conidia with hyaline equatorial rim, however they have larger conidia (10.5 × 10 and 9.6 × 8 vs. 7.1 × 6.4 µm) (Table 2).The conidiogenous cells of strain KUNCC 22-12409 are more similar to the type strain (LC 11683) in being subglobose to ampulliform to doliiform, but with smaller size (3.2 × 2.2 vs. 4.9 × 3.8 µm) (Table 2)., whereas the strain KUNCC 22-12408 has cylindrical to subcylindrical or ampulliform conidiogenous cells with larger size compared to other strains (6 × 3.5 µm) (Table 2).The morphological description of A. locuta-pollinis was based on cultures and its conidiophores were reduced to conidiogenous cells [61].Thus, the comparisons of conidiomata and conidiophores characteristics between these strains were not possible.In addition, we found some morphological differences between two new strains including strain KUNCC 22-12408 which had longer conidiophore mother cells and conidiogenous cells, but shorter conidiophores compared to strain KUNCC 22-12409 (Table 2).Apiospora locuta-pollinis was previously isolated from hive-stored pollen of Brassica campestris in Hubei province, China [61], whereas the new strains KUNCC 22-12408 were isolated from decaying bamboo and KUNCC 22-12409 was isolated from dead bamboo in Yunnan province, China.Although there were some morphological variations among the new strains and the type strain of A. locuta-pollinis, the multigene phylogeny and DNA sequence comparisons (ITS and TEF1-α gene regions) did not provide the necessary support to delineate them as a distinct species.Therefore, we treated these strains as new records of A. locuta-pollinis.It is possible that the strain KUNCC 22-12408 is a new species due to the significant morphological and phylogenetic differences, which caused it to form a separated branch to other A. locuta-pollinis strains.Further taxonomic studies are needed to resolve their conspecific status.

Host and Geographical Distribution of Apiospora Species
Based on species distribution data, Apiospora is widely distributed in temperate, subtropical, and tropical areas, including Africa, America, Asia, Australia, and Europe (Figure 5).The highest species diversity of Apiospora was found in Asia (e.g., China, India, Japan, Thailand) (Figure 5).However, the data reflect areas in which there have been reports of Apiospora species and may thus reflect hotspots of research, and not just species hotspots.Areas shown to be devoid of Apiospora species may just be areas that require further study.The host-substratum diversity of Apiospora species (Figure 6) showed that most species have been found exclusively associated with Poaceae (63%), including bamboo (31%) and non-bamboo (32%).The most common bamboo genera associated with Apiospora are Bambusa, Phyllostachys, and Arundinaria and the most common non-bamboo genera are Saccharum, Phragmites, and Arundo.
In the Apiospora marii clade, A. hispanica and A. mediterranea are not well-resolved (Figures 1 and 2).Morphologically, A. marii and A. hispanica have overlapping sizes of conidia (7.2-7.5 × 6.1-6.5 vs. 7.5-8.5 × 6.2-7.6 µm) (Table 3), whereas A. mediterranea has larger conidia (9-9.5 × 7.5-9 µm) (Table 3).The base-pair difference of ITS and TUB2 sequence data indicated that they are consistent.However, the LSU sequence data of A. hispanica and A. mediterranea are in short length (320 base pairs) and their TEF1-α sequence data were lacking.The morphological reexamination and molecular data of the type specimens of A. hispanica and A. mediterranea are required to confirm a putative synonymy.
In the Apiospora locuta-pollinis clade, two strains of A. marii CBS 113535 and CBS 114803 clustered together (Figure 2).Pintos et al. [5] and Garin et al. [62] also reported that A. marii CBS 114803 had a well-supported lineage distant from the A. marii clade.Tian et al. [12] synonymized A. pseudomarii GUCC 10228 as A. locuta-pollinis based on morphology and phylogeny.Crous and Groenewald [3] provided the sequence data of these strains in which CBS 113535 was isolated from oats in Sweden and CBS 114803 was isolated from the culm of Arundinaria hindsi in Hongkong.Thus, we treated these strains as A. locuta-pollinis based on the phylogenetic evidence.
TUB2 and TEF1-α gene regions are essential phylogenetic markers for accurate identification of Apiospora species [3][4][5]9,38].Most of the recent studies have used multigene phylogenetic analyses of integrated ITS, LSU, TUB2, and TEF1-α sequence data for Apiospora species delineation [2,5,[10][11][12]24,25,38].However, our phylogenetic analysis based on the integrated ITS, LSU, and TEF1-α sequence data also provided the necessary resolution to distinguish species of Apiospora (Figure S1).In addition, the TEF1-α gene region could support the species delineation between A. marii and A. locuta-pollinis.As they have a 10 bp difference (2.3%) in the TEF1-α gene region, whereas no difference was found in TUB2 gene region.It seems that the TUB2 gene region is uninformative for the separation of species in the A. locuta-pollinis/marii clade.Nevertheless, with the lack of TUB2 sequences in our new strains, the TEF1-α gene region was not enough to resolve their placements within A. locuta-pollinis lineage.We suggest that TUB2 gene and other protein-coding genes, such as RPB2, should be included for further phylogenetic analysis to confirm their actual identity and placements.
Our study also revealed significant morphological variation among Apiospora locutapollinis strains.This result was also observed for other Apiospora species.For instance, A. yunnana, introduced by Dai et al. [8], based on a sexual morph on bamboo culms, had larger conidia in cultures (15.5-26.5 µm diam.), compared to the strains CFCC 52311, CFCC 5231 which were isolated and described directly from bamboo culms (10-16 µm diam.)[9].Apiospora pseudoparenchymatica, introduced by Wang et al. [4], and was isolated from living leaves of bamboo and described by its asexual morph.Feng et al. [11] found a new record of A. pseudoparenchymatica from decaying bamboo culms and noted the significant difference in the characteristics of conidiophores and conidiogenous cells of the new strain, GZCC 20-0117 (on WA) compared to the type specimen, LC 8173 (on PDA and MEA) [4].The significant variation in morphology might be due to the differences in substrates (natural substrates or cultures), growth conditions, hosts, and habitats.This observation makes our finding all the more valuable, we found both new strains from the same host substrate and habitat.The only difference is strain KUNCC 22-12408, which was from the decaying state and strain KUNCC 22-12409 was from the dead state of bamboo.Therefore, our study highlighted the great impact of environmental factors on morphological variation.
Geographically, Asia was found to be home to greatest diversity of Apiospora (Figure 5) and is likely the result of the rich diversity of bamboo genera/species, especially those in China [63].More extensive sampling of these hosts will surely result in the discovery of additional new species.Although the host preference of Apiospora is the family Poaceae, there remains a number of species which have been found on other plant families (Figure 6), such as A. euphorbiae (on Euphorbiaceae, Lauraceae, Pinaceae, Zingiberaceae) [64], and A. jiangxiensis (on Lauraceae, Primulaceae, Theaceae) [4]., On the contrary, some species seem to be host-specific, such as A. pterosperma which has only been found on Cyperaceae (Lepidosperma and Machaerina) [3], and A. rhododendri has only been reported from Ericaceae (Rhododendron) [64].Some species are ubiquitous, with a diverse range of hosts and habitats (e.g., A. arundinis, A. marii, A. rasikravindrae, A. serenensis) [2][3][4][5]8,12,64].It is noteworthy that the asexual morph is more frequently discovered from different ecological habitats, and it is possible they are the cause of certain plant diseases [2,62].For example, A. arundinis which has been isolated from soil, water, and numerous plant hosts [4,64], is also reported as causing some plant diseases, such as brown culm streak of Phyllostachys praecox [65] and the leaf spot of rosemary (Salvia rosmarinus) [66].Apiospora marii has been isolated from air, sand, and different plant hosts [3,5], also reported as causing die-back of olives (Olea europaea) [62].

Conclusions
Based on the rate of discovery, their diverse host ranges and different life-styles, the species number of Apiospora is likely to continue to increase in the future [39,48].There are still a number of regions that have remained unstudied in terms of Apiospora, which will likely further add to the current list of species within this genus.A comprehensive survey of these unknown regions along with a polyphasic taxonomic study of Apiospora is necessary, especially focusing on Poaceae species, will enable a better understanding of host relationships and the ecological significance of this group of fungi.

Figure 1 .
Figure 1.Phylogenetic tree retrieved from RAxML analyses of a combined ITS, LSU, TUB2, and TEF1-α data sequence of Apiospora, and other related taxa in the family Apiosporaceae.Bootstrap support values for ML equal or greater than 60% and Bayesian posterior probabilities greater than 0.90 are indicated at the nodes as ML/PP.The ex-type strains are in bold.The new species are in red and new record and new combination species are in blue.The tree is rooted to Sporocadus trimorphus (CBS 114203).

Figure 2 .
Figure 2. Phylogenetic tree retrieved from RAxML analyses of a combined ITS, LSU, TUB2, and TEF1-α data sequence of taxa in Apiospora locuta-pollinis/marii clade.Bootstrap support values for ML and MP equal or greater than 60% and Bayesian posterior probabilities greater than 0.90 are indicated at the nodes as ML/MP/PP.The ex-type strains are in bold.The new record strains are in blue.The tree is rooted to Apiospora fermenti KUC 2189 and A. pseudospegazzinii CBS 102052.

Table 1 .
List of the taxa used in phylogenetic reconstruction and their corresponding GenBank numbers.