Serendipita is one of the main fungal genera that form mutualistic associations with species of orchids (Orchidaceae). Seven new Serendipita species associated with various Australian orchid genera are described. These Serendipita species were originally characterized by multilocus DNA sequence species delimitation analyses (three mtDNA and four nuclear genes) and confirmed as distinct with addition of further isolates and re-analysis of the nuc rDNA ITS1-5.8S-ITS2 (ITS) and nuc 28s rDNA (28S). Morphology and micro-features of cultures of each of the species are described. Three of the new species are binucleate, whereas the other four are multinucleate. For the ITS region, the seven species have within species sequence divergence between 1.07 and 4.31 % and all but one of the species pairs are separated by interspecific divergence of at least 4.35 %. The newly described Serendipita species, S. australiana, S. communis, S. occidentalis, S. rarihospitum, S. secunda, S. talbotii, and S. warcupii are shown to be separate species to S. vermifera on the basis of comparison against a sequence from the type. Isolates originally identified by Warcup as Sebacina “vermifera” from Caladenia orchids are reidentified as belonging to three of the species newly described here. Some non-Caladenia isolates identified by Warcup as S. “vermifera” are also shown to be non-conspecific with the type of S. vermifera. On the basis of ITS sequences, 346 isolates from 26 other studies, previously identified under provisional designations, are accomodated under the novel species. The species of Serendipta described here associate with the Australian orchid genera Caladenia, Cyanicula, Elythranthera, Ericksonella, Eriochilus, Glossodia, and Pheladenia. Most of the novel Serendipita species occur widely across Australia, often with widely distributed hosts, but one species, Serendipita rarihospitum, associates with narrowly distributed orchid species.

Fungal isolation.- To broaden the breadth of sampling for some Serendipita OTUs (B, D, E, F, H) identified in Whitehead et al. (2017), additional fungal isolates were obtained from several orchid species including Glossodia major (Australian Capital Territory, New South Wales, Queensland), Elythranthera brunonis (Western Australia), Caladenia attingens (Western Australia), and C. cairnsiana (Western Australia). Fungal pelotons were isolated from the collar regions of orchid roots by cutting this region into sections in sterile double-distilled water (ddH₂O) using a scalpel blade. Sections containing pelotons were serially rinsed in four changes of sterile dd.H₂O and torn apart in the last rinse with sterile forceps. The pelotons were then transferred to half-strength Fungal Isolation Medium agar (½ FIM) (Clements and Ellyard 1979) using a pipette, after which plates were sealed and incubated at 23 C for 2–4 d. The growing mycelium from the pelotons were then subcultured onto quarter-strength Potato Dextrose Agar (¼ PDA) or ½ FIM for approximately two wks. All cultures were stored and maintained in Mili-Q H₂O at 7 C. Voucher specimens of the fungi, as living cultures stored in a metabolically inactive state, were lodged at the Royal Botanic Gardens Victoria Fungal Culture Collection (RBGV-FCC) and the culture collection of the Department of Agriculture, Victoria (VPRI). In addition, specimens prepared by freeze-drying cultures grown in ½ FIM broth were lodged at the National Herbarium of Victoria (MEL).

DNA extraction, PCR, Sequencing, and phylogenetic analysis.- Protocols for DNA extraction, polymerase chain reaction (PCR), and sequencing followed Oktalira et al. (2019). The 11 additional Serendipita samples isolated in this study (SUPPLEMENTARY TABLE 1) were sequenced for the nuc rDNA ITS1-5.8S-ITS2 (ITS) using universal primers ITS1 and ITS4 (White et al. 1990). Sequences obtained in the present study were checked and edited using Sequencher 5.4.1 (Gene Codes Corp., Ann Arbor, Michigan). These new sequences were deposited in GenBank –(SUPPLEMENTARY TABLE 1).

Three analyes based on different sets of taxa were carried out to investigate the phylogenetic position and species delimitation of isolates of Serendipita from Australian orchids. First, we analysed the position of Australian orchid isolates in relation to the diversity in the family Serendipitaceae by constructing an alignment of the ITS-28S region including (1) up to three sequences representative of each of the eight OTUs defined by Whitehead et al. (2017), (2) sequences from described Serendipita species where available (S. herbamans, S. indica, S. restingae, S. vermifera, S. whamiae, and S. williamsii), (3) sequences from isolates originally identified as S. vermifera by Warcup, (4) sequences representing the major subclades of the genus Serendipita from previous studies (Weiss et al. 2011; Weiss et al. 2016), and (5) sequences of several Helvellosebacina and Sebacina representating Sebacinaceae. Auricularia auricula-judae (Bull.) Quél was used as the outgroup following Weiss et al. (2011). While we attempted to only include taxa for which both ITS and 28S sequences were available, both sequences were not always available for key reference taxa (19/79 taxa lacked an ITS sequence, while the 28S was not available for six taxa) (FIG. 1, SUPPLEMENTARY TABLE 2).

 For the second and third analyses, we placed additional isolates into the species framework generated by the first analysis, by analysing ITS (SUPPLEMENTARY FIG. 1) and 28S (SUPPLEMENTARY FIG. 2) alignments separately because Serendipita sequences from Australian orchids were mostly available for ITS only. The separate ITS and 28S analyses incorporated sequences from the study of Whitehead et al. (2017) and Serendipita sequences isolated from Australian orchids from GenBank (SUPPLEMENTARY TABLE 3), selected from BLAST searches against each OTU as identified in Whitehead et al. (2017) matching at least 92 % sequence similarity for ITS or 90 % for 28S. For ITS, we additionally incorporated 11 sequences newly obtained in this study (SUPPLEMENTARY TABLE 1). For 28S, the 123 sequences from Whitehead et al. (2017) are identified by numbers preceded by “CLM” (prefix for orchid fungal isolates in the culture collection of Celeste Linde Mycorrhizal collection – Australian National University) and the sequences were retrieved from Dryad (https://doi.org/10.5061/dryad.2hf40). In total there were 510 ITS sequences and 195 28S sequences.

Alignments were constructed in ClustalW (Thompson et al. 1994) and manually checked. Phylogenetic trees were estimated using both maximum likelihood (ML) and Bayesian inference (BI). A ML phylogenetic analysis with 1000 replicate bootstraps was performed using RAxML (Stamatakis et al. 2008). The BI analysis was performed with MrBayes 3.2.6 (Huelsenbeck and Ronquist 2001; Ronquist and Huelsenbeck 2003) with node support assessed by Bayesian posterior probability (BPP) and calculated after running two parallel runs of four heated chains. Convergence of runs was confirmed when the average standard deviation of the split frequencies reached < 0.01 and estimated sampling sizes were (ESS) > 200. For both ML and BI analyses, the best-fit substitution model was selected by ModelFinder (Kalyaanamoorthy et al. 2017) in IQ-TREE 2.0 according to the Bayesian Information Criterion (Nguyen et al. 2014) for each locus and partition.  The concatenated ITS-28S alignment was partitioned by each locus.