Identification and characterization of Colletotrichum species causing apple bitter rot in New York and description of C. noveboracense sp. nov.

Apple bitter rot caused by Colletotrichum species is a growing problem worldwide. Colletotrichum spp. are economically important but taxonomically un-resolved. Identification of Colletotrichum spp. is critical due to potential species-level differences in pathogenicity-related characteristics. A 400-isolate collection from New York apple orchards were morphologically assorted to two groups, C. acutatum species complex (CASC) and C. gloeosporioides species complex (CGSC). A sub-sample of 44 representative isolates, spanning the geographical distribution and apple varieties, were assigned to species based on multi-locus phylogenetic analyses of nrITS, GAPDH and TUB2 for CASC, and ITS, GAPDH, CAL, ACT, TUB2, APN2, ApMat and GS genes for CGSC. The dominant species was C. fioriniae, followed by C. chrysophilum and a novel species, C. noveboracense, described in this study. This study represents the first report of C. chrysophilum and C. noveboracense as pathogens of apple. We assessed the enzyme activity and fungicide sensitivity for isolates identified in New York. All isolates showed amylolytic, cellulolytic and lipolytic, but not proteolytic activity. C. chrysophilum showed the highest cellulase and the lowest lipase activity, while C. noveboracense had the highest amylase activity. Fungicide assays showed that C. fioriniae was sensitive to benzovindiflupyr and thiabendazole, while C. chrysophilum and C. noveboracense were sensitive to fludioxonil, pyraclostrobin and difenoconazole. All species were pathogenic on apple fruit with varying lesion sizes. Our findings of differing pathogenicity-related characteristics among the three species demonstrate the importance of accurate species identification for any downstream investigations of Colletotrichum spp. in major apple growing regions.

www.nature.com/scientificreports www.nature.com/scientificreports/ Phylogenetic analyses. The ITS phylogeny concurred with the multiplex PCR assay in that Colletotrichum isolates collected in this study fell into two Colletotrichum species complexes: C. acutatum (19 isolates) and C. gloeosporioides (25 isolates) with high support (Supplementary Fig. S2). The C. acutatum phylogeny dataset included 85 taxa (including 19 isolates from this study) and 1278 characters consisting of three loci (ITS, TUB2 and GAPDH). Two C. orchidophilum isolates, CBS 119291, and CBS 632.80, were used as an outgroup. All five major C. acutatum clades 42 were resolved with high support (BS ≥ 84, PP = 0.99; Fig. 1). Both Bayesian Inference (BI) and Maximum Likelihood (ML) analyses revealed that the 19 isolates collected in this study clustered with C. fioriniae as part of clade 3 of the CASC with full support (BS/PP: 100/1; Fig. 1) and are hereafter designated as C. fioriniae. We found that the majority of isolates in this study included in the CASC phylogentic analysis clustered with C. fioriniae type isolate CBS 128517, with high PP support (0.98), but lacking BS support ≥ 70. The remaining three isolates (ACFK3, ACFK6, ACFK205) fell outside of this group (Fig. 1), though remaining within the highly supported Clade 3 (C. fioriniae). ACFK3 and ACFK6 were placed well-within a different subclade with high BS support (85) but lacking PP support ≥ 0.90.While further analysis is required, we believe this separation of the isolates in this study may be similar to the previous fnding that the C. fioriniae clade is partitioned into two major subclades 42 .
The C. gloeosporioides phylogeny dataset included 201 taxa (including 25 isolates collected in this study and Coll940) and 4890 characters consisting of eight loci (ACT, ApMat, CAL, GAPDH, GS, APN2, ITS and TUB2). The outgroup included one member of the CASC, C. javanense CBS 144963, and two members of the C. boninense complex, C. boninense CBS 123755 and C. hippeastri ICMP17920.
The 26 isolates belonging to CGSC collected in this study from apple in New York, Virginia, and Pennsylvania, were found to group into three distinct clades, two of which represent previously described species within the CGSC. Twelve isolates, AFK17, AFK18, AFK22, AFK26, AFK28, AFK30 and AFK31 from New York, isolates AFK154 and PMAREC-1a from Virginia, and isolates PMKnsl-1, PMCMS-6760 and PMLynd-9a from Pennsylvania, grouped with the ex-type strain of C. chrysophilum with maximum support (BS/PP: 100/1; Fig. 2) and are hereafter designated as C. chrysophilum. To our knowledge, this is the first time that the C. chrysophilum species has been reported to cause bitter rot disease on apple. After C. fioriniae, C. chrysophilum was the second most abundant species causing bitter rot disease in New York.
Two isolates, AFK156 and PMCrwn1 from Virginia, grouped with the ex-type strain of C. fructicola (BS/PP: 91/1; Fig. 2) and are hereafter designated as C. fructicola. Of the two CGSC isolates originally collected from peach in South Carolina 49 , RR12-1 was found to group with C. fructicola, as previously reported 47 . The second isolate, RR12-3, previously recognized as C. fructicola using a multi-locus analysis (CAL, GAPDH and TUB2 49 ), was found clustered within the fully supported C. chrysophilum clade in our eight-gene multi-locus analysis (Fig. 2). Further, CGSC member GA253, isolated from avocado in Israel 69 , which was previously identified as C. nupharicola using an ApMat phylogeny as well as a six-gene multi-locus analysis 69 , was found to cluster within the C. chrysophilum clade (Fig. 2). No isolates belonging to C. fructicola were identified from apple fruit in New York and Pennsylvania.
The remaining 9 isolates, AFKH109, AFK65, AFK220, AFK289, AFK408 and AFK423 from New York and PMEssl-10a, PMCMS-6751 and PMBrms-1 from Pennsylvania, formed a separate, distinct clade with maximum support and independent from any recognized species in the CGSC (BS/PP: 100/1; Fig. 2). This distinct clade included isolate Coll940, which was originally isolated from leaves of black walnut (Juglans nigra) in Oklahoma and had an uncertain placement based on nrITS, TUB2, APN2 and ApMat analyses 53 . We pursued further analyses to determine if this unique cluster represented a new, undescribed lineage in the CGSC. For phylogenetic models and partitioning schemes see Supplementary Table S1. Species delimitation. All Colletotrichum isolates from apple were assigned to a lineage containing the ex-type of a previously described species using genealogical concordance phylogenetic species recognition approach (GCPSR) except for AFKH109, AFK65, AFK220, AFK289, AFK408, AFK423, PMEssl-10a, PMCMS-6751, and PMBrms-1. These isolates were strongly supported in the 8-locus concatenated analyses as monophyletic (BS = 100; PP = 1) and sister to C. fructicola, C. nupharicola and C. chrysophilum. Among the independent gene trees, these isolates were strongly supported as monophyletic in the ApMat (BS = 99; PP = 1), APN2 (BS = 100; PP = 1), and GS (BS = 92; PP = 1) phylogenies. These isolates were also inferred to be monophyletic in the ACT phylogeny, although with weak support in both the ML analysis (BS = 54) and BI analysis (PP = 0.85). While they were not monophyletic in the phylogenies inferred from TUB2, ITS, GAPDH and CAL, there was no strongly supported conflict in those trees. Our results are consistent with the criteria of GCPSR for recognizing these isolates as an independent lineage representing a novel species of Colletotrichum, named as C. noveboracense. Phylogenetic models and partitioning schemes used can be found in Supplementary Table S1. Morphology characterization. We described morphological characteristics including colony color, conidial shape, measurements of colony growth rate and conidial length and width for several isolates of each Colletotrichum species causing apple bitter rot in this study (C. fioriniae, C. chrysophilum, C. fructicola and C. noveboracense). The isolates of C. fioriniae produced salmon to red conidial masses on 7-day-old cultures on PDA in both front and reverse sides and produced fusiform conidia after 10 days on PDA (Fig. 3a-c). Isolates belonging to C. chrysophilum initially presented colonies in white to light gray and progressively turned to dark grey in the center covered with predominantly black acervuli, producing orange conidial masses with longer incubation time. Cylindrical conidia with rounded ends developed after 10 days of incubation on PDA for this species (Fig. 3d-f). C. fructicola formed off-white to slightly gray aerial mycelium and yellowish to grey in reverse, developing cylindrical conidia with rounded ends after 10 days of incubation on PDA (Fig. 3g-i). Comparisons of conidial dimensions and shape, colony growth rates, as well as the description of colony color are presented in Table 1 Detailed morphological description for C. noveboracense is provided in the Taxonomy section.     Diagnosis. Isolates of Colletotrichum noveboracense are strongly supported as monophyletic by the combined analysis of ACT, APN2, GS, CAL, ApMat, GAPDH, ITS and TUB2 and sister to C. nupharicola, C. chrysophilum and C. fructicola. C. noveboracense differs from C. nupharicola by having a faster growth rate on PDA as well as shorter and narrower conidia. C. nupharicola also differs in having an orange colony that turns black with age on PDA versus white to grey colony color for C. noveboracense. C. noveboracense differs from C. fructicola by having shorter conidia and lighter colonies on PDA and differs from C. chrysophilum by having a slower growth rate on PDA. Sequence data from ApMat, APN2, GS and ACT delimit C. noveboracense, but C. noveboracense could not be distinguished by sequences of GAPDH, CAL, TUB2 and ITS.
Additional specimens examined. USA

Notes.
A very low sporulation rate was observed among the isolates collected from New York. No sporulation was seen on PDA and ½ strength PDA except for few isolates including AFKH109 which sparsely produced conidia on OMA. However, isolates collected from Pennsylvania sporulated on ½ strength PDA and OMA.
Agar plate enzyme activity. All isolates belonging to C. noveboracense, C. fioriniae and C. chrysophilum showed lipolytic, amylolytic and cellulolytic activity after five days of incubation on PDA (Fig. 6). However, none of the isolates showed halos of degradation for the proteolytic activity on skimmed milk. All isolates evaluated in this study showed cellulolytic activity as a yellow halo around the colony in plates including CMC stained with Congo red and secured with NaCl. C. chrysophilum showed a significantly larger mean degradation halo of 8 mm in cellulolytic activity assay when compared to C. noveboracense and C. fioriniae (mean halo zone 6 and 6.5 mm, respectively) ( Fig. 6a,b). Colletotrichum isolates of all three species produced hallo around their colonies indicating their ability to produce lipase. C. chrysophilum isolates exhibited significantly the lowest lipid degradation with the mean halo diameter of 17 mm compared to isolates of C. fioriniae and C. noveboracense, with the mean halo diameter of 23 and 27 mm, respectively (Fig. 6c,d). The screening of Colletotrichum isolates on starch agar plates showed that all species produced halo zones reflecting amylase activity after exposure to iodine. The smallest and largest halo sizes for amylase belonged to C. fioriniae and C. noveboracense isolates, with an average of 3 and 8 mm, respectively. However, no significant difference was observed between the mean degradation halo of C. chrysophilum and C. fioriniae (Fig. 6e,f). www.nature.com/scientificreports www.nature.com/scientificreports/ Fungicide sensitivity. Isolates belonging to the three Colletotrichum species showed significantly lower sensitivity to natamycin (mean EC 50 values ranged from 4 to 5 µg/ml) compared to the other fungicides (mean EC 50 values less than 0.5 µg/ml) (Fig. 7). Relative to C. noveboracense and C. chrysophilum, C. fioriniae isolates exhibited greater sensitivity to difenoconazole (EC 50 value of 0.09 µg/ml), pyraclostrobin (EC 50 value of 0.04 µg/ ml) and fludioxonil (EC 50 value of 0.12 µg/ml), but had less sensitivity to thiabendazole and benzovindiflupyr with EC 50 values of 0.4 and 0.3 µg/ml, respectively. With respect to the relative fungicide sensitivity of individual species within the CGSC, we found that C. noveboracense isolates had significantly higher EC 50 values in response to the fungicide difenoconazole and fludioxonil compared to C. chrysophilum. While all members of the CGSC responded similarly to thiabendazole (mean EC 50 of 0.2 µg/ml), C. chrysophilum isolates were significantly less sensitive to pyraclostrobin and benzovindiflupyr (mean EC 50 values of 0.26 µg/ml) when compared to C. noveboracense with mean EC 50 value of 0.17 µg/ml (Fig. 7).
Pathogenicity. All isolates caused the typical symptoms of bitter rot as light to dark brown and sunken circular lesions on apple fruit of cultivar 'Honeycrisp' . To meet the requirements of Koch's postulates, Colletotrichum isolates were recovered from inoculated apple fruit and re-identified. Symptoms did not develop on the apple fruit inoculated with agar plugs. In the comprehensive pathogenicity test of selected Colletotrichum isolates of each species, the average diameter of lesions varied between the two species complexes and among the species  Colletotrichum fioriniae was represented with isolates ACFK145, ACFK15, ACFK16, ACFK205, ACFK29 and ACFK25; Colletotrichum noveboracense was represented with isolates AFK220, AFK289, AFK408, AFK423, AFKH109 and AFK65; and Colletotrichum chrysophilum was represented with isolates AFK17, AFK28, AFK154, AFK18, AFK31 and AFK22. Colletotrichum species followed by the same letter were not significantly different based on Bonferroni Comparison Posttest (p ≤ 0.05). Error bars represent standard deviation.
www.nature.com/scientificreports www.nature.com/scientificreports/ within the CGSC complex (Fig. 8). The average diameter of lesions caused by C. fioriniae isolates on apple fruit of 'Fuji' and 'Gala' was significantly smaller, 17.7 and 39.6 mm, respectively, compared to that of produced by C. chrysophilum and C. noveboracense (Fig. 8). However, in 'Red Delicious' , 'Golden Delicious' and 'Honeycrisp' , the average lesion diameter caused by C. fioriniae isolates were the same as that produced by C. chrysophilum isolates (53.2, 76.5 and 44.4 mm) but significantly larger than the lesions developed by C. noveboracense (29,57,23.5 mm) (Fig. 8).

Discussion
Effective control of plant diseases caused by Colletotrichum species and determination of host specificity and virulence factors are reliant on precise identification and accurate taxonomical delimitation of species boundaries. The assorting of Colletotrichum isolates recovered from apple fruit in this study to CASC and CGSC using a multiplex-PCR 49 , confirmed the reliability and affordability of this method to differentiate between these two species complexes. The ITS gene tree placed all the isolates in the CASC and CGSC with strong support, aligning with previous studies confirming the utility of ITS sequencing for classifying Colletotrichum isolates at the species complex level 49,52 . In addition, the placement of our isolates in this study into the CASC and CGSC supports previous evidence that species from these two species complexes are predominantly involved in causing apple bitter rot 16,17,29,43 . Although the application of ITS rDNA sequences to identify Colletotrichum species was used in studies in the 1990s [70][71][72] , ITS data are insufficient for identifying species in the CGSC 39,40,73 . The multi-locus analyses provided strong resolution and placed the Colletotrichum isolates causing bitter rot of apple in New York orchards in C. fioriniae clade from CASC and C. chrysophilum clade from CGSC. It also contributed to the identification of a new species in this study, C. noveboracense, causing apple bitter rot in New York and Pennsylvania.
In our study, C. fioriniae was the dominant species causing bitter rot on apple which is consistent with previous work in Kentucky where C. fioriniae was also the most abundant species 17 . C. fioriniae causes bitter rot on apple in the US, Korea and Croatia 17,43,74,75 . Seventy percent of isolates recovered from apple orchards in Arkansas, North Carolina and Virginia were identified as C. acutatum 76 , the taxon assigned to all Colletotrichum strains with acute conidia which later were assigned to CASC of over a dozen species 42 .
C. fructicola was reported as a causal agent of bitter rot of apple in the USA, Brazil, Korea and Uruguay 24,43 . However, in our work, C. fructicola was recovered only from symptomatic apple fruit received from Virginia, not from apple orchards in New York and Pennsylvania. C. fructicola was reported to represent the most biological and geographical diversity in the CGSC 3 . Its host range and distribution were reported from coffee berries in Thailand, peach in USA, avocado in Australia and apple in USA, Brazil and Korea, to name a few examples of the geographic and host diversity from which this species has been isolated 3 .
First, consisting of two strongly supported monophyletic subclades, C. ignotum was described as an endophyte of Genipa americana, Tetragastris panamensis and Theobroma cacao 52 . This species was later synonymized with C. fructicola 3 , with the ex-type of C. ignotum and C. fructicola nested within the same subclade. Later, it was determined that the second subclade within C. fructicola represented an independent evolutionary lineage and was described as C. chrysophilum 51 . We detected C. chrysophilum for the first time as pathogen on apple in New York and Pennsylvania. It ranks as the second most common species identified in apple orchards in New York, after C. fioriniae. In classification of Colletotrichum isolates causing anthracnose of peach, using CAL, GAPDH and TUB2, isolate RR12-3 clustered with C. fructicola reference strain ICMP 18645 with bootstrap value 94 49 . By adding five more partial gene sequences, RR12-3 was re-identified as C. chrysophilum in our phylogenetic analyses. In addition, we re-identified isolate GA253 as C. chrysophilum, that was previously identified as C. nupharicola 69 . These findings expand the known host range and geographic distribution of C. chrysophilum, which has been identified on cacao and genipa (Genipa americana; Panama 52 ), fern (Terpsichore taxifolia; Puerto Rico 53 ), avocado (Israel 69 ), peach (South Carolina 49 ) and banana (Brazil 51 ).
Colletotrichum noveboracense was identified as a new species in Colletotrichum genus causing apple bitter rot disease in New York and Pennsylvania. The nine C. noveboracense isolates from New York and Pennsylvania, as well as a single endophytic isolate from Juglans nigra in Oklahoma, formed a distinct clade with high support. In our initial phylogenetic analyses using Bayesian inference, the isolates later attributed to C. noveboracense formed a distinct clade in a three-gene multi-locus analysis (ITS, TUB2, ApMat) with full support (BI PP 1.0). Additionally, in our initial Bayesian analysis of seven loci (ACT, TUB2, CAL, GAPDH, GS, ITS and ApMat) and other different combinations of loci, C. noveboracense was sister to C. nupharicola (PP = 0.95). C. nupharicola is easily distinguished within the CGSC in terms of morphology. This host-specific species has very slow growth on PDA and both the length and width of the conidia are much greater than other species in CGSC 53,77 . The morphological differences between C. noveboracense and C. nupharicola prompted us to expand the analysis to include a much larger dataset, include an additional locus (APN2) known to provide better resolution in CGSC 53 , and evaluate the new clade under GCPSR criteria 78 . This led us to identify these isolates as a strongly supported clade, distinct from other taxa in CGSC.
Fungi have developed a plethora of adaptive mechanisms, including extracellular enzyme secretion 79 . In our study, using the skimmed milk agar plates to detect proteolytic activity, it was impossible to observe visible halos of degradation for the assessed isolates. Several possibilities might contribute to the lack of visualization of proteolytic activity in Colletotrichum species. First, the ability and the level of protease gene expression in fungi could differ based on the nutrient source used in agar medium. Aspergillus isolates showed ability to produce proteases in agar medium supplemented with gelatin and casein as two different sources of protein 80 . Second, the difference in range of pH in culture medium also affects the proteolytic activities 80 . Finally, although the degradation halo indicating the protease activity in C. fructicola isolates was detected easily on skimmed milk agar plate 55 , sometimes the detection of the degradation zones is not possible unless a developing agent like bromocresol green dye is used 81 . The three Colletotrichum species in our study showed different level of amylolytic, cellulolytic and lipolytic activities. Prior to the present work, only a few studies investigated the enzyme activity of www.nature.com/scientificreports www.nature.com/scientificreports/ Colletotrichum isolates. C. fructicola isolates causing bitter rot and leaf spot on apple in Brazil were compared for their ability to produce amylolytic, pectolytic, lipolytic and proteolytic activity, and showed higher amylolytic and pectolytic activity compared to the isolates causing leaf spot, while they were the same in lipolytic and proteolytic activity 55 . Our results show species variation in enzymatic activity, which might be related to variable ability of different Colletotrichum species to effectively penetrate and spread in host plant tissues and the higher level of virulence 55,82 . This hypothesis must be further evaluated by investigating the contribution of these enzymes in pathogenicity.
To control bitter rot disease, applications of different fungicides are recommended. We observed statistically different fungicide sensitivity between and within the complexes in our study, which is supported by the previous studies on apple where the CASC was more tolerant to thiophanate-methyl, myclobutanil, trifloxystrobin, captan and demethylation inhibitor (DMI) fungicides in comparison to the CGSC 17,83-85 . In addition, within the CASC, isolates from apple orchards in Brazil showed different levels of sensitivity (25-83%) to mancozeb, thiophanate-methyl and azoxystrobin 62 . All the fungicides in our study showed high mycelial growth inhibition against all the species. Several studies support our findings. Benzovindiflupyr was highly active against mycelial growth of C. gloeosporioides, C. acutatum, C. cereale and C. orbiculare with EC 50 values lower than 0.1 μg/ml 86 . Similar efficiency of this fungicide was seen in germination of conidia and germ tube growth of isolates with EC 50 values 0.1 and 1 μg/ml 86 . Few isolates belonging to C. fioriniae, C. fructicola and C. siamense were sensitive to fludioxonil, and benzovindiflupyr with EC 50 values < 0.1 μg/ml and <0.1 to 0.33 μg/ml, respectively 87 . In our study, in vitro toxicity of natamycin with an EC 50 of 5 μg/ml against Colletotrichum species was significantly lower than that of the other fungicides with EC 50 values ranging from 0.04 to 0.4 μg/ml. This is consistent with the previous work in which the toxicity of natamycin against mycelial growth of C. acutatum ranged from 0.5 to 1.9 μg/ ml in EC 50 values and was considerably lower compared to fludioxonil, azoxystrobin and cyprodinil 85 . Our data show strong in vitro activity of fungicides used in this study and likely would provide effective control of bitter rot in orchards or storages. Although the susceptibility profiles of Colletotrichum species against fungicides in apple orchards across the United States are limited, the efficacy of benzovindiflupyr and pyraclostrobin against bitter rot and GLS was reported in two recent trials in North Carolina 88,89 . Future studies should continue to validate the effectiveness of these and other fungicides against apple bitter rot.
In conclusion, three Colletotrichum species, C. fioriniae, C. chrysophilum and a novel species C. noveboracense, were identified as the causal agent of apple bitter rot in New York. Also, our study for the first time describes C. chrysophilum as the causal agents of bitter rot on apple in Virginia and Pennsylvania and C. noveboracense in Pennsylvania. We determined that the three species varied in pathogenicity, enzyme activity and fungicide sensitivity, which are important characteristics for bitter rot management. Our results highlight the significance of accurate identification of Colletotrichum species causing bitter rot in apple production regions in order to manage this economically important disease and secure the profitability of apple industry.

Methods
Sample collection and fungal isolation. In 2017 and 2018, apple fruit with typical symptoms of bitter rot disease were collected from a variety of apple cultivars in commercial and private apple orchards in the Hudson Valley area, New York (Table 2). Around 400 Colletotrichum isolates were obtained from apple fruit disinfected with 5% bleach for 2 min and rinsed with sterile distilled water. After removing the peel around the lesion, three small pieces of fruit pulp cut from the margin of each lesion were placed onto potato dextrose agar (PDA, Difco Laboratories, Detroit, MI, US). Plates were stored at 25 °C in the dark and colonies were purified by hyphal tip method.

Selection of isolates for molecular analysis.
Besides sample collection from New York, we also received bitter-rot infected apple fruit from commercial orchards in Pickerel and Cana, Virginia and Thurmond, North Carolina in 2017 (provided by Virginia Tech Research Station, Winchester, VA) and isolates from Pennsylvania State University's Fruit Research and Extension Center in Biglerville, PA, for identification and comparison. Moreover, two isolates (Cg)RR12-1 and (Cg)RR12-3 identified as C. fructicola recovered from peach fruit 49 were received from School of Agricultural, Forest and Environmental Sciences, Clemson University, SC, for re-identification and comparison. All isolates collected in this study were placed into two morphological typesbased on growth rate, colony texture and color, sporulation and conidial shape on PDA. In total, 44 isolates (31 from New York and 13 from other states) from the two morphologically distinct typeswere selected based on geographical distribution and apple cultivar for identification to the species complex using ITS sequencing and multiplex PCR assay, and subsequently to the species level using multi-locus phylogenetic analyses (Table 2). Isolates collected from New York were used for enzyme activity assay, fungicide sensitivity and pathogenicity test.
Multiplex PCR assay. DNA from mycelia of 7-day-old Colletotrichum cultures was extracted using the DNeasy Plant Mini Kit (Qiagen Inc., Valencia, CA, USA) according to the manufacturer's instructions. A multiplex PCR assay was performed to differentiate isolates of the CGSC and CASC by partial amplification of the GAPDH and CAL genes using primer pairs GDF1/C-GAPDH-R, CALF1/Cg-R, and CALF1/Ca-R1 49     www.nature.com/scientificreports www.nature.com/scientificreports/ in Table 3 Phylogenetic analyses. Consensus sequences were obtained by assembling forward and reverse reads using Geneious Pro v. 11.1.4 90 . In order to confirm the placement of the isolates within species complexes, ITS sequences collected from 44 isolates and references from representatives of each of the nine major clades 10 were used to construct the ITS phylogeny. To evaluate the placement of isolates at the species level, the C. acutatum phylogeny was constructed using three loci (ITS, TUB2 and GAPDH), whereas the C. gloeosporioides phylogeny was constructed using eight loci (ACT, CAL, GAPDH, GS, ITS, ApMat, APN2 and TUB2).
All three phylogenies were constructed using Bayesian inference (BI) and maximum likelihood (ML) approaches. Reference sequences (Supplementary Table S2) were downloaded from GenBank and aligned using MAFFT v7 on-line 91,92 , specifying the G-INS-i iterative refinement strategy. The alignments were trimmed using Gblocks v0.91b 93 specifying the less stringent criteria. Model selection was conducted using PartitionFinder 2 94 , specifying the Greedy algorithm 95 , "MrBayes" models for BI using PhyML 96 , or "all" models for ML analysis using RAxML 97 and the AICc metric. Bayesian inference was conducted using MrBayes v3.2.6 98 implementing the BEAGLE library 99 . For the ITS and C. acutuatum phylogenies posterior probabilities were estimated using two runs of 2,000,000 generations with 25% burn-in. For the C. gloeosporioides phylogeny, 10,000,000 generations were used. ML analysis was conducted using RAxML v8.2.12 97 (Table 2) and the taxonomic novelty in MycoBank. The alignment files and trees were deposited in TreeBase (http://purl.org/phylo/treebase/ phylows/study/TB2:S25647).

Species delimitation.
Initial phylogenetic analyses revealed that several CGSC isolates collected in this study, in addition to isolate Coll940 53 , clustered together with high support and were distinct from any clade containing the ex-type of any previously described species, suggesting that these isolates may represent a novel lineage. In order to determine whether this new clade formed a distinct phylogenetic lineage, we applied GCPSR 78 . In this approach, a clade is determined to represent an independent evolutionary lineage if the clade satisfies one of two criteria: genealogical concordance or nondiscordance. The genealogical concordance criterion is satisfied if the clade is found well-supported (e.g. both ML and BI analysis ≥70% and ≥0.95, respectively) in most individual gene trees. The nondiscordance criterion is satisfied if the clade is found well-supported in at least one gene tree and members not found strongly supported in contradictory placement (e.g. clustering with the type isolate of another species) in any other individual gene trees.  www.nature.com/scientificreports www.nature.com/scientificreports/ To apply the GCPSR approach, individual gene trees were constructed for each of the eight genes used in the multi-locus C. gloeosporioides phylogeny. Evolutionary model selection and gene tree constructed were as described above, except that 5,000,000 generations were used to infer posterior probabilities for the Bayesian approach. Placement of clade members in each Bayesian and ML tree were evaluated for each individual gene tree.
Morphological characterization. Colony color, growth rate, conidial shape, length and width of Colletotrichum spp. in this study was evaluated by transferring 4-mm diameter plugs from the periphery of 5-day-old cultures, grown at 25 °C in dark, onto PDA and ½ strength PDA. Colony color was described after 7 days of incubation on PDA at 25 °C in dark. Colony growth rate was determined by measuring the colony diameter of each isolate grown on PDA daily over the course of 7 days at 25 °C in dark. To study the morphology of isolates belonging to the novel species, slide culture technique 108 was used to induce the isolates to produce appressoria. Synthetischer nahrstoffarmer agar (SNA i.e. synthetic nutrient-poor medium) 109 and oatmeal agar (OMA) 110 were used to induce sporulation.
Microscopic observations, with 25 measurements per each structure, were viewed with an Olympus BX51 microscope (Olympus Corporation of the Americas, Center Valley, PA, US) using the differential interference contrast (DIC) setting. Statistical analysis was conducted by one-way analyses of variance (ANOVA) using Graph Pad Prism software v5 (San Diego, CA, U.S.A).
Agar-plate enzyme activity. To perform the qualitative enzyme activity, isolates were grown on PDA at 25 °C for 7 days in the dark. For lipolytic and proteolytic activities, we transferred a mycelial plug from the growing part of each colony onto peptone agar medium (10 g peptone, 5 g NaCl, 0.1 g CaCl 2 2H 2 O, 15 g agar, pH 6.0) supplemented with 1% Tween 20 111 and onto PDA containing 1% soluble skim milk 112 , respectively. After five days of incubation at 25 °C in dark, the size of the clear zone indicating lipolytic and proteolytic activity around each colony was measured in millimeters (mm) using a caliper. For amylolytic activity, isolates were transferred to starch hydrolysis agar medium (pH 7) and kept at 25 °C for 5 days in dark 111 . After flooding with 1 ml of Gram Iodine solution, the clear halo around each colony was measured. Isolates were cultured on PDA supplemented with 0.5% carboxy-methylcellulose (CMC) for 5 days at 25 °C in dark for cellulolytic activity. The plates were treated with 1% Congo red solution and shaken for 15 min. After removing Congo red, cultures were treated with 1 M NaCl and shaken for 15 min. Subsequently, clear zones indicating cellulolytic activity were measured 113 . Three replicates per each isolate was used. Data were analyzed by one-way ANOVA with Graph Pad Prism software v5 (GraphPad Software, San Diego, CA, US).
Pathogenicity assay. Pathogenicity of all isolates was first tested on apple fruit of cultivar 'Honeycrisp' to reproduce bitter rot symptoms. Later, six Colletotrichum isolates from each species were inoculated on the apple fruit of cultivars 'Golden Delicious' , 'Honeycrisp' , 'Red Delicious' , 'Fuji' and 'Gala' obtained from a grocery store and washed with detergent and water to ensure that no fungicide residues remain on the surface. Three fruit per each cultivar were disinfected for 2 min in 5% bleach, rinsed twice with sterile distilled water and then wounded with a 3-mm corkborer 43 . Two opposite sides of each fruit were inoculated with 3-mm mycelial plugs of each isolate with aerial mycelia facing the flesh. Control apple fruit received uninoculated agar plugs. Plastic boxes containing inoculated apple fruit placed on moist paper towels were incubated at 25 °C in the dark. Lesion diameter was measured 15 days after inoculation . Data were analyzed by two-way ANOVA with Bonferroni Posttest using Graph Pad Prism software v5 (GraphPad Software, San Diego, CA, US). P-values ≤ 0.05 were considered significant. To fulfill Koch's postulates, strains were re-isolated and morphologically re-identified.