Epidemiology, pathology and identification of Colletotrichum including a novel species associated with avocado (Persea americana) anthracnose in Israel

Anthracnose disease caused by Colletotrichum species is a major constraint for the shelf-life and marketability of avocado fruits. To date, only C. gloeosporioides sensu lato and C. aenigma have been reported as pathogens affecting avocado in Israel. This study was conducted to identify and characterize Colletotrichum species associated with avocado anthracnose and to determine their survival on different host-structures in Israel. The pathogen survived and over-wintered mainly on fresh and dry leaves, as well as fresh twigs in the orchard. A collection of 538 Colletotrichum isolates used in this study was initially characterized based on morphology and banding patterns generated according to arbitrarily primed PCR to assess the genetic diversity of the fungal populations. Thereafter, based on multi-locus phylogenetic analyses involving combinations of ITS, act, ApMat, cal, chs1, gapdh, gs, his3, tub2 gene/markers; eight previously described species (C. aenigma, C. alienum, C. fructicola, C. gloeosporioides sensu stricto, C. karstii, C. nupharicola, C. siamense, C. theobromicola) and a novel species (C. perseae) were identified, as avocado anthracnose pathogens in Israel; and reconfirmed after pathogenicity assays. Colletotrichum perseae sp. nov. and teleomorph of C. aenigma are described along with comprehensive morphological descriptions and illustrations, for the first time in this study.


Results
Fungal isolates and host infections. A total of 576 Colletotrichum isolates were recovered from different tissue (fruits, fresh leaves, fresh twigs, dry leaves and dry twigs) of avocado (Supplementary Table 1). Percentage of Colletotrichum isolates obtained from different host tissues is presented in Fig. 1. The Colletotrichum isolates were most readily isolated from infected fruits (94.88%), as compared to green leaves (19.87%), green twigs (10.93%) and dry leaves (18%) showing typical anthracnose symptoms. Low infection values of 0.9% were recorded for dry and dead twig tissues. Morphologically identical isolates recovered from the same tissue samples during isolation were discarded and 538 isolates were then selected for further molecular characterization.
Assessment of genetic diversity. Amplification products were obtained for all the 538 Colletotrichum isolates from this study using four arbitrarily primed PCR (ap-PCR) primers: (CAG) 5 , (GACA) 4 , (AGG) 5 and (GACAC) 3 . A high level of genetic diversity was observed, categorizing the isolates into eight distinct genetic groups (Supplementary Text). Genetic variability between the representative isolates from each group is presented in Supplementary Fig. 1. Thirty-three isolates were then selected from the eight genetically distinct groups based on their geographical location and isolation-tissue, for further multi-gene phylogenetic analyses, pathogenicity testing and morphological characterization.
Identification of species complex using ITS gene region. Following assessment of genetic diversity, the ITS gene region of the 33 isolates was sequenced for their preliminary identification to the species complex level. Based on NCBI-BLAST search results of the ITS sequences, 31 isolates belonged to the C. gloeosporioides species complex and two isolates belonged to the C. boninense species complex. Further phylogenetic analyses were performed according to the described set of gene markers for the respective species complexes 9,11 . NCBI accession numbers for the sequences generated in this study appear in Tables 1 and 2. ApMat marker based phylogenetic analysis of the C. gloeosporioides species complex members. The ApMat dataset included 58 sequences and 944 characters including gaps. Forty-one characters from the ambiguously-aligned regions were excluded from the analysis. Of the remaining 903 characters, 398 characters were constant, 320 characters were parsimony-informative and 185 characters were parsimony-uninformative. MP analysis resulted in two trees and based on the KH test, the second tree was not significantly different as compared to the best tree (details not shown). One tree (TL = 832, CI = 0.773, RI = 0.932, RC = 0.721, HI = 0.227)  Table 1. GenBank accession numbers of the Colletotrichum isolates belonging to the C. gloeosporioides species complex sequenced for the ITS, gapdh, act, tub2, cal, ApMat and gs gene sequences from this study (N.S. = not sequenced, T = type strain of the newly described species, in bold).   Fig. 3 (length = 2107, CI = 0.728, RI = 0.844, RC = 0.614, HI = 0.272). The topologies of the 240 trees were not significantly different (details not shown). The clustering pattern for the 31 isolates from 6-gene analysis was comparable to ApMat-based phylogeny and the overall bootstrap support for individual branches was typically strong [C. aenigma (88%), C. fructicola (100%), C. gloeosporioides (100%), C. nupharicola (69%), C. perseae sp. nov. (100%), C. siamense (78%), C. theobromicola (100%)]. The 5-gene (act, cal, gapdh, ITS, tub2) phylogenetic analysis (data not shown) was incapable of resolving C. siamense, C. gloeosporioides, C. theobromicola, C. alienum, C. fructicola and C. nupharicola into strongly supported clades. However, C. aenigma and C. perseae sp. nov. were resolved with 88% and 100% bootstrap support values, respectively.
The ApMat-gs dataset included 1867 characters including gaps (gene boundaries ApMat: 1-944, gs: 945-1867). The analysis involved 55 sequences, including 31 sequences from this study. Ninety-one characters from the ambiguously aligned regions were excluded from the analysis. Of the remaining 1776 characters, 965 were constant, 463 were parsimony-informative and 348 were parsimony-uninformative. The MP analysis resulted in one tree (TL = 1290, CI = 0.763, RI = 0.905, RC = 0.690, HI = 0.237) as shown in Supplementary Fig. 2. The bootstrap support values exceeding 50% for the observed branching pattern are indicated next to the branches. The ApMat-gs tree is strongly supported with high bootstrap values as compared to the 6-gene phylogeny [C. aenigma Pathogenicity assay. The inoculated fruits developed typical anthracnose lesions around the wound (Fig. 5); however, disease development at the unwounded site was very limited or absent even after seven days post inoculation (data not shown). To validate Koch's postulates, pathogens were re-isolated from the infected host tissues. The control fruits did not develop anthracnose symptoms. All Colletotrichum isolates from this study caused 100% disease incidence. Colletotrichum aenigma proved to be the most virulent pathogen of avocado in Israel with 92.6 ± 7.7% disease severity ( Table 3). The next two virulent pathogens were C. alienum and C. theobromicola with 90.1 ± 6.7 and 88.9 ± 3.7% disease severity scores, respectively. The percent disease severity (PDS) scores for other isolates were: C. siamense (85.9 ± 4.3%), C. fructicola (85.2 ± 4.3%), C. gloeosporioides (82.7 ± 5.0%), C. perseae sp. nov. (80.2 ± 2.7%), C. karstii (67.9 ± 6.5%) and C. nupharicola (63.0 ± 14.7%). Pathogenicity assays were performed in triplicate, and similar results were obtained for each experiment (data not shown). Calculations for the results of PDS are provided in Table 3.
Geographic distribution: Known only from avocado (Persea americana Mill.) from Israel. Genetic identification: ITS sequences are not sufficient to distinguish C. perseae from the species in the Fructicola clade. Colletotrichum perseae was well resolved using ApMat and gs markers.

Discussion
Phenotypic plasticity within Colletotrichum species complexes is a key limiting factor in species delimitation. Although a polyphasic approach towards characterization of Colletotrichum species is a recommended strategy 18 ; there is a lack of consensus among mycologists regarding the choice of markers to be used for multi-locus phylogeny 37,44 . Thus far, 11 species complexes of Colletotrichum have been distinguished: C. acutatum, C. boninense, C. caudatum, C. dematium, C. destructivum, C. gigasporum, C. gloeosporioides, C. graminicola, C. orbiculare, C. spaethianum, and C. truncatum [8][9][10][11][12]31,[39][40][41]45 . Besides these 11 species complexes, 23 single species and some independently evolved small clusters have been described e.g. C. dracaenophilum, C. yunnanense, C. cliviae and C. araceaerum 8,31,46 . Each species complex is recognized by the specific epithet of a historically known or well-studied species 37 . Colletotrichum gloeosporioides sensu lato remains the most confusing taxa within the Colletotrichum genus. Following extensive taxonomic revisions [8][9][10][11][12]39,40 , attempts were made to investigate a potential secondary barcode for Colletotrichum. To date, ApMat and gs are reported to be efficient in species delimitation within the C. gloeosporioides species complex 21,22,24,26,27,42 . It is crucial to accurately identify a pathogen, for effective plant quarantine purposes, breeding programs and disease control 47,48 . This is especially important for economically important agricultural commodities such as avocado.
This study highlights the genetic heterogeneity of Colletotrichum populations associated with avocado anthracnose in Israel. Nine dominant Colletotrichum spp. causing avocado anthracnose were identified, including one new species, C. perseae sp. nov., based on multigene phylogeny, pathogenicity assays and morphology. Among the nine species identified in this paper; C. aenigma, C. alienum, C. fructicola, C. gloeosporioides, C. karstii and C. siamense have been previously reported from Persea americana 11,26,43,49 . To date, only C. gloeosporioides sensu lato and C. aenigma have been reported from avocado in Israel 3,11,14,43,50 . This is the first report demonstrating pathogenicity of C. persea sp. nov., C. nupharicola and C. theobromicola, specifically in avocado. However, other Colletotrichum species affecting avocado in this study in Israel have been reported to attack diverse hosts.   In congruence with previously published studies, the ApMat marker proved to be superior in resolving species within the C. gloeosporioides species complex 21,22,24,26,27,42 . The Apn2-Mat1-2 locus was first used for delineating populations within the C. graminicola species complex 55 . The ApMat gene exhibited the following advantages which established it as a promising marker for molecular systematics of the Colletotrichum species complexes: (a) the apn2-Mat1-2 intergenic region is flanked by relatively conserved regions, useful for the design of specific primers, (b) the overall phylogenetic resolution provided by this single marker is more informative than a ScIentIfIc REPORtS | 7: 15839 | DOI:10.1038/s41598-017-15946-w multigene phylogeny, and (c) the Apmat gene is a highly variable marker which can surpass the problems of gene tree discordance 42 . In combination with ApMat, gs was also efficient for species resolution within the C. gloeosporioides species complex, as previously observed 26 .
Pathogenic variability was also observed among the nine representative Colletotrichum species isolates. Pathogenicity assays confirmed that all the nine species cause anthracnose disease in avocado (Table 3). Based on percent disease severity (PDS) calculations, C. aenigma was the most virulent pathogen of avocado in Israel with 92.6 ± 7.7% values. C. perseae sp. nov. emerged as the most dominant pathogen of avocado anthracnose in Israel, occurring in all the sampled areas, totaling 354 of the 538 isolates (65.8%) included in this study. This suggests that C. perseae sp. nov. could be used as a model species for studying population structure and evolution of anthracnose in Israel. Furthermore, the occurrence of the teleomorph in C. perseae sp. nov and C. aenigma species is noteworthy, as no reports of the appearance of the sexual stage have been previously reported, albeit under artificial culture conditions; but may also appear in nature, thus contributing to the genetic diversity of these and other Colletotrichum species. The isolates also exhibited a pattern in geographic distribution (Supplementary Text), C. theobromicola was recovered only from samples collected from Central Israel; C. persea sp. nov. was mainly recovered from samples collected from Central and Northern Israel while C. aenigma was mainly recovered from samples collected from Southern Israel; C. gloeosporioides was mainly recovered from samples collected from Central Israel and C. siamense was mainly recovered from samples collected from the Northern and Southern Israel.
In this study percentage of isolations from different avocado tissues was significantly higher from fruits (94.9%), as compared to green/fresh leaves (19.9%), dry leaves (18%) and green/fresh twigs (10.9%), whereas only a trace percentage originated from dry and dead twigs. This is in contrast to an earlier report 56 , where large numbers of C. gloeosporioides conidia were produced on dead leaves and infected/mummified fruits in avocado trees in Australia; while minimal numbers were produced on dead twigs and none isolated from branches and green leaves. This may be associated with a dryer environment in certain areas in Australia as opposed to the constant wet, rainy winter season occurring in Israel, when avocado fruits are predominantly harvested. This elevated humidity may contribute to the relatively high percent of infected green and fresh leaves and allow survival of inoculum of the pathogen in the form of quiescent germinated apperssoria in these tissues during the dry summer season prevailing under Israeli cultivation conditions. Similarly, increased rainfall resulted in increased levels of rots in harvested avocados in certain regions of Australia 57 , and more quiescent infections became established in the temperate regions of South Africa during the rainy part of the year rather than in the dry winter months 58,59 . While spores of C. gloeosporioides were released from dead leaves entangled in the main canopy 60 , removal of this material and dead twigs from the canopy did not consistently reduce the numbers of postharvest rots in avocados while the principal means of spread to avocado fruit occurred via rain-borne inocula 61 . In another report, C. gloeosporioides conidia and perithecia were found in margins of leaf and twig lesions and in the bark from the trunks of trees 59 . The teleomorph may also occur more commonly in orchards in Israel which has remained unreported, similar to that detected in vitro in plates.
In summary, this work has shed light on aspects of epidemiology of avocado anthracnose in Israel, which may lay the foundation for future studies related to management of the pathogen under field conditions. The diverse genetic structure of the pathogen in Israel, further attests to the probability of the teleomorph existing under field conditions in avocado specifically, and in other economically important crops in general, affected by members of C. gloeosporioides s. l.

Materials and Methods
Sample collection. From November 2014 to April 2015, plant samples (fruit, fresh leaves and twigs, dry leaves and twigs) were collected from avocado orchards in the following locations in Israel: Kfar Yuval (Northern Israel), Beit Haemek (Northern Israel), ARO, Bet Dagan (Central Israel), Mikve Israel experimental farm (Central Israel) and Kfar Aza (Southern Israel). Initial samples were collected from plantations of Ettinger and Hass cultivars located in Mikve Israel and ARO, Bet Dagan, (Central Israel). In each orchard, five trees were selected randomly and samples of fruit, leaves (fresh green and dry leaves from the ground) and twigs (fresh green and dry twigs) were collected. During the initial isolation, only 18.6 and 3.4% of Colletotrichum isolates were recovered from dry leaves and twigs, along with other common fungal saprophytes such as Aspergillus and Alternaria; therefore in further samplings from Northern and Central Israel, only fruits, green leaves and green twigs were collected. In Northern Israel (from Kfar Yuval and Beit Haemek), Reed and Hass cultivars were sampled, while in Southern Israel (from Kefar Aza) Hass cultivar was sampled. From each tree five fruits, twigs and leaves of each were sampled. Samples were brought to the laboratory and maintained in a moist chamber at room temperature (20 to 25 °C) and inspected regularly for the appearance of anthracnose symptoms.
Fungal isolation and culture conditions. Anthracnose symptoms were observed after 7-10 days in the collected fruits, leaves and twigs maintained under humid conditions. From each fruit two necrotic disease spots were selected for fungal isolation; while from each twig and leaf, five disks were removed (see below). Colletotrichum strains were isolated from the visible sporulation obtained on fruit lesions using the single spore method 62 . Isolation from leaves and twigs was performed initially using a tissue isolation method, whereby five sections of 1 cm 2 size were cut from each leaf or twig near the infected area, surface sterilized with 70% ethanol for 20 seconds, 1% sodium hypochlorite (NaOCl) for 3.5 minute, washed with sterile water and dried on sterilized tissue paper. The plant tissue was then placed aseptically on Mathur's MS semi-selective (M3S) agar medium    Table 1). Percentage occurrence of Colletotrichum isolates from fruit, leaves and twigs was calculated to determine recovery of the pathogen from the different plant parts (Fig. 1). PCR reactions were carried out in a thermocycler (Biometra, Germany) with the following cycling parameters: initial denaturation at 95 °C for 5 minutes, followed by 29 cycles of denaturation at 95 °C for 30 seconds, annealing for 30 seconds (60 °C for CAG and AGG; 48 °C for GACAC and GACA), and extension at 72 °C for 1 minute and 30 seconds, and a final extension at 72 °C for 15 minutes. The PCR amplification and the reaction results were maintained at 4 °C until further processed. The PCR products were separated in 1.8% agarose gel (15 × 10 cm, W × L) in Tris-Acetate-EDTA buffer, at 80 V, 400 mA for 2 hours and stained with ethidium bromide (0.5 µg/ml) to visualize the banding patterns using ENDURO TM GDS gel documenting system (Labnet, USA). PCR reactions were repeated three times with consistent results. Variation based on ap-PCR analysis was not quantified but diversity was interpreted according to overall banding patterns for all the 538 Colletotrichum isolates used in this study (Supplementary Text). Representative isolates of the different groups selected after ap-PCR were then used for sequence based analyses. PCR amplification and sequencing. Thirty-three representative Colletotrichum spp. isolates were selected according to ap-PCR for multi-locus phylogenetic analyses. PCR amplification of act, cal, gapdh, tub2 and ITS regions was performed for all the isolates. In addition, gs and ApMat were amplified for isolates belonging to the C. gloeosporioides species complex, while chs1 and his3 gene regions were amplified for isolates belonging to the C. boninense species complex. The PCR reactions were carried out as described 11,38,42 . PCR products were purified with the Gel/PCR DNA fragments extraction kit (Geneaid, Catalogue# DF100, Taiwan), and quantified using a Nanodrop Spectrophotometer ND-1000 (Thermo, USA). Purified PCR products were sequenced by Macrogen Europe (http://www.macrogen.com) and submitted to NCBI-GenBank (Tables 1 and 2).

Phylogenetic analyses.
Phylogenetic analyses were carried out using the multigene dataset for the C. gloeosporioides species (act, cal, gapdh, ITS, tub2) and C. boninense species complexes (act, cal, chs1, gapdh, his3, ITS, tub2) using reference sequences 9,11 . In addition, analyses for ApMat marker, gs gene, 2-markers (Apmat, gs), 6-genes (act, cal, gapdh, gs, ITS, tub2) and 7-gene (act, ApMat, cal, gapdh, gs, ITS, tub2) were also performed for the C. gloeosporioides species complex isolates using recently published reference sequences 24,26 . Reference sequences for the newly described Colletotrichum species within the C. gloeosporioides species complex (C. chengpingense, C. conoides, C. grossum, C. hebeinse, C. helleniense, C. henanense, C. hystricis, C. jiangxiense, C. liaoningense, C. wuxiense) were also added to the dataset 31,65,66 . Maximum Parsimony (MP) analysis was conducted using PAUP version 4.0b10 67 . Ambiguous regions within the alignment were removed from the analyses and the gaps were considered as missing data. The trees were inferred using the heuristic search option with Tree Bisection Reconnection (TBR) branch swapping and 20 random sequence additions. Maxtrees were set to 10000, zero length branches were collapsed and all multiple parsimonious trees were saved. In addition, descriptive tree statistics, including tree length (TL), consistency index (CI), retention index (RI), rescaled consistency index (RC), and homoplasy index (HI) were recorded. The strength of clades was assessed by a bootstrap analysis with 100 replicates. The resulting trees were viewed using TreeView 68 and edited in MEGA version 7.0.14 69,70 and Microsoft PowerPoint version 2007. The alignment files and trees were deposited in TreeBase (www.treebase.org; Study ID: 20611). The MP trees generated in this study are shown in Figs 2-4, and Supplementary Fig. 2. Additionally, Maximum likelihood (ML) trees were generated for each dataset using "one click mode" available at the online platform for phylogenetic analysis, www.phylogeny.fr 70 . The bootstrap support values generated using two methods (MP and ML) and the resulting tree topologies were also compared (data not shown).
Pathogenicity assays. Pathogenicity assays were performed for representative Colletotrichum spp. isolates (C. aenigma -GA050, C. alienum -GA524, C. fructicola -GA186, C. gloeosporioides -GA070, C. karstii -GA206, C. nupharicola -GA253, C. perseae sp. nov. -GA100, C. siamense -GA331, C. theobromicola -GA002), essentially as described 3 . Isolates were cultured in M3S agar plates at 25 °C to induce sporulation. After 7 days, conidia were harvested by flooding the culture with 0.85% NaCl (normal saline, with 100 µl/lt Tween 80) and dislodging the conidia with a glass spreader. The conidial solution was filtered through sterile gauze to remove hyphal filaments and concentrated by centrifugation at 8000 rpm for 10 minutes at 4 °C. The supernatant was discarded and the pellet was washed and re-suspended in 1 ml of cold normal saline solution. The conidial concentration was adjusted to a final working concentration of 1 × 10 7 conidia/ml. Disease free, fresh avocado fruits (Reed and Hass cultivars) were collected from Eyal orchard, Central Israel for pathogenicity assays. Assays were conducted in three replicates containing three fruits each, with two infection sites per fruit. The fruits were surface sterilized using 1% sodium hypochlorite solution, washed and dried using sterile filter paper. Fruits were inoculated with 7 µl of conidial suspension (1 × 10 7 conidia/ml) at wounded (pin-pricked) and unwounded sites (3 sites per fruit). Control fruits were mock-inoculated with sterile normal saline solution. Inoculated fruits were maintained under humid conditions at 25 °C. The fruits were monitored regularly for the appearance of anthracnose symptoms. Disease severity was scored using the 0-9 point scale 65,66,71 at 7 days post inoculation (dpi) and calculations were made as previously described 27 .