Morphology Characterization, Molecular Phylogeny, and Pathogenicity of Diaporthe passifloricola on Citrus reticulata cv. Nanfengmiju in Jiangxi Province, China

The Nanfengmiju (Citrus reticulata cv. Nanfengmiju), a high-quality local variety of mandarin, is one of the major fruit crops in Jiangxi Province, China. Citrus melanose and stem-end rot, two common fungal diseases of Nanfengmiju, are both caused by Diaporthe spp. (syn. Phomopsis spp.). Identification of the Diaporthe species is essential for epidemiological studies, quarantine measures, and management of diseases caused by these fungi. Melanose disease was observed on Nanfengmiju fruit in Jiangxi Province of China in 2016. Based on morphological characterization and multi-locus phylogenetic analyses, three out of 39 isolates from diseased samples were identified as D. passifloricola. Since these three isolates did not cause melanose on citrus fruit in the pathogenicity tests, they were presumed to be endophytic fungi present in the diseased tissues. However, our results indicate that D. passifloricola may persist as a symptom-less endophyte in the peel of citrus fruit, yet it may cause stem-end if it invades the stem end during fruit storage. To the best of our knowledge, this is the first report of D. passifloricola as the causal agent of the stem-end rot disease in Citrus reticulata cv. Nanfengmiju.


Introduction
As the earliest citrus producer in the world, China has over 4000 years of history of citrus cultivation. The citrus industry of China covers more than 20 provinces [1]. Recently, the cultivation area reached 2.5 million ha, and the production was about 38 million tons [2]. Melanose, one of the most common fungal diseases of citrus worldwide [3,4], generally occurs in many citrus-growing regions of China, such as Chongqing, Fujian, Guangdong, Guangxi, Hunan, Jiangxi, Shaanxi, Shanghai, Zhejiang, and so on [5][6][7]. All commercial citrus varieties are susceptible to melanose. Typical symptoms of melanose disease are small, discrete, sunken spots with a yellowish, reddish-brown to black color. Symptoms begin as tiny pustular lesions, then, pustular lesions disappear and become hardened gummed areas with a sandpaper-like surface [3,8,9]. Diaporthe spp. (syn. Phomopsis) are the causal agents of melanose and can also cause stem-end rots on fruit during the storage period. Since 95% of citrus is consumed as fresh fruit in China, melanose and stem-end rots diseases reduce the economic value of this crop seriously.
At present, Diaporthe citri is the only known causal agent of citrus melanose disease in the world. The species was first found as the causal agent of stem-end rot of citrus fruit in Florida, USA [10]. After that, D. citri was also associated with melanose of citrus fruit, leaves, and shoots and gummosis of perennial branches worldwide [11][12][13][14]. All Citrus

Pathogenicity Test
In pathogenicity tests, non-wounded Nanfengmiju fruit were used to test the ability of three isolates to cause citrus melanose and stem-end rot diseases. At 15 days after inducing melanose symptom, three isolates of NFIF-3-11, NFIF-3-19, and NFIF-3-21 did not

Pathogenicity Test
In pathogenicity tests, non-wounded Nanfengmiju fruit were used to test the ability of three isolates to cause citrus melanose and stem-end rot diseases. At 15 days after inducing melanose symptom, three isolates of NFIF-3-11, NFIF-3-19, and NFIF-3-21 did not cause any symptoms, while the positive control D. citri strain caused typical reddish-brown to black lesion spots symptoms ( Figure 3B). On the contrary, all the fruit inoculated with conidial suspension of isolates NFIF-3-11, NFIF-3-19, and NFIF-3-21, as well as positive control fruit inoculated with D. citri strain showed typical rot symptoms at 7 days after inoculation. No significant symptom was observed on negative control fruit inoculated with sterile water ( Figure 3C). Re-isolation was performed following Koch's postulation method. The strains were re-isolated from the experimentally inoculated fruit with stem-end rot symptoms. The identity of the re-isolated strains was confirmed by amplification and sequencing of ITS, TUB, TEF, HIS, and CAL molecular markers.

Phylogenetic Analyses
For preliminary identification, the MegaBlast search was performed for ITS region of three isolates in NCBI's GenBank nucleotide database. All three isolates (NFIF-3-11, NFIF-3-19, and NFIF-3-21) showed 100% identity to Diaporthe ueckerae (KY565426) and Multi-locus phylogenetic analyses were carried out based on the sequences of ITS, TUB, TEF, HIS, and CAL. To verify if these five loci were congruent and could be combined together, single locus analysis was also performed for each locus. The results indicated that the topology of single-locus trees was congruent (Supplementary Figure  S2 Of the 3,165 analyzed characters, 1,036 characters were parsimony-informative, 431 variable characters were parsimony uninformative, and 1,698 characters were constant. Data of each region/loci were shown in Supplementary Table S1. Using the best scoring RA×ML analysis, a final optimization tree with a likelihood value of -30,716.492582 was

Phylogenetic Analyses
For preliminary identification, the MegaBlast search was performed for ITS region of three isolates in NCBI's GenBank nucleotide database. All three isolates (NFIF-3-11, NFIF-3-19, and NFIF-3-21) showed 100% identity to Diaporthe ueckerae (KY565426) and Multi-locus phylogenetic analyses were carried out based on the sequences of ITS, TUB, TEF, HIS, and CAL. To verify if these five loci were congruent and could be combined together, single locus analysis was also performed for each locus. The results indicated that the topology of single-locus trees was congruent ( Supplementary Figures S2-S6). Fifteen new sequences were generated from three isolates in this study. Of the 3165 analyzed characters, 1036 characters were parsimony-informative, 431 variable characters were parsimony uninformative, and 1698 characters were constant. Data of each region/loci were shown in Supplementary Table S1. Using the best scoring RA×ML analysis, a final optimization tree with a likelihood value of −30,716.492582 was generated. The matrix data had 1837 distinct alignment patterns in the ML analysis, with 39.30% of gaps and completely undetermined characters. Estimated base frequencies were as follows:

Discussion
Diaporthe passifloricola was identified from leaf spots on Passiflora foetida in Malaysia [39]. The colonies of this species on MEA, OA, and PDA are dirty white. Alpha conidia are aseptate, hyaline, smooth, guttulate, fusoid-ellipsoid, tapering towards both ends, apex subobtuse, base subtruncate, (5-) 6-7 (-9) × 2.5 (-3) µm. Gamma conidia are not observed. Beta conidia are spindle shaped, aseptate, smooth, hyaline, apex acutely rounded, base truncate, tapering from lower third towards apex, curved, (20-) 22-25 (-27) × 1.5 (-2) µm. In this study, the colonies of the isolates on PDA were dirty white, which are similar to those of D. passifloricola [39], D. durionigena [40], D. rosae [41], and D. ueckerae [42], while that of D. miriciae is buff [23]. Morphological characteristics of alpha (bi-guttulate) and beta conidia of three isolates are consistent with those of D. passifloricola ex-type strain (CBS 141329) [39]. The sizes of alpha and beta conidia of three isolates are larger than those of D. durionigena [40] and D. rosae [41]. The alpha conidia of D. miriciae are not described of guttulate characterized [23], and the beta conidia of D. ueckerae are not observed in a previous study [42]. Thus, morphological characteristics of the three isolates are the most consistent with those of D. passifloricola. Taking into account that morphological characteristics sometimes vary with environmental conditions, they are not always reliable to identify the isolates to species level in genus of Diaporthe [37]. Thus, further molecular identification is necessary.
The sequence of the ITS region was once used alone to identify Diaporthe species. However, there are many intraspecific variations in ITS locus of certain Diaporthe species. Sometimes the intraspecific variation is even greater than interspecific variation, which makes it difficult to identify Diaporthe species with ITS sequence alone [43,44]. Currently, multi-locus phylogenetic analyses have been applied for the identification of Diaporthe species [37,45]. Thus, although ITS sequences of all three isolates showed 100% similarity with D. ueckerae (KY565426) in this study, it was unreliable, due to many intraspecific variations in ITS regions of Diaporthe species.
The combined use of the five loci (i.e., ITS, TUB, TEF, HIS, and CAL) is shown to be the best way to generate a phylogenetic tree to determine the boundaries of Diaporthe spp. [21,31,33,37,38,45]. After preliminary identification with ITS locus, four species of D. passifloricola, D. rosae, D. ueckerae, and D. miriciae were found to have high identity to the three isolates obtained in this study. Thus, five loci of ITS, TUB, TEF, HIS, and CAL were further employed to perform phylogenetic analysis.
The taxonomy of Diaporthe is complex. Many Diaporthe spp. were classified according to different criteria, i.e., host associations, morphological characteristics [26,28,46,47], or sequences of ITS region [22,26,48]. It is suggested that only those type strains, whose identification has been widely recognized, should be accepted as references for the taxonomy of this genus [37,49,50]. Moreover, several isolates included type strains from previous publications are selected for references with phylogenetic analysis in this study. While MegaBlast search was performed for each locus on NCBI, the Diaporthe species showing the highest similarity with the sequencing of each locus of the isolates were not the type strains. Thus, the species identified by us are different from those retrieved by a single locus MegaBlast search on NCBI.
In previous studies, 15 Diaporthe spp. have been reported to be associated with citrus in China [5,6]. Of them, three species are pathogens on citrus, i.e., D. citri, D. citriasiana, and D. citrichinensis. D. citri is identified as the causal agents of melanose disease as well as stem-end rot disease. In addition to being a pathogen, D. citri is also found as an endophyte in non-symptomatic twigs and as a saprobe on dead twigs. Two species, D. citriasiana, and D. citrichinensis, can only cause stem-end rot symptom on ponkan fruit (Citrus reticulata) [5]. The other 12 Diaporthe spp. were identified as endophytes or saprobes on citrus [6]. All of these indicate that the symbiotic relationship and ecological function of Diaporthe spp. with citrus plants is complex and variable.
Endophytes are defined as all organisms inhabiting plant organs which, at some time in their lives, can colonize internal plant tissues without causing significant damage to the host [55]. So defined, endophytes may also encompass asymptomatic latent pathogens. Sometimes asymptomatic fungi can cause diseases on their host plants under certain conditions. It's reported that several Plectosphaerella spp. isolated from symptomless tomatoes and peppers can cause disease symptoms on tomato and pepper, and even basil and parsley when artificially inoculated [56,57]. Epichloë festucae is a well-known endophytic fungus of perennial ryegrass (Lolium perenne). However, a E. festucae noxA mutant is associated with severe stunting of the host as a result of hyphal hyper-branching and increased biomass [58]. Some fungal saprobes and pathogens can be isolated from rice (Oryza sativa) as endophytes [59]. In this study, since D. passifloricola isolates failed to cause melanose on citrus fruit, they are supposed to be the endophytic fungi colonizing diseased tissues with melanose symptoms. However, our results show that this species can induce stem-end rot symptoms on artificially inoculated citrus fruit. Thus, D. passifloricola could be a potential causal agent of stem-end rot disease during transportation and storage.
The disease spots of citrus melanose are formed by host hypersensitive response (HR). When the pathogens penetrate epidermal cells of the citrus, they are arrested and killed at the infection sites by hosts along with the development of melanose symptoms [60][61][62]. As a result, it is difficult to isolate pathogens in old disease spots. The disease spots were not newly formed, which might be the reason why we failed to isolate the pathogen causing melanose symptoms.

Fungal Isolation
In 2016, 10 citrus fruit of Nanfengmiju with typical symptoms of melanose were collected from a citrus orchard in Fuzhou City of Jiangxi Province ( Figure 3A). The discrete and sunken black spots were observed on the fruit surface. Pieces of small sections about 5 mm 2 from the margin of the lesion were cut off and soaked in 75% ethanol solution for 1 min. The sections were surface disinfested with 1% sodium hypochlorite solution (NaClO) for 1 min, rinsed three times with sterilized water, dried, and then incubated on PDA plates amended with 100 µg/mL streptomycin and 100 µg/mL ampicillin at 25 • C for 2 to 5 days. Hyphal tips growing from the pieces of the sample were transferred onto fresh PDA plates and incubated at 25 • C for 30 days as previous methods [7]. After sporulation, single-spore-isolation was performed as previously described [63]. All single-spore cultures were stored on half strength PDA slants in Eppendorf tubes at 4 • C, and on dried filter paper discs at −20 • C, respectively. A living culture of D. passifloricola in this study was deposited in China Center for Type Culture Collection (CCTCC), Wuhan, China.

Morphological Characterization
Sporulation was induced on PDA, MEA, CMA, OMA, and PNA. After inoculation, isolates were incubated at 25 • C with 12 h of light and 12 h of dark for 30 days. Conidia were harvested from the top of mature pycnidia. Pycnidia were picked up from pine needles with sterile toothpicks. The length and width of 30 conidia were measured with a stage micrometer under a Motic BA200 light microscope (Motic China Group Co., Ltd., Xiamen, China). The morphology of conidiomata was observed under OLYMPUS SZX16 stereo microscope (Olympus Corporation, Tokyo, Japan). Images of conidia were captured using a digital camera Nikon Eclipse 80i on a compound light microscope (Nikon Corporation, Tokyo, Japan) imaging system. Images of culture plates were captured using Cannon 600D digital camera (Cannon Inc., Tokyo, Japan). Colony and pycnidia color was investigated with a color chart according to the method of Rayner [64].

Pathogenicity Test
Pathogenicity tests were carried out on detached Nanfengmiju fruit (Citrus reticulata cv. Nanfengmiju). Non-wounded citrus fruit were washed with tap water, then surface disinfested with 75% of ethanol and rinsed with sterile water. Pycnidia with alpha conidia were induced as mentioned above and diluted to 10 6 conidia/mL with sterile water. To stimulate melanose symptoms, 300 µL of conidial suspensions was dropped on a piece of cotton, and then placed on the bottom of the fruit as previously described with a slight modification [65]. The inoculated fruit were placed in a plastic chamber with 95% relative humidity, incubated under the condition of 12 h of light and 12 h of dark at 25 • C for 15 days. Since Diaporthe spp. were the causal agents of both melanose and stem-end rot diseases on citrus fruit, their ability to cause stem-end rot symptom was also determined. The stems of citrus fruit were removed carefully, and 10 µl of conidial suspension (10 6 conidia/mL) of each strain was inoculated onto stem ends as previously described [5]. Then, the inoculated fruit were placed in a plastic chamber with wet towel tissues at the bottom. The chamber was wrapped with plastic film to maintain 95% relative humidity and incubated at 25 • C in the dark for 7 days. In all the pathogenicity tests, the conidial suspension (10 6 conidia/mL) of D. citri strain NFHF-8-4 [7] and sterile distilled water were used as positive and negative controls, respectively. Symptoms on fruit were observed. Four fruit were inoculated for each strain, and the experiments were repeated at least twice.
To authenticate the causal agent, tissue pieces from the margin of lesions on the experimentally inoculated and diseased fruit were placed on PDA to re-isolate the fungus. Molecular identification of the isolate was performed using the sequence of ITS, TUB, TEF, HIS, and CAL loci as mentioned below.

Phylogenetic Analyses
The preliminary identifications of the isolates obtained in this study were determined using newly generated ITS sequences with all available type-derived sequences listed in previous studies [6,24,25,37,51]. Based on the result of preliminary identification, Diaporthe species with the closest genetic distance to the isolates in this study were selected. Sequences (ITS, TUB, TEF, HIS, and CAL) of them were downloaded from NCBI's GenBank nucleotide database (www.ncbi.nlm.nih.gov). All sequences used in this study are listed in Table 1, including 15 sequences of three new isolates. The reference isolates were selected from ex-type, ex-epitype, and holotype cultures. Five-locus phylogenetic analyses were conducted to identify isolates to species level according to previous studies [21,30,37]. Sequences of five loci (ITS, TUB, TEF, HIS, and CAL) were assembled. Alignments of assembled sequences were performed with L-INS-i iterative refinement method by MAFFT alignment, a version available online [71], and manual adjustment was conducted where it was necessary by BioEdit v.7.2.5 [72]. ML trees were generated with 1,000 replicates using RA×ML-HPC BlackBox v.8.2.10 [73], which was available on the CIPRES Science Gateway v.3.3 Web Portal [74]. The RAxML software selected general time reversible model of evolution including estimation of invariable sites (GTRGAMMA+I). MP analyses were carried out with 1,000 replicates using Phylogenetic Analyses Using Parsimony (PAUP*) v.4.0b10 [75], with tree bisection and reconnection (TBR) branch-swapping algorithm. All characters were weighted equally, and the alignment gaps were treated as missing characters. Descriptive tree statistics including TL, CI, RI, RC, and HI were calculated for parsimony analyses. MrModeltest v.2.3 [76] was used to perform statistical selection of the best-fit model of nucleotide substitution and the corrected Akaike information criterion (AIC) determined above was incorporated into evolutionary models in the analysis (Supplementary Table S1). BI analysis was performed by using MrBayes v.3.2.2, with Markov Chain Monte Carlo (MCMC) algorithm. Four simultaneous of MCMC chains were run for 20,000,000th generations, and trees were sampled frequency every 100th generations, resulting in a total of 20,000 trees, and started from a random tree topology. The calculation of BI analyses was stopped when the average standard deviation of split frequencies fell below 0.01. The first 10% of trees were discarded as burn-in phase of analysis, and the remaining 180,000 trees were summarized to calculate the PP in the majority rule consensus tree. Phylogenetic analyses and full alignment of datasets were submitted to TreeBASE (www.treebase.org) with the study ID: 27334.

Conclusions
Our results indicate that D. passifloricola, may occur as an asymptomatic endophyte in the peel of citrus fruit. If is manages to invade the fruit stalk, however, it may induce typical stem-end rot symptoms during transportation and storage. To the best of our knowledge, this is the first time D. passifloricola has been isolated from Citrus reticulata cv. Nanfengmiju in China and identified as a causal agent of stem-end rot disease in this crop.