Acrogenospora (Acrogenosporaceae, Minutisphaerales) Appears to Be a Very Diverse Genus

During a study of diversity and taxonomy of lignicolous freshwater fungi in China, nine species of Acrogenospora were collected. Seven of these were new species and they are described and illustrated. With morphology, additional evidence to support establishment of new species is provided by phylogeny derived from DNA sequence analyses of a combined LSU, SSU, TEF1α, and RPB2 sequence dataset. Acrogenospora subprolata and A. verrucispora were re-collected and sequenced for the first time. The genus Acrogenospora is far more species rich than originally thought, with nine species found in a small area of Yunnan Province, China.


INTRODUCTION
Freshwater fungi are an ecological group that are defined by their presence in freshwater for the whole or part of their life cycle (Thomas, 1996;Wong et al., 1998), and include any species that grow on predominantly aquatic or semi-aquatic substrates (Goh and Hyde, 1996). Freshwater fungi play an important role in nutrient and carbon cycling, biological diversity and ecosystem functioning (Zhang et al., 2008;Swe et al., 2009). There have been many studies of freshwater fungi, especially on diversity, taxonomy and phylogeny (Tsui et al., 2000;Cai et al., 2002;Vijaykrishna et al., 2005;Vijaykrishna and Hyde, 2006;Hirayama et al., 2010;Ferrer et al., 2011;Barbosa et al., 2013;Raja et al., 2013Raja et al., , 2015 and recently from China Yang et al., 2017;Huang et al., 2018a,b;Su et al., 2018;Guo et al., 2019;Luo et al., 2019). In this study, we report nine Acrogenosporaceae species that were collected from freshwater habitats in China. Acrogenosporaceae was established by Jayasiri et al. (2018) to accommodate Acrogenospora within Minutisphaerales, with the latter being a freshwater ascomycetes order, comprising two families, Acrogenosporaceae and Minutisphaeraceae (Wijayawardene et al., 2020). Members of these two families are mostly reported from freshwater habitats Raja et al., 2015;Bao et al., 2019;Hyde et al., 2019).
The sexual morph of Acrogenospora has been linked with Farlowiella. Mason (1941) showed the connection between A. megalospora and Farlowiella armichaeliana based on cultural studies. Ellis (1971) reported the asexual morph of F. carmichaeliana as A. carmichaeliana. Ellis (1972) introduced A. australis as the asexual morph of F. australis based on morphological characters. Goh et al. (1998) accepted these two asexual morphs of Farlowiella and synonymized A. megalospora under F. carmichaeliana and A. altissima under F. australis. Jayasiri et al. (2018) carried out phylogenetic analyses with seven isolates of Acrogenospora and showed that A. sphaerocephala clustered with the sexual morph Farlowiella carmichaeliana. This confirmed the connection between Acrogenospora and Farlowiella. Hyde et al. (2019) also supported the asexual-sexual connection between these two genera based on a phylogenetic study. Based on recent nomenclatural changes with regards to one fungus one name, Acrogenospora was given priority (Wijayawardene et al., 2014;Rossman et al., 2015).
During our investigation of freshwater fungi on submerged wood along a north/south gradient in the Asian/Australasian region , nine isolates of Acrogenospora were collected from freshwater habitats in China. Among them, two are identified as existing species, A. subprolata and A. verrucispora, and another seven are introduced as new species by comparing their morphology with known species of the genus, as well as performing phylogenetic analyses of on LSU, SSU, TEF1α, and RPB2 DNA sequence data. The objectives of this study are as follows: (i) describe and illustrate the newly collected Acrogenospora spp. from freshwater habitats in China; (ii) provide molecular data for Acrogenospora species and understand their phylogenetic relationships.

Isolation and Morphology
Samples of submerged wood were collected from Yunnan and Tibet provinces, China and taken to the laboratory in plastic bags. The samples were incubated in plastic boxes lined with moistened tissue paper at room temperature for one week. Specimen observations and morphological studies were conducted following the protocols provided by Luo et al. (2018).
Single spore isolations were carried out following the method described in Chomnunti et al. (2014). Germinating conidia were transferred aseptically to PDA and MEA plates supplemented with 100 mg of streptomycin and grown at room temperature in daylight. Colony color and other characters were observed and measured after 1 week and again after 3 weeks. The specimens were deposited in the Mae Fah Luang University (MFLU) Herbarium, Chiang Rai, Thailand. Living cultures are deposited in the Culture Collection at Mae Fah Luang University (MFLUCC). Facesoffungi numbers (FoF) were acquired as in Jayasiri et al. (2015) and Index Fungorum (2020). New species are established following recommendations outlined by Jeewon and Hyde (2016).

DNA Extraction, PCR Amplification, and Sequencing
Fungal mycelium was scraped from the surface of colonies grown on potato dextrose agar (PDA) or malt extract agar (MEA) at 25 • C for 4 weeks, transferred into a 1.5 mL centrifuge tube and ground using liquid nitrogen. The EZ geneTM fungal gDNA kit (GD2416) was used to extract DNA from the ground mycelium according to the manufacturer's instructions. Primers for PCR amplification used were LSUrDNA = LR0R/LR7 (Vilgalys and Hester, 1990), SSUrDNA = NS1/NS4 (White et al., 1990), (TEF1α) = 983F/2218R and (RPB2) = fRPB2-5F/fRPB2-7cR (Liu et al., 1999). The PCR mixture was prepared as follows: 12.5 µl of 2 × Power Taq PCR MasterMix, 20 mM Tris-HCl pH 8.3, 100 Mm KCl, 3 mM MgCl 2 , stabilizer, and enhancer), 1 µl of each primer, 1 µl genomic DNA extract and 9.5 µl deionized water. The PCR of LSU, SSU and TEF1α gene was processed as follows: 94 • C for 3 min, followed by 35 cycles of denaturation at 94 • C for 30 s, annealing at 56 • C for 50 s, elongation at 72 • C for 1 min and a final extension at 72 • C for 10 min, and finally kept at 4 • C. The RPB2 gene region was amplified with an initial denaturation of 95 • C for 5 min, followed by 40 cycles of denaturation at 95 • C for 1 min, annealing at 54 • C for 40 s, elongation at 72 • C for 90 s, and the final extension at 72 • C for 10 min. PCR amplification was confirmed on 1% agarose electrophoresis gels stained with ethidium bromide. Purification and sequencing of PCR products were carried out using the above-mentioned PCR primers at Beijing Tsingke Biological Engineering Technology and Services Co., Ltd. (Beijing, P.R. China).

Sequencing and Sequence Alignment
Sequences were assembled with BioEdit and those with high similarity indices were determined from a BLAST search to find the closest matches with taxa in Acrogenospora and from recently published data (Jayasiri et al., 2018;Hyde et al., 2019). All consensus sequences and the reference sequences were automatically aligned with MAFFT v. 7 and the strategy was using Auto (Katoh and Standley, 2013) 1 . Aligned sequences of each gene region (LSU, SSU, TEF1α and RPB2) were combined and manually improved using BioEdit v. 7.0.5.2 (Hall, 1999). Ambiguous regions were excluded from the analyses and gaps were treated as missing data. Phylogenetic analyses were performed using Maximum Likelihood (ML) and Bayesian tree building criteria.

Phylogenetic Analyses
Maximum likelihood analysis was performed at the CIPRES Science Gateway v.3.3  2 using RAxML v. 8.2.8 as part of the "RAxML-HPC2 on XSEDE" tool (Stamatakis, 2006;Stamatakis et al., 2008). All model parameters were estimated by RAxML. The final ML search was conducted using the GTRGAMMA + I model which was estimated by using MrModeltest 2.2 (Nylander, 2004), Maximum likelihood bootstrap support was calculated from 1000 bootstrap replicates.
Bayesian analysis was performed using MrBayes v 3.1.2. (Ronquist and Huelsenbeck, 2003). The model of each genes was estimated using MrModeltest 2.2 (Nylander, 2004), GTR + I + G model was the best-fit model of LSU, SSU, TEF1α and RPB2 for Bayesian analysis. Posterior probabilities (PP) (Rannala and Yang, 1996) were performed by Markov chain Monte Carlo sampling (BMCMC) in MrBayes v.3.1.2 (Liu et al., 2012). Six simultaneous Markov chains were run for 50 million generations, and trees were sampled every 5000th generation (resulting in 10,000 trees). The first 2000 trees representing the burn-in phase of the analyses were discarded and the remaining 8000 (post burning) trees were used for calculating posterior probabilities (PP) in the majority rule consensus tree (Cai et al., 2006;Liu et al., 2012).
Maximum-parsimony analyses were performed using PAUP v.4.0b10 (Swofford, 2003). Gaps were treated as missing data with the heuristic search option with 1000 random sequence additions and tree bisection reconnection (TBR) branchswapping. Maxtrees were unlimited, branches of zero length were collapsed and all parsimonious trees were saved. The consistency indices (CI), tree length (TL), homoplasy index (HI), rescaled consistency indices (RC), retention indices (RI) were calculated for each tree. Clade stability was assessed using a bootstrap (BT) analysis with 1000 replicates, each with 10 replicates of random stepwise addition of taxa. Other details are as provided by Jeewon et al. (2002Jeewon et al. ( , 2003 Phylogenetic trees were represented by FigTree v. 1.4.4 (Rambaut, 2014) and edited in Microsoft Office PowerPoint 2016 (Microsoft Inc., United States). Newly generated sequences in this study were deposited in GenBank (Table 1) and the alignment used for the phylogenetic analyses were submitted to TreeBASE 3 under the accession number: 26373.

Phylogenetic Analyses
The combined LSU, SSU, TEF, and RPB2 sequence dataset included 101 taxa (ingroup) and two outgroup taxa (Diploschistes ocellatus and Stictis radiata) with a total of 3853 characters (LSU: 867 bp; SSU:1020 bp; TEF1α: 915 bp; RPB2: 1051 bp) after alignment including the gaps. The RAxML and Bayesian analyses of the combined dataset resulted in phylogenetic reconstructions with largely similar topologies and the result of ML analysis with a final likelihood value of -49015.757408 is shown in Figure 1. The matrix had 1149 distinct alignment patterns, with 29.85% undetermined characters or gaps. Estimated base frequencies were: A = 0.253075, C = 0.235518, G = 0.278280, T = 0.194616; substitution rates AC = 1.387195, AG = 3.679854, AT = 1.133462, CG = 0.233127, CT = 7.472473, GT = 1.000000; gamma distribution shape parameter α = 0.303701. Bootstrap support values for RAxML and MP greater than 60% and Bayesian posterior probabilities greater than 0.95 are given at each node (Figure 1).
Frontiers in Microbiology | www.frontiersin.org  to subglobose basal cell. Acrogenospora aquatica can be distinguished from A. basalicellularispora by the size of conidiophores (259-395 × 8-12 vs. 202-250 × 7.8-9.3 µm). In addition, conidia of A. basalicellularispora are pale orange-brown to olivaceous-brown, with several small to large guttules, while conidia of A. aquatica are dark brown to black and lack guttules. Notes: In the phylogenetic analysis, Acrogenospora basalicellularispora clustered with A. sphaerocephala (CBS 206.36) with low support (66% ML and 0.98 BYPP). Unfortunately, CBS 206.36 lacks a morphological description and only LSU sequence data is available in GenBank. Morphologically, our new isolate can be distinguished from other Acrogenospora species by its pale orange-brown to olivaceous brown, broadly obovoid to spherical conidia with several small to large guttules and a small, hyaline, subcylindrical to subglobose basal cell. In our study, A. aquatica also has conidia with a basal cell. However, we can distinguish them by the shape (broadly obovoid to spherical vs. subprolate to broadly ellipsoidal) and color (pale orange-brown to olivaceous brown vs dark brown to black) of conidia and size (260-395 × 8-12 vs. 200-250 × 7.5-9.5 µm) of conidiophores. Holotype-MFLU 20-0289 Etymology-Referring to the large guttule in the conidia.
Notes: Acrogenospora subprolata is characterized by conidiophores that are macronematous, mononematous, solitary or in groups of 2-4 with multiple percurrent proliferations and by acrogenous, subprolate to broadly ellipsoidal, pale orangebrown to olivaceous brown, aseptate, thick-walled conidia. Our isolate fits well with the characters of A. subprolata as described by Goh et al. (1998). Therefore, we identify this collection as A. subprolata.
Saprobic on submerged decaying wood. Notes: In the phylogenetic analysis, Acrogenospora yunnanensis shares a sister relationship to A. submersa. Morphologically, A. yunnanensis can be distinguished from A. submersa by the longer conidiophores ( Table 2) and color of conidia. Acrogenospora yunnanensis has dark brown to black conidia with a large guttule, while conidia of A. submersa are pale orange-brown to olivaceous brown at maturity and lack a guttule.
Morphologically, A. yunnanensis is most similar to A. gigantospora and A. subprolata in having similar conidial shape. However, they differ in size of conidiophores and conidia ( Table 2).

DISCUSSION
In this study, we provide new descriptions and illustrations for seven new species and two known species of Acrogenospora. This study contributes to a better taxonomic understanding and proposes that there could be a number of additional new species within the genus and its diversity could be much higher than anticipated. Acrogenospora species are cosmopolitan with worldwide distribution, they are mainly found on dead and submerged wood especially in freshwater habitats (Hughes, 1978;Goh et al., 1998;Ma et al., 2012;Hyde et al., 2019). Among the 20 Acrogenospora spp., 15 species were reported from freshwater habitats and only 5 of them were recovered from terrestrial habitats. Our study has shown that in a small area of Yunnan Province there are 14 species of Acrogenospora in streams alone and indicates that the genus is highly diverse, and has been found to occur with other genera in the region . Previous studies have also reported that there could be an amazing fungal diversity hidden in the South East Asian region .
Acrogenospora species are quite similar to each other, and previous studies suggested to distinguish them based on conidial shape, size, and color and the degree of pigmentation of the conidiophores (Hughes, 1978, Table 2). We found that guttules and basal cells of conidia are also important characters to distinguish species and a morphological comparison of all Acrogenospora species is provided ( Table 2).
Previous publications on submerged wood in freshwater have lumped several Acrogenospora collections and identified them based on morphology as A. sphaerocephala ( Table 3) perhaps because of the difficulty of using morphs alone to delineate species and due to a lack of DNA sequence data. It is likely that the collections of A. sphaerocephala in older publications ( Table 3) are wrongly named and further taxonomic work is necessary.
Before this study, there were 13 species of Acrogenospora but only three of them had sequence data available in GenBank, and there was no data for the ex-type strains. Our phylogenetic sampling included 12 strains of Acrogenospora (Acrogenosporaceae), and all strains grouped with four species of Minutisphaera (Minutisphaeraceae) within Minutisphaerales (99% ML and 0.99 PP, Figure 1). The results were similar to the analyses by Jayasiri et al. (2018). In our analyses, Acrogenospora carmichaeliana (CBS 206.36) did not cluster with other strains of A. carmichaeliana, instead clustered with our new isolate A. basalicellularispora with low statistical support. Unfortunately, there are no morphological descriptions for CBS 206.36, so we are unable to compare its morphology with our new isolate. Further collections and phylogenetic studies of Acrogenospora are needed to better understand the phylogenetic placement of those species which lack sequence data and, undoubtedly, many more novel species can be found.
The phylogenetic analysis provide clear resolution to the taxonomic complexities within this group. Protein-coding genes have been shown to be essential to identify a taxon up to species level (Tang et al., 2007(Tang et al., , 2009Jeewon et al., 2017). In our study, we sequenced the RPB2 and TEF1α sequence data to distinguish Acrogenospora species and the phylogenetic trees are provided in Supplementary Files 1, 2. In our phylogenetic tree (LSU + SSU + TEF1α + RPB2, Figure 1), Acrogenospora aquatica, A. guttulatispora, A. submersa and A. yunnanensis grouped together, but they constitute different clades based on phylogenies derived from the TEF and RPB2 data which clearly support that they are phylogenetically distinct species. Acrogenospora verrucispora clustered with A. carmichaeliana (Figure 1), but there are 9 bp differences in TEF1α gene region. In addition, they can be easily distinguished from each other by the shape, color and wall of conidia, (conidia of A. verrucispora are spherical or subspherical, orange-brown to olivaceous brown, distinctly verrucose-walled, while A. carmichaeliana has broadly ellipsoidal to obovoid, brown to dark brown, smooth-walled conidia). Acrogenospora olivaceospora is close to A. sphaerocephala, however, there are 12.3% nucleotide differences in RPB2 gene region between them. These results support our establishment of the new taxon as recommended by Jeewon and Hyde (2016). As for A. basalicellularispora and A. subprolata there are no DNA sequences available from protein-coding gene but they can be easily distinguished from other species based on morphological characters ( Table 2). Hyde et al. (2019) discussed whether Acrogenospora sphaerocephala was the asexual morph of Farlowiella carmichaeliana and whether A. megalospora was wrongly introduced as the asexual morph of F. carmichaeliana. In our phylogenetic analyses, Acrogenospora sphaerocephala did not cluster with A. carmichaeliana, forming different clades within Acrogenosporaceae. DNA sequences of Ex-type strains of both A. megalospora and Farlowiella carmichaeliana are unavailable in GenBank. Therefore, the connection of sexual and asexual morph of Farlowiella carmichaeliana is not clear and this needs further morpho-molecular evidence.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.