Re-Evaluation of the Podosphaera tridactyla Species Complex in Australia

The Podosphaera tridactyla species complex is highly variable morphologically and causes powdery mildew on a wide range of Prunus species, including stone fruit. A taxonomic revision of the Po. tridactyla species complex in 2020 identified 12 species, seven of which were newly characterised. In order to clarify which species of this complex are present in Australia, next generation sequencing was used to isolate the fungal ITS+28S and host matK chloroplast gene regions from 56 powdery mildew specimens of stone fruit and ornamental Prunus species accessioned as Po. tridactyla or Oidium sp. in Australian reference collections. The specimens were collected in Australia, Switzerland, Italy and Korea and were collected from 1953 to 2018. Host species were confirmed using matK phylogenetic analysis, which identified that four had been misidentified as Prunus but were actually Malus prunifolia. Podosphaera species were identified using ITS+28S phylogenetic analysis, recognising three Podosphaera species on stone fruit and related ornamental Prunus hosts in Australia. These were Po. pannosa, the rose powdery mildew, and two species in the Po. tridactyla species complex: Po. ampla, which was the predominant species, and a previously unidentified species from peach, which we describe here as Po. cunningtonii.


Introduction
Stone fruit such as peaches (Prunus persica), cherries (Pr. avium), apricots (Pr. armeniaca), plums (Pr. domestica, Pr. salicina and Pr. cerasifera) and almonds (Pr. dulcis) belong to the genus Prunus (Rosaceae), which contains approximately 250 species distributed across temperate regions worldwide [1]. In Australia, there are only two native Prunus species, Pr. brachystachya and Pr. turneriana; both are in the subgenus Cerasus and found in tropical rainforests of north east Australia. A further 15 species of Prunus have been introduced to Australia as horticultural crops and for use in gardens [2]. On the basis of phylogenetic analysis of multiple genes, the genus is subdivided into three subgenera: Cerasus (cherries), Padus (bird cherries, including species formerly placed in Laurocerasus) and Prunus (plums) and the latter subgenus is further subdivided into seven sections: Amygdalus, Armeniaca, Emplectocladus, Microcerasus, Persicae, Prunocerasus and Prunus [3]. The three subgenera are also distinguished on morphology according to inflorescence structure, where subgenus Cerasus has corymbose inflorescences, Padus has racemose inflorescences and Prunus has solitary inflorescences [4].
The Australian stone fruit industry was established in the late 1800s by European and Chinese settlers who introduced apricot, peach, nectarine and plums that were in

Fungal Sampling and Morphological Characterisation
Powdery mildew conidia, mycelia and chasmothecia were collected from the specimens by using a 6 mm leaf punch or scraping with a blade. Sampling was completed under clean room conditions to minimise contamination with modern DNA.
The VPRI specimens collected between 1993 and 1995 were morphologically examined while fresh at the time of collection. The hyphae, conidiophores and conidia were collected off the leaf surface with clear cellotape, which was then mounted on a microscope slide with lactofuchsin mountant and observed using light microscopy. Lactofuchsin preparation was as follows: lactic acid 20 g, glycerol 40 g, water (H 2 O) 20 mL and acid fuchsin 0.1 g; this was adapted from the Johnson and Booth [13] lactophenol mountant recipe minus the phenol. Germination patterns were studied on host tissue by pressing a sporulating colony onto a fresh, healthy leaf and incubating the leaf in a petri dish with moist filter paper, then examined by cellotape strip after 24 and 48 h. Reference collection specimens collected between 1977 and 1992 were rehydrated before examination, as described by Shin [14] and Shin and La [15], by placing a small section of infected leaf upside down in a drop of lactic acid on a microscope slide, then placing a flame underneath to boil. Once boiled, the rehydrated fungal material was scraped off the leaf surface and mounted in lactic acid for examination by light microscopy [14,15].
Morphological characters were recorded for both fresh and reference collection specimens as follows: mycelium growth pattern and hyphal structure, mycelial appressoria shape, conidiophore characteristics (length, shape, position of basal septum), conidial chain edge line, conidial size and shape, presence or absence of fibrosin bodies and germination characters.

DNA Extraction and Quantification
DNA was extracted following the E.Z.N.A. ® Forensic DNA protocol [12]. The DNA extraction initial incubation step was increased to one hour and included a repeated final elution step (50 µL of elution buffer were added to the filter column with a 5 min incubation for a final volume of 100 µL). A NanoDrop 2000™ (Thermo Fisher Scientific, Waltham, Massachusetts) was used to assess DNA quality using the 260/280 nm absorbency ratio (1.8-1.9). DNA concentrations were quantified using two methods: Quantus™ fluorometer (Promega) and Agilent Tapestation ® electrophoresis (Agilent Technologies, Santa Clara, CA, USA).

Library Preparation and Sequencing
Library preparation was as described in Smith et al. [12]. Libraries comprising VPRI DNA were paired-end sequenced on the Illumina HiSeq 3000 platform (San Diego, CA, USA). Libraries comprising BRIP and DAR DNA were paired-end sequenced on the Illumina MiSeq platform, due to sequencing platform accessibility constraints at the time.

Sequence Analysis
Sequence reads were assigned to each sample based on their indices. Raw DNA-Seq files were filtered using the Nuclear program to trim adapters and retain only the sequences above the cut-off value of 100 bp read length and 20 mismatches. The library preparations were expected to contain Podosphaera DNA and plant host DNA, as well as DNA from microbes present on the leaf surface at the time of collection. Therefore, the filtered, high-quality (HQ) reads were mapped as paired-end reads to Podosphaera and plant host reference sequences. For Podosphaera, the reference sequences used were the internal transcribed spacer ITS (ITS1-5.8S-ITS2) and 28S rDNA regions of Po. ampla was identified by mapping to matK chloroplast plastid regions. Reference sequences were those identified by Chin et al. [4] or taken directly from GenBank. The matK accessions from Chin et al. [4] were: Pr. armeniaca (HQ235100), Pr. cerasifera (HQ619834), Prunus domestica (HQ235146), Pr. laurocerasus (HQ235181), Pr. mahaleb (HQ235184), Pr. padus (HQ235216), Pr. persica (HQ235409), Pr. salicina (HQ235252), and the others from GenBank were Pr. yedoensis (GQ248191) and Malus prunifolia (JQ391019). The sequence mapping was performed by the program Nuclear, generating reference-initiated sequence alignments to be viewed in Vision software (version 3.3.6 Gydle Inc. Bioinformatics Service, Québec City, Québec; http://www/gydle.com, accessed on 3 February 2021). In Vision, ITS+28S and matK sequences were edited to incorporate single nucleotide polymorphisms (SNPs) and indels which related to each individual specimen's sequence data. A mapping threshold was determined by a minimum of 5x coverage of the complete reference sequence and sequence files which did not meet this criterion were excluded. Mapping success was determined from the Vision images of each sequence file by calculating the total number of aligned DNA sequence reads and converting it into a percentage from the total number of HQ reads. The mapped ITS+28S and matK sequences generated for this study were exported from Vision for identification to species by BLASTn and phylogenetic analysis.

Phylogenetic Analysis
Based on preliminary phylogenetic analysis, two data sets were generated for the sequences originally identified as Po. tridactyla. The first data set ( Figure 1) was constructed for ITS (rDNA ITS1-5.8S-ITS2) sequences from VPRI Po. tridactyla specimens reidentified as Po. leucotricha and Po. pannosa, which included sequences from closely related Podosphaera. The second data set contained ITS+28S (rDNA ITS1-5.8S-ITS2 + rDNA 28S large subunit) sequences from the Po. tridactyla complex ( Figure 2). Both data sets included sequences generated from this study and sequences of Podosphaera obtained from GenBank that were selected from published studies [11,16,17] and NCBI searches of Podosphaera spp. ITS and ITS+28S sequences. Phylogenetic analyses of these two data sets used Cystotheca lanestris (GenBank accessions AF011289 and AF011288) as the outgroup taxon, in line with Meeboon et al. [11]. The phylogenetic analysis of host matK sequences included sequences from Prunus species subgenera Cerasus, Padus and Prunus obtained from Chin et al. [4] and the outgroup was Oemleria cerasiformis (matK AF288110). Alignments were generated in Geneious Prime using the Muscle 3.8.425 [18,19] alignment tool with suggested settings followed by manual refinement. Phylogenetic trees were obtained from the aligned sequence data by maximum likelihood (ML) and Bayesian inference (BI) methods. ML analysis was performed in PhyML (3.3.20180621) [20] with the general time-reversible (GTR) substitution model with optimization for topology/length/rate with the proportion of invariable sites set at 0 and number of substitution rate categories at 4. The bootstrap analysis was set at 1000 replications with the stepwise addition option set as simple. BI analysis was completed in MrBayes (3.2.6) [21] using two runs with four chains each under the GTR model and run assuming a gamma distribution variation. Four heated chains and a single cold chain were used in all Markov chain Monte Carlo (MCMC) analyses, which were run for 1,100,000 generations and sampled one tree every 200 generations. Burn-in length was set at 100,000.
To identify characteristic bases for sets of closely related species, summaries of variable base positions were generated for (1) Po. ampla and sister taxa and (2) species closely related to Po. pruni-avium. For each set of sequences, outgroup sequences were removed, and the sequences of interest were realigned. Masking of the amended alignments by removing sites containing identical bases, leaving only sites with variable bases (including gaps), was performed by the Mask Alignment tool in Geneious Prime. In addition, positions where only one sequence within a species varied from the consensus were removed manually in Geneious Prime. Base position numbers of the variable sites were manually edited in Microsoft PowerPoint to reflect base positions in the original alignment.  Branch support values for maximum likelihood and Bayesian inference analyses are shown when >70% and 0.85, respectively. *Type indicates sequences obtained from isotype or holotype specimen for that species. Sequences generated in this study are shown in bold with names as originally listed in reference collections. Colours represent Prunus subgenera; Cerasus is pink, Padus is blue and Prunus is yellow.

Sequencing
DNA from 56 of the 58 collections were successfully mapped to reference sequences of the powdery mildew (Podosphaera) ITS+28S region and host plant chloroplast gene matK. The exceptions were DAR 35281 and DAR 64667, which failed to meet the mapping threshold. The ITS+28S and matK sequences generated from this study were confirmed as belonging to Podosphaera and Prunus, respectively, using BLASTn. The BLASTn analysis of the powdery mildews detected six collections which did not match with species of the Po. tridactyla complex and matched to either Po. pannosa or Po. leucotricha. These six sequences were placed in a data set with other sequences identified at Po. leucotricha and Po. pannosa as well as closely related Podosphaera species, as they are not part of the Po. tridactyla species complex. The remaining ITS+28S sequences returned BLASTn percentage identities of at least 98% and up to 100% (E values were 0.0) with members of the Po. tridactyla species complex.

Phylogeny
Phylogenetic analysis of the first data set compared ITS sequences of the six collections reidentified as Po. pannosa and Po. leucotricha with closely matched Podosphaera from GenBank and published sources [16,17] Table 2).
A summary of variable bases for the Po. ampla sequences shows that the subclade containing the Po. ampla sequences from Germany, when compared to the remaining members of the clade, exhibits a one base pair difference at position 445 where a T is present instead of a C ( Figure S1). Sequences MK530450 and MK530451 from Germany have two additional base changes at positions 813 (G instead of T) and 816 (A instead of G) ( Figure S1). Sequence AF154321 displayed identical base pairs to other Po. ampla sequences, although it was significantly shorter at 479 bp in length. Sequence AY833656 was also shorter in length (503bp) but included three variants at positions 545 (A instead of G), 546 (T instead of A) and 550 (A instead of G).  Separate analyses of alignments of each of ITS (ITS1-5.8S-ITS2), including 134 sequences covering 486 base pairs, and of 28S (large subunit) including 73 sequences of 808 base pairs, formed the same overall structure regarding all major clades. All Australian sequences generated from this study fell into two well-supported clades (Po. ampla and the unknown species of Podosphaera). Support values for the Po. ampla clade, including the Australian sequences, were as follows: ITS (BS 80, BI 1) and 28S (BS 100, BI 1) ( Figures S2 and S3). In the ITS tree, two Australian sequences (AY833656 and AF154321) published by Cunnington et al. [10] and labelled as Podosphaera sp. by Meeboon et al. [11] fell into the Po. ampla clade.
Eight Australian Podosphaera sequences on Pr. persica and Pr. mahaleb hosts, including two Australian sequences (AY833651 and AY833653) published by Cunnington et al. [10] and also labelled as Podosphaera sp. by Meeboon et al. [11], formed an independent lineage from other species of the Po. tridactyla complex. In the ITS+28S phylogeny, support for this new Podosphaera species is high (BS 84 and BI 1) (Figure 2). The individual ITS and 28S phylogenies also have strong support for this clade: ITS (BS 87, BI 1) and 28S (BS -, BI 0.98) ( Figures S2 and S3). A summary of variable bases for the novel Podosphaera species shows five base-pair positions in 1234 characters at which there are differences from the two sister taxa, with differences at three positions compared to Po. pruni-avium and two positions compared to Po. pruni-japonicae ( Figure S4). The new sequence generated for VPRI 19591 by next generation sequencing in the current study was taken from the same specimen that was used to generate sequence AY833653 by Sanger sequencing in 2005; the two sequences are identical.

Fungus-Host Relationships
The ITS+28S phylogeny of the fungi showed several sequences from hosts belonging to different Prunus subgenera compared to the hosts detected by Meeboon et al. [11] (Figure 2). Three species, Po. pruni-avium, Po. pruni-japonicae and Po. prunigena, were all originally described from Prunus hosts within the subgenus Cerasus, but in the ITS+28S phylogeny, the clades for these species included VPRI 22159, VPRI 20490, VPRI 20233 and VPRI 20491 on Pr. padus, which is in subgenus Padus. The Podosphaera pruni-japonicae clade in the ITS tree also included sequences from fungi collections on Prunus davidiana (subgenus Prunus) ( Figure S2). The ITS tree includes additional fungi sequences from GenBank, in which Po. pruni-avium forms two separate lineages; the first lineage consists of collections on both Pr. subgenus Cerasus and Padus and the second lineage has collections only on Pr. subgenus Cerasus ( Figure S2). The clade consisting of collections of the undescribed powdery mildew on peach, Prunus persica (subgenus Prunus), also included a sequence from a collection on Pr. mahaleb (subgenus Cerasus). Both Po. prunina and Po. ampla clades contained powdery mildew sequences generated from Prunus hosts within subgenus Prunus in all three phylogenetic analyses.

Plant Host Phylogeny
There were nine different Prunus species originally listed as the plant hosts for Australian Po. tridactyla complex collections (Pr. armeniaca, Pr. cerasifera, Pr. domestica, Pr. laurocerasus, Pr. mahaleb, Pr. padus, Pr. persica, Pr. salicina and Pr. yedoensis). The plant host species were analysed in a matK phylogenetic analysis containing 133 sequences of 1288 base pairs, including representatives of the three subgenera of Prunus (Figure 3).
The phylogeny forms three main clades which represented the Prunus subgenera Cerasus (clade 1), Padus (clade 2) and Prunus (clade 3 There are minimal differences between these species within the matK gene region and therefore, for the purpose of identification, this group is referred to as the Pr. domestica group which includes the previously mentioned Prunus species as well as Pr. brigantina, Pr. consociiflora, Pr. simonii and Pr. spinosa. Additionally included in the clade of Eurasian plums were sequences from BRIP 15118 and VPRI 19868, which were both previously misidentified as Pr. persica (Figure 3). There were 14 specimens with plant hosts originally listed only as Prunus sp., in the matK phylogeny, these sequences fell within the clade of Eurasian plums in section Prunus (Table 2).
There were four VPRI specimens, VPRI 12495, 18452, 18514 and 20687, which are labelled with plant hosts listed as Prunus persica, Pr. sp., Pr. domestica and Pr. sp., respectively. However, the matK sequences identified the hosts as Malus prunifolia (plum-leaf crab apple) and the associated powdery mildews were all re-identified as Po. leucotricha.

Morphological Characterisation
The phylogenetic analysis showed a well-supported independent lineage of an undescribed species which is sister to Po. pruni-avium and Po. pruni-japonicae. No sexual morphs were found among material of this species. The morphologies of the sister taxa barely differ from Po. tridactyla in the strict sense and the asexual characters of this undescribed species on Prunus persica and Pr. mahaleb are the shorter conidiophore foot cells, crenate conidial chains and smaller conidia when compared to characters of the asexual morph of Po. tridactyla, which has long foot cells and ellipsoid(-doliform) conidia catenescent with crenate edge line (Figure 4).

Discussion
This study re-evaluated reference collection specimens of Po. tridactyla and Oidium sp. on Prunus hosts and, through phylogenetic analysis, found seven different powder mildew species and corrected misidentified plant hosts of eight specimens. The main spe cies present on Prunus in Australia was Po. ampla and we also found that Australian co lections of Po. pannosa and Po. leucotricha had been misidentified as Po. tridactyla and Oid ium sp. in two reference collections. Finally, we characterised a new species in the Po. tr dactyla complex which we name Po. cunningtonii from Pr. persica and Pr. mahaleb plan hosts.
The phylogenetic analysis based on sequences derived from NGS shows that 37 o the 56 specimens fell within a clade alongside sequences identified as Po. ampla by Mee boon et al. [11]. Most Australian sequences assigned to Po. ampla had a one base pair di ference compared to the Po. ampla sequences analysed in previous studies. The revision o the Po. tridactyla species complex by Meeboon et al. [11] included a number of sequence containing several bases varying from the Po. ampla holotype MK530453 but were sti included under Po. ampla. Therefore, most of the Australian powdery mildew fungi o stone fruit (Prunus armeniaca, Pr. cerasifera, Pr. domestica and Pr. salicina) formerly ident fied as Po. tridactyla are in fact Po. ampla. Etymology: the epithet commemorates Dr. James Cunnington, the first person to conduct molecular examination of this species.
Natural distribution: currently unknown as isolates were collected in quarantine glasshouses in Australia.

Discussion
This study re-evaluated reference collection specimens of Po. tridactyla and Oidium sp. on Prunus hosts and, through phylogenetic analysis, found seven different powdery mildew species and corrected misidentified plant hosts of eight specimens. The main species present on Prunus in Australia was Po. ampla and we also found that Australian collections of Po. pannosa and Po. leucotricha had been misidentified as Po. tridactyla and Oidium sp. in two reference collections. Finally, we characterised a new species in the Po. tridactyla complex which we name Po. cunningtonii from Pr. persica and Pr. mahaleb plant hosts.
The phylogenetic analysis based on sequences derived from NGS shows that 37 of the 56 specimens fell within a clade alongside sequences identified as Po. ampla by Meeboon et al. [11]. Most Australian sequences assigned to Po. ampla had a one base pair difference compared to the Po. ampla sequences analysed in previous studies. The revision of the Po. tridactyla species complex by Meeboon et al. [11] included a number of sequences containing several bases varying from the Po. ampla holotype MK530453 but were still included under Po. ampla. Therefore, most of the Australian powdery mildew fungi on stone fruit (Prunus armeniaca, Pr. cerasifera, Pr. domestica and Pr. salicina) formerly identified as Po. tridactyla are in fact Po. ampla.
The ITS phylogeny and 28S phylogenies also confirmed that most of the Australian sequences were Po. ampla and fell within a well-supported clade with Po. ampla sequences analysed by Meeboon et al. [11] (Figures S2 and S3). Podosphaera tridactyla sequences AF154321 and AY883656 generated by Cunnington et al. [10] were included in the ITS phylogeny and summary of variable sites and they both fell in with the other Po. ampla sequences, the only differences being that the sequence lengths were significantly shorter than the ITS+28S sequences which were generated by this study and Meeboon et al. [11] and AY883656 contained three variable base pairs towards the end of the sequence. We are confident in reidentifying these two sequences as Po. ampla.
The range of hosts for Po. ampla observed in this study included Pr. cerasifera and Pr. domestica as well as other species within subgenus Prunus section Prunus (Figure 3). In Meeboon et al. [11], Po. ampla was identified from Pr. domestica and Pr. spinosa, as well as from Pr. armeniaca and Pr. cerasifera. These Prunus hosts for Po. ampla were observed in this study as well as Pr. salicina, which is also within subgenus Prunus sect. Prunus. It would be expected in ideal conditions such as in a glasshouse for Po. ampla to infect a close relative of its listed hosts.
The results of the three phylogenetic analyses from this study contradict host specificity at the subgenus level in Prunus for Po. prunigena, Po. pruni-japonicae and Po. pruni-avium species, as suggested by Meeboon et al. [11] because each fungus species clade contained Prunus hosts from subgenera Cerasus and Padus. Chin et al. [4] suggested that species within the subgenus Cerasus have a close alliance with some temperate racemose species found within subgenus Padus, such as Pr. padus and Pr. laurocerasus. This might offer an explanation as to the inclusion of supposedly different Prunus host subgenera in these host genera-specific powdery mildew fungi, as the remaining species of the Po. tridactyla complex are specific at the host subgenus level.
The remaining species which were identified from the Australian reference collections were: Po. pannosa, Po. leucotricha, Po. pruni-japonicae, Po. pruni-avium, Po. prunina and Po. prunigena. Podosphaera pannosa and Po. leucotricha were identified from Australian collections which had misidentified fungus and plant hosts recorded. The phylogeny showed four VPRI specimens were actually Po. leucotricha or the apple powdery mildew on plum-leaf crab apple (M. prunifolia) and two specimens from BRIP and DAR were Po. pannosa, which is more commonly seen on roses. The remaining Po. tridactyla complex species were identified from collections made in Switzerland and South Korea but held at VPRI.
While most Australian specimens were Po. ampla, six of the sequences newly generated by NGS were of an unknown species which did not match any identified sequences on GenBank. Cunnington et al. [10] and Meeboon et al. [11] both stated that two of the six sequences represented in their studies were an independent genetic lineage within the Po. tridactyla complex. In the Cunnington et al. [10] study, the two sequences AY833651 and AY833653 formed a clade which was sister to a clade comprising Podosphaera on Pr. japonica, which was renamed Po. pruni-japonicae by Meeboon et al. [11], within clade 2, which contained sequences of Po. tridactyla not on Prunus subgenus Prunus. In the Meeboon et al. [11] study, the phylogeny showed sequences AY833651 and AY833653 forming a separate clade sister to a clade comprising Po. pruni-japonicae and Po. pruni-avium. The results of the present phylogenetic study align with both Cunnington et al. [10] and Meeboon et al. [11], where the ITS+28S phylogeny shows that sequences of Po. cunningtonii including additional sequences generated in this study form a well-supported clade which is sister to a clade comprising Po. pruni-japonica and Po. pruni-avium.
A GenBank search for sequences of Podosphaera species on Prunus persica returned 26 sequences of two powdery mildew species, Po. pannosa and Po. leucotricha (peach rusty spot), which are both found on peach [22][23][24]. However, the sequences of this new species do not match either of these powdery mildew species, further confirming a previously undescribed powdery mildew. Braun and Cook [7] list three powdery mildew species with Pr. persica included in the host ranges, these are Po. tridactyla in the strict sense, Po. prunicola and Po. pannosa, which are quite separate to Po. cunningtonii in the phylogeny.
At the time of the first collections of Po. cunningtonii in Victoria (mid-1990s), it was assumed that the quarantine glasshouse-grown Pr. persica had been infected by powdery mildew blowing into the glasshouse (Cunnington 2020 pers. comm.). However, in this study, we identified the same powdery mildew fungi on three specimens from New South Wales, two of which (DAR 28962 and DAR 28963) were also collected in quarantine glasshouses. The two specimens from quarantine glasshouses in N.S.W. are on two different hosts-Pr. persica (subgenus Prunus) and Pr. mahaleb (subgenus Cerasus). Notes in the specimen packet reported that the Pr. mahaleb seedlings were growing next to heavily powdery mildew-infected Pr. persica seedlings in the same glasshouse. The determiner, Dr. John Walker, noted at the time of collection (1977), "mildew development very light and feel that this is an adventitious development of mildew on this host from the heavily mildewed peach seedlings growing nearby". We suggest that the primary host of this new powdery mildew is Pr. persica but under glasshouse conditions it can infect other Prunus species from different subgenera as accidental hosts.
A third specimen of Po. cunningtonii, from N.S.W. (DAR 71638), was collected from Pr. persica on a private property in 1996, in a locality with several orchards that grew peach and other stone fruit in the near vicinity. This is the only report of Po. cunningtonii from outside of quarantine glasshouses and therefore we cannot confirm if it is established in Australia. Two specimens of powdery mildew on Pr. persica collected in 1991 from a quarantine facility in Alice Springs, Northern Territory, are lodged in the Northern Territory Plant Pathogen Collection (DNAP). Unfortunately, we were unaware of them at the time of commencement of the present study, so they were not included. At the time of collection, they were identified morphologically as Po. pannosa by determiner Dr. Jose Liberato. The two native Prunus species, Pr. brachystachya and Pr. turneriana, are only found in tropical rainforests in the far north of Queensland of Australia. Horticultural Prunus species, including stone fruit, are grown in the temperate regions of Australia further south [2,6]. Geographical distance and climatic differences would prevent the spread of a "native" Prunus powdery mildew fungus onto commercial stone fruit crops and to date there are no records of powdery mildew fungi infecting these native hosts. It is highly unlikely that Po. cunningtonii has a native area of distribution in the natural distribution of Pr. persica which originates from Northwest China. New peach varieties are imported into Australia as budwood and grown under quarantine prior to release. Potential explanations as to why Po. cunningtonii has not been observed more widely in Australia could be that infected plants would have not been released from quarantine. Another reason could be that the local climate in peach-growing regions in Australia is not climatically suitable for Po. cunningtonii development, whereas the quarantine glasshouse conditions provided an artificial environment for Po. cunningtonii growth [25,26].
Australia is believed to be a continent without native powdery mildew fungi, with accidental human-assisted powdery mildew introductions occurring post-European settlement on non-native agricultural, horticultural, ornamental and pasture plant species [27]. Further evidence for the lack of native powdery mildew fungi in Australia is the evolution of native plant species in long isolation from continents that now make up the northern hemisphere, which is where powdery mildew fungi appear to have originated and subsequently co-evolved with plant hosts [27]. Walker [28] investigated the distribution of plant-parasitic fungi across Australia and observed that most accounts of powdery mildew infection were on introduced monocot and dicot plant species and only a few on native plant species, with these infections occurring in artificial settings such as glasshouses or nurseries. After analysis of records of powdery mildew fungi from Australia, Kiss et al. [27] concluded that through European settlement, almost all of the agricultural and horticultural crops grown in Australia were recently introduced and with them powdery mildew. They also found that freshly collected powdery mildew specimens from Australia were phylogenetically similar to species known overseas and were collected mostly on imported plant hosts but were also identified from 13 native plant species [27].
Over the last 50 years, there have been changes in the way that specimens are collected, preserved, studied and recorded in reference collections [29]. These changes include increased use of molecular analysis for identification, publication of sequence data in online resources such as GenBank, digitisation of reference collection specimens and exposure via associated online catalogues. We detected numerous misidentified Po. tridactyla specimens and we also identified and described a previously unrecorded powdery mildew species, Po. cunningtonii. Outcomes such as these highlight the need to re-evaluate important plant pathogen groups held in reference collections, in particular unculturable pathogens and complexes of cryptic species [30]. Australia implements a high level of biosecurity and quarantine measures to prevent entry of unwanted pests and pathogens which can affect the Australian horticultural industry and trade (Hyde et al. 2010). In order to provide this pest and pathogen information, the specimen records must be up to date with current taxonomic classification and names in order to provide accurate species lists of pathogens currently present in Australia [31].
Through NGS applications, re-examination of selected powdery mildew specimens from Australian plant pathogen reference collections has demonstrated that one species, Po. ampla, is the dominant powdery mildew infecting stone fruit in Australia. In addition, the phylogenetic study confirmed the presence of Po. pannosa in Australia and allowed delimitation of a new powdery mildew species within the Po. tridactyla species complex, Po. cunningtonii on Pr. persica, for which we provide a morphological and molecular description. This study resolved powdery mildew species but also identified Prunus plant hosts and reclassified previously misidentified or incompletely identified hosts of specimens held in Australian plant pathogen reference collections. The information generated in this study will be used to update BRIP, DAR and VPRI powdery mildew specimens to provide accurate data for Australian biosecurity agencies.
Supplementary Materials: The following are available online at https://www.mdpi.com/2309-6 08X/7/3/171/s1, Figure S1: Summary of variable sites for Podosphaera ampla and sister taxa Po. prunina and Po. prunigena, Figure S2: Maximum likelihood phylogenetic analysis of nuclear rDNA ITS1-5.8S-ITS2 sequences for the Po. tridactyla species complex and closely related species, Figure  S3: Maximum likelihood phylogenetic analysis of nuclear rDNA 28S sequences for the Po. tridactyla species complex and closely related species, Figure S4: Summary of variable sites for Po. cunningtonii sequences compared to sister taxa Podosphaera pruni-avium and Po. pruni-japonicae. phylogenetic analyses. I.G.P. provided the morphological data and taxonomic drawings for the new species description. Data analyses and interpretation were conducted by R.L.S. with assistance and advice from all authors. R.L.S. wrote the manuscript, which was reviewed by T.W.M. and J.E. to shape the final version. All authors have read and agreed to the published version of the manuscript.

Funding:
The research received no external funding.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The data sets generated and analysed during the current study are available from the corresponding author on reasonable request.