Colletotrichum species associated with mango in southern China

Mango (Mangifera indica L.) is an economically significant fruit crop in provinces of southern China including Hainan, Yunnan, Sichuan, Guizhou, Guangdong and Fujian. The objective of this study was to examine the diversity of Colletotrichum species infecting mango cultivars in major growing areas in China, using morphological and molecular techniques together with pathogenicity tests on detached leaves and fruits. Over 200 Colletotrichum isolates were obtained across all mango orchards investigated, and 128 of them were selected for sequencing and analyses of actin (ACT), chitin synthase (CHS-1), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), the internal transcribed spacer (ITS) region, β-tubulin (TUB2) genomic regions. Our results showed that the most common fungal isolates associated with mango in southern China involved 13 species: Colletotrichum asianum, C. cliviicola, C. cordylinicola, C. endophytica, C. fructicola, C. gigasporum, C. gloeosporioides, C. karstii, C. liaoningense, C. musae, C. scovillei, C. siamense and C. tropicale. The dominant species were C. asianum and C. siamense each accounting for 30%, and C. fructicola for 25%. Only C. asianum, C. fructicola, C. scovillei and C. siamense have previously been reported on mango, while the other nine Colletotrichum species listed above were first reports associated with mango in China. From this study, five Colletotrichum species, namely C. cordylinicola, C. endophytica, C. gigasporum, C. liaoningense and C. musae were the first report on mango worldwide. Pathogenicity tests revealed that all 13 species caused symptoms on artificially wounded mango fruit and leaves (cv. Tainong). There was no obvious relationship between aggressiveness and the geographic origin of the isolates. These findings will help in mango disease management and future disease resistance breeding.

morphological data is now recommended as an alternative method for accurate species identification in the genus Colletotrichum 6,7,9 . Colletotrichum species complexes such as C. acutatum, C. amagnum, C. aorchidearum, C. boninense, C. dracaenophilum and C. gloeosporioides have been separated into many different species using morphological and molecular techniques [9][10][11][12][13] . However, in China the identification of Colletotrichum species on mango has relied mainly on conidial morphology and ITS sequences 14 , and the combined data were still limited and insufficient to distinguish closely related taxa in the six species complexes mentioned above. Previous research regarding Colletotrichum species on mango were mostly samples from small and fragmented provincial areas, with poor representation of the actual species diversity in the vast mango growing regions in China.
In our previous study, three Colletotrichum species causing mango anthracnose in Guangxi province in China (C. asianum, C. fructicola and C. siamense) were reported 15 . However, the growing season and the harvesting time of mango in seven provinces growing mangoes vary in environment, climatic conditions and varieties grown, which may lead to variations in composition and distribution of causal agents of mango anthracnose.
Genetic diversity is often related to evolutionary potential, and an ability to adapt to variable environmental conditions 16 . Based on our current understanding of the gene-for-gene hypothesis, resistance genes lose their effectiveness when faced with complex and evolving pathogen populations. An understanding of the population structure and species distribution of pathogens can provide insights into optimal breeding strategies for durable resistance to anthracnose in mangoes and improved cultural controls.
The objectives of this study were as follows: (1) to identify Colletotrichum species associated with mango anthracnose symptoms in six provinces of China using cultural characteristics such as conidial and appressorial morphology as well as multi locus sequencing data for phylogenetic analyses; (2) to test the aggressiveness of isolates from different geographic locations in the major mango growing areas in mainland China; and (3) to examine the variation of isolates within and between regions.

Results fungal isolations.
In 2016 and 2017, all the leaves, fruits and branches of mango with anthracnose symptoms collected from the six mango-growing provinces in China were used for fungal isolation. Based on morphology observation and ITS sequences, a total of 205 Colletotrichum isolates were obtained from all mango growing regions investigated, and 128 representative isolates were selected for further study (one isolate from each orchard, and more than one for some big orchards; 34 from Yunnan, 23 from Hainan, 18 from Guizhou, 13 from Sichuan, 16 from Guangdong and 24 from Fujian province). The selected isolates are listed in Supplementary  Table S1. Isolated species in other genera include Alternaria, Phomopsis, Botryosphaeria, and Pestalotia and were retained for further study. phylogenetic analyses. These isolates from different areas were used for sequencing and analyses of ACT, CHS-1, GAPDH and TUB2 loci. All sequences were first compared with the NR database from NCBI-GenBank using BLASTn, and then submitted to GenBank (see Supplementary Table S1). For phylogenetic analysis, sequences of extype or ex-epitype strains of Colletotrichum species (Supplementary Table S2) were selected. According to the ILD test, partitions of the data into ACT, CHS-1, GAPDH, ITS and TUB2 were homogeneous, and the data sets were combined for Maximum likelihood (ML) and Bayesian analyses. For the Bayesian analyses, a JC model was selected for ACT, a GTR + I + G model for GAPDH, ITS and CHS-1, a HKY + G model for TUB2 and incorporated in the analysis.
Concatenated sequences from the five loci (ACT, TUB2, CHS-1, GAPDH and ITS) were generated for each of the 128 Colletotrichum isolates, plus ex-type or type strains downloaded from GenBank to generate a phylogenetic tree (Fig. 1). As the topology of the Bayesian analysis of the combined data set was nearly identical to that of the ML consensus tree, only the Bayesian tree is shown with bootstrap (from ML analysis) on the left and posterior probability (from Bayesian analysis) on the right of corresponding nodes (Fig. 1).
Morphological observations. The colonies of the Colletotrichum isolates were white initially, later turning pale gray, with sparse aerial mycelium, hairy, sometimes producing abundant orange (#FF8040) pionnotes (Fig. 2). There was high variation in mycelial growth rate among the 128 Colletotrichum isolates (Supplementary  Table S3). The mean conidial length and width ranged from 11.5 μm to 25.5 μm and 4.0 μm to 8.2 μm. The conidial morphology of Colletotrichum species on mango differed significantly between and within species. For examples, C. fructicola (GZ15-1) had profuse sporulation, C. gigasporum (HN42-2) produced the largest spores, and the shortest spores were generated by C. scovillei (YN51-1).   www.nature.com/scientificreports www.nature.com/scientificreports/ Conidial appressoria were grey cloud (#B6B6B4) to black (#000000), nearly elliptical or irregular. The length of appressoria for each species varied from 4.9 μm to 14.9 μm and the width of appressoria varied from 3.6 μm and 10.6 μm ( Table 1). The mean length and width calculated for each isolate ranged from 6.4 μm to 12.4 μm. Mycelial appressoria were pale brown to dark brown, rod-shaped, occasionally irregular. The length of appressoria for each species varied from 5.6 μm to 20.7 μm and the width of appressoria varied from 3.5 μm to 15.4 μm.
Provincial geographic distribution of Colletotrichum species on mango in china. The diversity of species and geographical distribution for various Colletotrichum species associated with mango anthracnose across different provinces in China are shown in Fig. 3. Hainan and Yunnan had the most abundant and diverse Colletotrichum species on mango. Nine Colletotrichum species (C. asianum, C. cordylinicola, C. endophytica, C. fructicola, C. gigasporum, C. karstii, C. scovillei, C. siamense, and C. tropicale) were recovered from mango with anthracnose symptoms in Hainan province while eight species (C. asianum, C. cliviicola, C. endophytica, C. fructicola, C. karstii, C. liaoningense, C. scovillei and C. siamense) were obtained from mango in Yunnan. Four species (C. fructicola, C. gloeosporioides, C. karstii and C. musae) were associated with mango anthracnose in Guizhou province. Only two Colletotrichum species (C. asianum and C. siamense) were isolated in Sichuan and Guangdong provinces. C. asianum and C. fructicola were found in Fujian province. As a whole, C. asianum, C. fructicola, C. siamense were the dominant species widely spread across the major mango growing areas in southern China, accounting for 84.4% of all isolates. C. endophytica, C. gigasporum, C. gloeosporioides and C. liaoningense were only found on wild mango trees (Supplementary Table S1).
The aggressiveness of the tested isolates on mango leaves was also significantly different, and the diameters of disease lesions varied from 0.60 cm to 4.34 cm (Supplementary Table S3). The aggressiveness of the tested strains in mango differed considerably. According to the diameter size of the disease lesions, the aggressiveness scale can be classified as strongly virulent (>2 cm), moderately virulent (>1.0 to <2 cm) and weakly virulent (<1.0 cm). Using this scale, 84 isolates (65%) were moderately virulent, 38 isolates (30%) were strongly virulent, 6 isolates (5%) were weakly virulent. The least virulent strain was C. siamense YN56-1-1, while the most virulent strain was C. siamense SC2-1 (Supplementary Table S3).

Discussion
In the past, common identification systems in Colletotrichum included traditional morphological taxonomy, molecular classification based on ITS sequences, presence of secondary metabolites and some other traits. These previously used identification systems alone or in combination were unable to fully resolve the species identity of individual Colletotrichum isolates 6 . Additional characteristics such as other aspects of morphology, and multi-gene phylogenetic analyses are needed to identify Colletotrichum species 6 .
Based on morphology and a concatenated five-gene phylogenetic analysis, 128 Colletotrichum isolates on mango from southern China were separated into 13 species: C. asianum, C. cliviicola, C. cordylinicola, C. endophytica, C. fructicola, C. gigasporum, C. gloeosporioides, C. karstii, C. liaoningense, C. musae, C. scovillei, C. siamense, and C. tropicale. Four species complexes (C. acutatum, C. boninense, C. gigasporum and C. gloeosporioides) were found to be associated with mango. The morphological descriptions of the isolates were in agreement with previous descriptions of Colletotrichum species, although cultural characteristics may be diverse due to varied cultural conditions, and conidial morphology alone could not distinguish all species within species complexes 9 .
The present study revealed high Colletotrichum species diversity associated with mango from the six major mango-grown areas in China. The Colletotrichum isolates examined here also showed high diversity, which may correlate with the diversity of environmental conditions such as temperature and rainfall 26 and the sample collection time and isolate source. Colletotrichum asianum, C. fructicola and C. siamense were found to be the major species associated with mango anthracnose in China. This result agrees with what was reported by Vieira et al. 18 (2014) who found that C. asianum was the most common endophytic species from mango trees. Colletotrichum asianum and C. siamense, were report as pathogens of a wide diversity of hosts such as Coffea arabica 9 , Carica papaya 9 and M. indica 15,19,27 . Our study showed that a single host species may be attacked by several Colletotrichum species, and this was previously reported by Phoulivong et al. 28 . Colletotrichum fructicola was initially reported on coffee berries, and it also has a wide host range (Arachis sp., Citrus bergamia and M. indica) as well as wide geographical distribution 29 . Similarly, C. tropicale has been reported on Annona muricata (Annonaceae), Viola surinamensis (Myristicaceae) and M. indica in Brazil 29 .
Colletotrichum musae is a major causal agent of banana anthracnose and associated with fruit disease spots for Musa sp. in many countries 29 . Colletotrichum cliviicola was found to be associated with anthracnose of Clivia www.nature.com/scientificreports www.nature.com/scientificreports/ miniata, Cymbidium hookerianum in China 12,30 and of Phaseolus sp. and Saccharum sp. in India 12 . It was also reported as an endophyte on M. indica in Brazil 18 .
Colletotrichum gloeosporioides was reported as a main pathogen from Citrus sp., Mangifera sp., and Vitis vinifera among many other host species, since it is considered a species complex 9 . It has a particularly wide host range based on publications because many Colletotrichum isolates were previously labelled C. gloeosporioides 29 . In the present study, C. asianum, C. cordylinicola, C. fructicola, C. gloeosporioides, C. kahawae, C. musae, C. siamense and C. tropicale, can all be placed into the C. gloeosporioides species complex 9 , and these results are consistent with previous reports that the C. gloeosporioides species complex is a preponderant Colletotrichum species associated with mango anthracnose 23,31 .
Colletotrichm scovillei was reported as a pathogen on mango 22 and Capsicum sp. (chilli "Django") 32 , and has been placed into the C. acutatum species complex 10 . Colleotrichum karstii was first reported from orchids and was also known from some other hosts such as Carica papaya 11 , Citrus sp. 11 , and M. indica 11,17 . Colletotrichum karstii has been found on mango and avocado fruit 17,33 , and it has been placed in the C. boninense species complex 11 .
Colletotrichum gigasporum belongs to the C. gigasporum species complex 13 . The species has large conidia and its wide range of hosts include Acacia auriculiformis and Coffea sp. (in Vietnam), and Diospyros kaki and Musa sp. (in Japan). Colletotrichum endophytica was found on Pennisetum purpureum in Thailand and tea plant in China 34 .
Colletotrichum liaoningense was first reported as a pathogen of Capsicum annuum var. conoides in China 35 .
Pathogenicity test using thirteen species of Colletotrichum isolates showed that all species were pathogenic to mango leaves and fruits in the wound-inoculation experiment. The pathogenicity tests indicated that aggressiveness of Colletotrichum species on fruit was not completely consistent to that on leaves, but the aggressiveness of different isolates was found to be significantly different. The difference in aggressiveness of Colletotrichum within species was independent of a particular geographical origin for sample collection and varied, perhaps resulting from adaptations to the geographic and environmental origins of the isolates, although no specific correlations were found between aggressiveness and province of origin. However, there were more isolates from Yunnan province with weak aggressiveness compared to those from Hainan province. Symptoms may vary considerably with factors such as cultivar, fruit physiological state, inoculum factors such as concentration, and environmental conditions such as humidity and temperature 33,36,37 . Thus, the pathogenicity test results from this study may not reflect the full aggressiveness potential of the isolates examined. Additional research should be conducted to determine the aggressiveness for Colletotrichum species with conidial inoculation in planta rather than wounded detached leaves and fruits with mycelial plugs.
This study has enhanced our understanding on the diversity of Colletotrichum species associate with mango anthracnose in China. Further study is required to determine differences in biological and infection characteristics among various mango Colletotrichum species, as well as the molecular mechanisms responsible for such differences and their effects on mango anthracnose disease in the field. There have been few studies on the etiology and epidemiology of Colletotrichum species on mango. Therefore, future work on pathogenic mechanisms could be helpful for prevention and control of mango anthracnose disease.

Materials and Methods
Collection of sample and isolation of fungi. During 2016 and 2017, mango leaves, fruits and branches with symptoms or signs of anthracnose were gathered from six provinces (Fujian, Guangdong, Guizhou, Hainan, Sichuan and Yunnan) in China, covering an area 1950 km by 800 km. Mango orchards were sampled with minimum 2 km distance between samples. Lesion margins on mango leaves, branches and fruits were cut into 3 mm × 3 mm pieces, surface sterilized with ethanol (75%) for 15 s, sodium hypochlorite (1% vol/vol) for 2 min, followed by three rinses in autoclaved distilled water for 30 s. Isolated fungus were cultured on potato dextrose agar (PDA) and kept at 25 °C. Over two weeks, the cultures were examined every day. Isolates were placed on PDA slants for longer term storage at 4 °C in the dark.
Morphological characterization. Before morphological examination, isolates were cultured under continuous fluorescent light on PDA for week at 25 °C. Mycelial plugs (5 mm diam) were cut from colony margins and placed 9-cm-diam Petri dishes with four plates per isolate. The colony characteristics (color of upper and lower surfaces) plus growth rate (diameters in two perpendicular directions) were recorded after 3, 5, 7 and 10 days. Hyphal growth rate (cm/day) was calculated based on colony diameter. Conidial production, shape and the size of conidia were examined up to three weeks on PDA incubated at 25 °C. Conidia (120 per isolate) from three-week-old PDA cultures were measured for length and width. Appressoria were produced from conidia placed on glass slides inside moist chambers. Appressoria were produced from hyphae using a slide culture technique described by Cai et al. 6 . PDA plugs (1 cm × 1 cm) were placed in clean Petri dishes, and each agar plug was treated with conidia, and a cover slip was placed over each plug. After 7 days, the shape and size of the appressoria were recorded. For each isolate, at least 30 appressoria were examined, and photographs were taken under a Nikon Eclipse Ni-E microscope.
DNA extraction, polymerase chain reaction (PCR), and sequencing. Isolates were incubated at 25 °C on PDA in Petri plates overlaid with cellophane. After 7 days of growth, mycelia were harvested by scraping the cellophane with autoclaved spatulas, and DNA was extracted using a CTAB method 38 . DNA was amplified with primers for partial actin (ACT), partial chitin synthase (CHS-1), partial glyceraldehyde-3-phosphate dehydrogenase (GAPDH), rDNA-ITS (ITS), and partial β-tubulin (TUB2) genomic regions (Supplementary  Table S4). PCR amplification of ACT, CHS-1, GAPDH, TUB2 and ITS was amplified with the primer-pairs shown in Supplementary Table S4 www.nature.com/scientificreports www.nature.com/scientificreports/ each primer and 1 μl of template DNA (20 ng/μl). The PCR programs for ACT, CHS-1, GAPDH, ITS and TUB2 were set following Mo et al. 15 The PCR products were sequenced by Shanghai Sangon Company in China. phylogenetic analyses. Consensus sequences were obtained from forward and reverse primer sequencing of the same genomic region for each sequenced isolate using DNAMAN version 7.0. The consensus sequences were compared by BLASTn 39 against the NCBI NR database. Among the top matches, select sequences of extype or ex-epitype isolates of Colletotrichum species were selected for the phylogenetic analyses (Supplementary  Table S2). Alignments for each gene were carried out using MUSCLE v. 3.8.3 40 . We tested whether the five gene data sets were combinable using the Partition Homogeneity Test as implemented in PAUP* 4.0b4 and described by Farris et al. 41 as the incongruence length difference (ILD) test 41 . Sequences from the five genes were then manually concatenated and re-aligned. Phylogenetic analysis was conducted using MrBayes version 3.2.6 42 . Models of nucleotide substitution for each gene determined by jModeltest v. 2.1.7 were included for each gene partition 43 . Following Drummond and Rambaut 44 four runs with four chains for 5,000,000 generations were set up while the first 25% were discarded. The analyses were sampled every 1000 generations until the average standard deviation of split frequencies fell below 0.01. Maximum likelihood analyses were conducted with the RAxML GUI v. 1.3.1 45 using a GTRGAMMA substitution model, and bootstrap support was calculated based on 1,000 iterations.
Pathogenicity and aggressiveness on mango tissues. Pathogenicity tests were conducted following Mo et al. 15 . Select isolates were used for pathogenicity and aggressiveness tests on detached leaves and fruits of mango (cv. Tainong) under controlled conditions. Each isolate was incubated on PDA for 7-10 days at 25 °C. Freshly harvested mango fruits and young leaves without visible disease from Tiandong (Guangxi province) were used for the tests. The tissues were surface sterilized in 75% ethanol for 30 s and 1% NaClO for 1 min, and then with three rinses in autoclaved distilled water.
After air drying, detached young healthy leaves (12-15 cm) were placed into plastic containers (9 cm × 17 cm × 25 cm) lined with paper towelling, and six stab wounds were made forming a 5-mm-diam circle. Mycelial plugs (6 mm diameter) from margins of PDA cultures were placed on each wound. The completely randomized trial with three replicates per isolate. For controls, sterile agar plugs were used. The containers were sealed and incubated in the dark at 25 °C. Aggressiveness was assessed one week after inoculation by measuring lesion diameter.
After washing and air drying, newly ripened mango fruits (∼100 g) were wounded by stabbing to a depth of 3 mm to form a 5-mm-diam circle. Mycelial plugs (6 mm diameter) from PDA cultures were placed on the wounds. Sterile agar plugs were used as control treatments. Inoculated fruits were placed in plastic containers on top of moist paper towelling and sealed. The containers were placed in a growth chamber and incubated at 25 °C in the dark. Symptom development of the fruits was checked daily up to 10 days. Isolates were considered pathogenic when lesions expanded beyond the 5 mm circle of the initial wounding. Aggressiveness was evaluated at 10 days after inoculation (dpi) by measuring lesion diameter.
Statistical analyses. Statistical analyses were done using Data Processing System software (DPS 7.0) for Windows (Zhejiang University, Hangzhou, China) 46 , and data were subjected to analysis of variance (ANOVA) followed by Fisher's Least Significant Difference" (LSD) tests for means separation.

Data availability
All data can be available to Editorial Board Members, referees and readers.