Molecular and Morphological Assessment of Septoria Species Associated with Ornamental Plants in Yunnan Province, China

The Karst landform is the main geographic characteristic in South China. Such areas are rich in vegetation and especially suitable for growth of shrubs and herbaceous plants. In this study, 11 Septoria strains were obtained from different plants’ leaves collected in the Kunming Botanical Garden, Yunnan Province, China. Based on single-gene and multi-gene analyses of five gene loci (tef1, rpb2, tub2, ITS, and LSU) and four gene regions (without LSU), these strains were found to belong to three independent phylogenetic lineages representing five species, including four novel taxa, and one new record for China. Five single gene trees were also provided to evaluate the effectiveness of each gene for discriminating the species, as a result of which tub2 was found to have the most suitable DNA barcode for rapid identification. Morphological descriptions, illustrations, and comparisons are provided for a more comprehensive assessment. Genealogical Concordance Phylogenetic Species Recognition (GCPSR) with a pairwise homoplasy index (PHI) test was used to evaluate the conclusions of the phylogenetic analyses.


Introduction
Septoria Sacc., established by Saccardo in 1884, belongs to the Mycosphaerellaceae family of fungi and accommodates around 1000 species [1,2], although only 200 species have been confirmed by molecular data [2]. Many of these species cause leaf spot diseases of numerous cultivated and wild plants [3]. According to its morphology at the primary generic level, Septoria includes coelomycetous asexual morphs, which produce pycnidial conidiomata having holoblastic, hyaline, smooth, filiform-to-cylindrical multi-septate conidia [4][5][6][7][8][9]. On the basis of a polyphasic approach to taxon delimitation, Verkley et al. [3] pointed out that septoria-like fungi preserved in CBS were in fact distributed over three main clades and introduced a novel genus: Caryophylloseptoria Verkley, Quaedvlieg and Crous. Quaedvlieg et al. [10] re-defined Septoria as having pycnidial to acervular conidiomata and hyaline conidiophores that give rise to conidiogenous cells that proliferate both sympodially and percurrently to form hyaline, filiform conidia with transverse eusepta. Crous et al. [11] introduced Acervuloseptoria on account of its black, erumpent conidiomata, and the old name Septoria capensis G. Winter was transferred to this genus [12]. More DNA sequence data are necessary to support the morphological characters in this species identification [10].
In this study, 11 Septoria strains were obtained from different ornamental plants in a South China Karst region. Morphological comparisons, phylogenetic analyses based on five gene loci, DNA base-pair differences, and GCPSR evaluation confirmed that they formed three phylogenetic lineages representing five Septoria species comprising four novel species and one new Chinese record.

Fungus Collection and Isolation
The isolates included in this study were collected from the Kunming Botanical Garden, Yunnan Province, China, in 2018. Pure cultures were obtained by single-spore isolations following the methods of surface sterilization and incubation of specimens [13]. After 24 h of incubation, germinated conidia were transferred to the new potato-dextrose agar (PDA) medium and incubated at 25 • C. The holotype specimens were deposited in the Herbarium of the Department of Plant Pathology, Agricultural College, Guizhou University (HGUP). The type cultures were deposited in the Culture Collection at the Department of Plant Pathology, Agriculture College, Guizhou University, China (GUCC), and the Mae Fah Luang University Culture Collection (MFLUCC) in Thailand (Table 1). Ex-type isolates are labeled with " T ".

Morphological Studies
Morphological characters were recorded from cultures that had been incubated for 2 to 3 weeks. For light microscopy, the relevant structures were mounted in Shear's liquid, distilled water or lactic acid and examined with an Olympus BX53 microscope. Measurements of 30 conidia and other structures were made at a magnification of 1000× [14]. Taxonomic information of the new taxa was submitted to the MycoBank database (www.mycobank.org, accessed on 24 March 2021).

DNA Extraction, Amplification (PCR), and Sequencing
Methods outlined in [15] were followed for DNA extraction, amplification (PCR), sequencing, and phylogenetic analysis. Fresh fungal mycelia of strains were harvested using a sterile scalpel, and the genomic DNA was isolated using A BIOMIGA Fungus Genomic DNA Extraction Kit (GD2416) according to the manufacturer's protocol. The DNA was amplified in a 25 µL reaction volume containing 2.5 µL 10× PCR buffer, 1 µL of each primer (10 µM), 1 µL template DNA, 0.25 µL Taq DNA polymerase (Promega, Madison, WI, USA), and 18.5 µL ddH 2 O. Five gene regions-loci β-tubulin (tub2), internal transcribed spacer (ITS), Translation elongation factor 1-alpha (tef1), 28S nrDNA (LSU), and RNA polymerase II second largest subunit (rpb2)-were targeted for Polymerase Chain Reaction (PCR) amplification and subsequent sequencing. The primers used and amplification conditions of the genes are listed in Table 2. The DNA sequences were submitted to GenBank and their accession numbers are provided in Table 1. The generated sequences for each locus and the reference sequences of ex-type or representative sequences of Septoria species downloaded from GenBank (Table 1) were aligned with the online version of MAFFT v. 7.307 [16,17].

Phylogenetic Analyses
The alignments were checked and manually improved where necessary using MEGA v. 5 [27]. Phylogenetic analyses were performed by maximum parsimony (MP), maximum likelihood (ML), and Bayesian methods for individual and combined locus datasets. Ambiguous regions were excluded from the analyses and gaps were treated as missing data. Maximum parsimony analysis was performed in PAUP v. 4.0b10 [28] using the heuristic search option with 100 random taxon additions and tree bisection and re-connection (TBR) as the branch-swapping algorithm with Maxtrees = 5000. Branches of zero length were collapsed and all multiple, and equally most parsimonious trees were saved. The robustness of the trees obtained was evaluated by 1000 bootstrap replications [29]. Other measures calculated included tree length (TL), consistency index (CI), retention index (RI), and rescaled consistency index (RC).
The resulting PHYLIP file was used to generate the ML tree on the CIPRES Science Gateway [30] using RAxML-HPC2 black box with 1000 bootstrap replicates and GTRGAMMA as the nucleotide substitution model. Bayesian analyses were launched with random starting trees for 10,000,000 generations. The heat parameter was set at 0.15 and trees were saved every 1000 generations until the average standard deviation of split frequencies reached 0.01 (stop value). Burn-in was set to 25% after which the likelihood values were considered to be stationary. All resulting trees were visualized with FigTree v. 1.4.3 (Institute of Evolutionary Biology, University of Edinburgh, UK) [31].

Genealogical Concordance Phylogenetic Species Recognition Analysis
The Genealogical Concordance Phylogenetic Species Recognition (GCPSR) concept with a pairwise homoplasy index (PHI) test was used to analyze the new species, their species boundaries, and their most closely related taxa as described by Quaedvlieg et al. [32]. The recombination level within phylogenetically closely related species was determined with the PHI test performed using SplitsTree4 [33,34]. The concatenated datasets (tef1, rpb2, tub2, ITS, and LSU) were used. The relationships between different taxa were visualized in splits graphs with both the Log-Det transformation and splits decomposition options. A pairwise homoplasy index below a 0.05 threshold (Fw < 0.05) indicated the presence of significant recombination in the dataset.

Phylogenetic Analyses
Eleven Septoria strains isolated from different plant hosts were sequenced. PCR products of 450-536 bp (tef1), 440-453 bp (tub2), 458-524 bp (ITS), 799-863 bp (LSU), and 718-1083 bp (rpb2) were obtained. By alignment with the single-gene region and then in combination in the order of tef1, rpb2, tub2, ITS, and LSU with Cercospora beticola (CBS 124.31), 2434 characters were obtained: tef1, 1-479; rpb2, 480-824; tub2, 825-1149; ITS, 1159-163; and LSU, 1636-2434. Among these characters, 1672 were constant, while 195 variable characters were parsimony-uninformative and 567 were parsimony informative. The parameters of the MP phylogenetic trees are shown in Table 3, and the procedure yielded a single most parsimonious tree ( Figure 1). Similar topologies were obtained by MP, ML, and Bayesian methods. In the Septoria phylogenetic tree (Figure 1), all Septoria isolates were grouped together, but only the BI support was high (BPP = 1), while the three major clades received greater statistical support (  We also compared the DNA base-pair differences in five different loci between our strains and related species (Supplementary Table S1). This revealed that the LSU gene region was too conserved for species-level identification, and the ITS had little value, but tef1, tub2, and rpb2 provided more than 80% of the DNA base-pair differences (Supplementary Table S1). We also built a phylogenetic tree based on four loci, excluding the LSU region (Figure 2), using the parameters for MP analysis in Table 3. The topology showed highly similar placements of our strains in the Septoria in Figure 1; however, in Figure 2 only two branches were formed and all members of Branch 3 were integrated with Branch 1. To evaluate the distinctive effectiveness of different DNA markers, five single gene trees were constructed (Supplementary Figures S1-S5) and all MP parameters were as indicated in Table 3. Through comparison, we found that only tub2 and tef1 included more parsimonious characters (50.7% and 49.2%), and the sequence of tub2 was shorter than that of tef1. Moreover, the topology originating from the tub2 gene region was more similar to Figure 2. Maximum Parsimony (MP) topology of Septoria generated from a combination of tef1, rpb2, tub2, ITS, and LSU sequences. Cercospora beticola (CBS 124.31) was used as outgroup taxon. MP and ML above 50% and BPP above 0.90 were placed close to topological nodes and separated by "/", otherwise were labeled with "-".

Figure 2.
Maximum parsimony (MP) topology of Septoria generated from a combination of tef1, rpb2, tub2, and ITS sequences. Cercospora beticola (CBS 124.31) was used as an outgroup taxon. MP and ML above 50% and BPP above 0.90 were placed close to topological nodes and separated by "/"; otherwise, they were labeled with "-".

Genealogical Concordance Phylogenetic Species Recognition
In order to determine evolutionary independence, the GCPSR concept was applied to the GUCC 2164.2, GUCC 2131.4, GUCC 2131.1, and related taxa S. chrysanthemella (CBS 128716), S. exotica (CBS 163.78), and S. posoniensis (CBS 128645). A pairwise homoplasy index (PHI or Fw) less than 0.05 provided evidence of the presence of significant recombination within a dataset. According to the GCPSR analysis, our dataset showed PHI of 0.116, indicating no significant genetic recombination among our strains and related taxa. Hence, it was concluded that these taxa were significantly different from each other.
For GUCC 2164.3 and GUCC 2127.4 and related species S. coprosmae (CBS 113391) and S. verbenae (CBS 113438), the pairwise homoplasy index (PHI or Fw) was 1.173 × 10 −8 , which provided evidence for the presence of significant recombination within a dataset. The four strains could belong to a single species.  Etymology: The name refers to the plant host, from which the fungus was collected.
(2) Septoria pileicola Y.Y. An & Yong Wang bis sp. nov. (Figure 4) MycoBank MB 839126 Etymology: The name refers to the plant host from which the fungus was collected. Description in vitro: Colonies: on PDA up to 10-15 mm diameter, with an even, glabrous, colourless margin in 2 weeks. Mycelium: greenish grey to dark slate-blue, immersed, throughout covered by well-developed, tufty whitish-grey aerial mycelium that later attains a reddish haze; reverse black, but margin paler; in the central part of the colony numerous pycnidia develop, releasing pale vinaceous to rosy-buff conidial ball. Conidiomata: pycnidial, epiphyllous but sometimes also visible from the underside of the lesion, one to a few in each leaf spot, subglobose to globose, brown to black, usually fully immersed, 80-120 µm diam. Ostiolum: central, initially circular and 15-30 µm wide, later becoming more irregular and up to 45 µm wide, surrounding cells concolorous to pale brown. Conidiogenous cells: hyaline, discrete, doliiform, or narrowly to broadly ampulliform, holoblastic, with a relatively narrow elongated neck, proliferating percurrently several times with distinct annellations, often also sympodially after a few percurrent proliferations, 5.5-12 × 2-3.5 µm. Conidia: cylindrical or filiform-cylindrical, straight to slightly curved, narrowly to broadly rounded at the apex, narrowing slightly or more distinctly to a truncate base, (0-)1-2-septate, not or slightly constricted around the septa, hyaline, contents with a few minute oil-droplets and granular material in each cell in the rehydrated state, 8.5-30 × 1.5-3.5 µm. Sexual morph unknown.  (Figure 1), but formed a subclade with S. sanguisorbigena. After morphological comparisons, we found that Septoria pileicola can be distinguished from S. sanguisorbigena by its wider conidia, and from S. posoniensis and S. exotica by its shorter conidia with obviously fewer septa [36,37]. For S. chrysanthemella and S. longipes, the species had apparently shorter conidia [35]. The two strains of Septoria pileicola had nearly the identical sequences (only one different ITS base pair); however, the tub2 gene provided enough base distinction to separate it from related species (Supplementary Table S1) according to the guidelines of Jeewon and Hyde [38]. The PHI value was 0.116 (>0.05), indicating no significant genetic recombination among S. pileicola, S. sanguisorbigena, S. chrysanthemella, S. exotica, and S. posoniensis. Thus, they should belong to different species [39].
(4) Septoria dispori Y.Y. An & Yong Wang bis sp. nov. (Figure 6) MycoBank MB 839128 Etymology: The name refers to the plant host from which the fungus was collected. Description in vitro: Colonies: on PDA 2.0-3.5 mm diameter, with an even to slightly ruffled, glabrous, dull yellow margin in 2 weeks, spreading, remaining almost plane, immersed mycelium yellowish brown to brown; aerial mycelium well-developed, goose feather flocculent on the surface of the colony; numerous conidiomatal initials developing at the surface, mature ones releasing cirrhi of conidia that first are milky white, later salmon, sometimes merging to form slimy masses covering areas of the colony surface. Conidiogenous cells: hyaline, broadly or elongated ampulliform, normally with a distinct neck, hyaline, holoblastic, proliferating percurrently, annellations indistinct, 10-15 ×1.5-2.5 µm. Conidia: cylindrical to filiform-cylindrical, slightly to strongly curved, rarely somewhat flexuous, narrowly rounded to pointed at the apex, attenuated gradually or more abruptly towards a narrowly truncate base, 3-5-8-septate, later with secondary septa dividing the cells, sometimes breaking up into smaller fragments in the cirrhus, not or slightly constricted around the septa, hyaline, 14-41.5 × 1.5-2.5 µm. Sexual morph unknown. Note: Four strains (GUCC 2127.1, GUCC 2127.4, GUCC 2164.3, and GUCC 2164.4) of Septoria dispori clustered together with high statistical support (MP: 95%, ML: 100%, BPP: 1.00) adjacent to S. coprosmae and S. verbenae. Thus, we consider these four strains to be a single species. Septoria coprosmae produced spermatogonia of an Asteromella-state, but this species did not [40]. Conidia of S. verbenae possessed fewer septa than those of Septoria dispori [41]. GUCC 2127.4 showed some phylogenetic distance from the other three strains, however DNA base comparison (Supplementary Table S1) revealed only 11 bases that had tub2 differences. The PHT test confirmed significant recombination between strains GUCC 2164. 4 and GUCC 2164.3 and they were morphologically similar. Thus by combining the above evidence, we established the four strains as a new taxon.  Table S1), revealed that sequences of strain GUCC 2127.3 were identical to the ex-type strain of S. protearum (CBS 778.97) in four gene regions. Conidial shape and size range of S. protearum (12-22 × 1.5-2 µm) were similar to the present strain [42]. The number of conidial septa of the two strains was also the same (0-4 septa). Thus, we conclude that GUCC 2127.3 is S. protearum.

Discussion
Verkley et al. [3] pointed out that for the identification of the Septoria species, morphological description must be integrated with sequences analyses. Quaedvlieg et al. [10] treated species in Septoria within a modern taxonomic framework and pointed out that Septoria spp. formed a well-defined phylogenetic clade. Regarding morphology, the species concept was to produce pycnidial, ostiolate conidiomata; conidiophores reduced to conidiogenous cells that proliferate sympodially; and hyaline, filiform conidia with transverse eusepta that fit the original concept of [4]. We followed this system and applied morphological and phylogenetic approaches to the present study. After comparing the topologies of five single-gene and two multi-gene trees (Figures 1 and 2 and Supplementary Figures S1-S5), we showed that Septoria forms two branches (Branch 1 and Branch 2), mainly because only the phylogenetic trees based on the LSU region and five DNA fragments (including LSU) supported three branches, whereas the conserved LSU sequences included the least parsimonious characters (31/799) ( Table 3). In morphology, all species in Branch 3 produced filiform or fusiform, sub-straight to slightly curved conidia mainly with 3 septa, which was not a unique characteristic. Thus, we proposed exclusion of the LSU region for multi-gene analyses of Septoria at the species level, but always as the primary DNA barcode with more parsimonious characters (43/486), the ITS fragment was conserved in the present phylogenetic analysis.
The S. protearum complex accommodated eight members: S. citri, S. citicola, S. chamaecisti, S. gerberae, S. hederae, S. lobelia, S. limonum, and S. protearum, according to Verkley et al. [3]. Apart from S. protearum, the other species were old names without ex-type cultures, and thus no sequences were available. The base comparison of DNA sequences originated from Verkley et al. [3], who indicated that among these eight species there were only approximately 10 base-pair differences on the rpb2 fragment (434 characters) of S. gerberae, S. hederae, and S. lobelia compared to the other five species, while for the other four gene regions, their sequences were nearly identical (≤1 base difference) (Supplementary Table  S1). On the other hand, in the literature these seven species are depicted only by simple descriptions often without drawings or photographs, which does not strongly support them as different taxa. Comparing with the sequences from Verkley et al. [3] and in the absence of type materials, we were more willing to believe that they belonged to the same species, S. protearum.
Our 11 strains isolated from Disporum bodinieri, Pilea cadierei, Sanguisorba officinalis all from the Botanical Garden of Kunming county represented five Septoria species and included four novel species supported by morphology and phylogeny. Septoria sanguisorbigena was obtained from two plant hosts (Sanguisorba officinalis and P. cadierei), and S. dispori was also on two hosts (D. bodinieri and Sanguisorba officinalis). Septoria pileicola and S. longipes were only discovered on one host (P. cadierei). Our S. protearum strain was on D. bodinieri. Verkley et al. [3] recalled that trans-family host jumping must be a major force driving the evolution of Septoria. Our results support this hypothesis as we found the same species on different hosts. However, our findings revealed that the Septoria species did not show any host specialization, which differs from the view of Verkley et al. [3].

Conclusions
In this study, our 11 Septoria strains represented five species including four novel taxa, and one new record for China by morphological comparison and multi-gene analyses. The Septoria species are pathogens often causing leaf spot diseases of many plant hosts worldwide [10]. Based on previous studies, relatively sufficient reference sequences are available for rapid identification of Septoria pathogens. By comparing the parsimoniousinformative characters of different DNA fragments (Supplementary Table S1), we showed that either tef1 or tub2 is suitable as a secondary DNA barcode, and that the latter was more discriminating than the former. Moreover, the DNA fragment of tub2 (≈300 bp) was shorter than that of tef1 (≈450 bp) with a high PCR amplification success rate. Consequently, a standardized approach including morphological characters and phylogenetic analysis is needed for the correct and precise identification of Septoria isolates.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/jof7060483/s1. Figure S1: The phylogenetic tree based on ITS region. Figure S2: The phylogenetic tree based on LSU region. Figure S3: The phylogenetic tree based on rpb2 region. Figure  S4: The phylogenetic tree based on tef1 region. Figure S5: The phylogenetic tree based on tub2 region. Cercospora beticola (CBS 124.31) was used as outgroup taxon. MP and ML above 50% and BPP above 0.90 were placed close to topological nodes and separated by "/", otherwise were labeled with "-". Taxa from this study are highlighted in green. Table S1: DNA base difference between our Septoria strains and related species.   Table S1.