Aspergillus korhogoensis, a Novel Aflatoxin Producing Species from the Côte d’Ivoire

Several strains of a new aflatoxigenic species of Aspergillus, A. korhogoensis, were isolated in the course of a screening study involving species from section Flavi found contaminating peanuts (Arachis hypogaea) and peanut paste in the Côte d’Ivoire. Based on examination of four isolates, this new species is described using a polyphasic approach. A concatenated alignment comprised of nine genes (ITS, benA, cmdA, mcm7, amdS, rpb1, preB, ppgA, and preA) was subjected to phylogenetic analysis, and resulted in all four strains being inferred as a distinct clade. Characterization of mating type for each strain revealed A. korhogoensis as a heterothallic species, since three isolates exhibited a singular MAT1-1 locus and one isolate exhibited a singular MAT1-2 locus. Morphological and physiological characterizations were also performed based on their growth on various types of media. Their respective extrolite profiles were characterized using LC/HRMS, and showed that this new species is capable of producing B- and G-aflatoxins, aspergillic acid, cyclopiazonic acid, aflavarins, and asparasones, as well as other metabolites. Altogether, our results confirm the monophyly of A. korhogoensis, and strengthen its position in the A. flavus clade, as the sister taxon of A. parvisclerotigenus.


Introduction
The presence of mycotoxins in agricultural commodities poses serious economic and health risks [1][2][3]. Among the mycotoxins, aflatoxins are by far the most studied since their ingestion can cause deleterious health effects in humans and animals including hepatic cancer and, in some instances,  [4]. Aflatoxin B 1 is the potent compound of this chemical family as it displays mutagenic, teratogenic and hepatocarcinogenic effects in humans and animals [5]. To date, it is considered as the most carcinogenic, teratogenic and genotoxic substance of natural origin [6,7]. Species that have so far been reported to produce aflatoxins are all classified in Aspergilllus subgenus Circumdati and section Flavi, with the exception of two other species originally sampled in the Côte d'Ivoire, A. ochraceoroseus and A. rambelli [8,9]. Species from Aspergillus section Flavi represent a well-known group of saprophytic filamentous fungi, several of which have the ability to produce beneficial secondary metabolites or enzymes used in food fermentation and biotechnology, such as kojic acid and α-amylase [10]. Conversely, some of these species have the potential to produce one or more harmful mycotoxins, such as aflatoxins, cyclopiazonic acid, versicolorins, and aflatrems [11,12]. Due to extensive research into their aflatoxin production, A. flavus, A. parasiticus, and A. nomius are considered major species in section Flavi. Among them, A. flavus is the most important because of its worldwide distribution, and it represents the largest source of aflatoxin B 1 contamination of several staple crops, including maize, tree nuts, peanuts, cottonseed, grains, cassava, and spices [13][14][15][16].
Although some species relationships in section Flavi are still unclear, the section can be separated into one of seven main clades based on a polyphasic approach: A. flavus clade, A. parasiticus clade, A. tamarii clade, A. nomius clade, A. alliaceus clade, A. togoensis clade, or A. avenaceus clade, and A. mottae and A. bertholletius [17][18][19]. Each clade may contain cryptic species that are difficult to identify, based solely on morphological characters or extrolite profiles, but can be delineated using a polyphasic approach that also includes molecular analyses [12,17,18]. It has been suggested that two cryptic species, A. minisclerotigenes and A. parvisclerotigenus, belong to the A. flavus clade [9,20]. However, in addition to B-aflatoxins these species produce G-aflatoxins, which A. flavus is incapable of producing [17,21] due to a deletion of genomic sequence between the aflatoxin pathway genes, norB (aflF) and cypA (aflU), thereby altering the promoter and coding regions [22].
In this paper, we describe A. korhogoensis sp. nov. as a novel cryptic species within the A. flavus clade, based on a polyphasic analysis of four strains isolated from peanuts collected in the region of Korhogo, Côte d'Ivoire.

Multilocus Phylogenetic Analysis
The phylogenetic tree inferred from nine concatenated genes (ITS, benA, cmdA, mcm7, amdS, rpb1, preB, ppgA, and preA), obtained from Bayesian and ML analyses, yielded largely similar topologies, particularly congruent for Aspergillus flavus clade. Here, we chose the Bayesian topology as hypothesis of phylogenetic relationships because the results were of greater robustness ( Figure 1). Our results support previous phylogenetic inferences involving species from section Flavi. Aspergillus bertholletius was used as the outgroup taxon. The A. nomius clade, which included A. bombycis and A. nomius, was monophyletic and appeared as a basal group (Posterior Probability, PP = 1). The topology then split in two robust groups, one formed by the A. tamarii clade, which included A. caelatus, A. pseudocaelatus and A. pseudotamarii, and a second monophyletic group, which included the A. parasiticus and A. flavus clades, as well as A. mottae. This latter was placed as the ancestral taxon of the group including A. parasiticus and A. flavus clades. Aspergillus parasiticus clade was consistent with Soares et al. results [18] and included A. parasiticus, A. sojae, A. arachidicola, A. novoparasiticus, A. sergii and A. transmontanensis. A. sergii and A. transmontanensis are basal taxa, respectively. The A. flavus clade included A. flavus, A. oryzae, A. minisclerotigenes, A. parvisclerotigenus and A. korhogoensis, the herein described new species.
The A. flavus clade is comprised of two main groups: one that includes A. flavus, its domesticated species A. oryzae and A. minisclerotigenes (PP = 1); and the other group encompasses A. parvisclerotigenus and A. korhogoensis sp. nov. (PP = 1). The four isolates (MACI46, MACI219, MACI254 and MACI264) putatively identified as A. korhogoensis were tightly clustered, suggesting they were a distinct species from A. parvisclerotigenus. Strains from the latter species included isolates from different populations, which were clustered together, suggesting a monophyletic group with no major differences among populations.
Aspergillus novoparasiticus was represented by two well-supported groups, which segregated based on geography since one group corresponds to the strains isolated from Brazil [23], while the other group corresponds to strains isolated from Benin [14]. Besides their geographical distributions, both groups were isolated from different environments, such as hospital environments (Brazil samples) or foodstuffs (Benin samples).

Mating Type Analysis
Results from our mating type diagnostic PCR revealed that isolates MACI46, MACI254 and MACI264 contained a single Mat1-1 gene, and that isolate MACI219 contained a single Mat1-2 gene. These findings demonstrate that A. korhogoensis sp. nov. is likely a heterothallic (self-infertile) fungus. Whether these mating-type genes are functional is unknown. Future mating tests will be required to determine this.
The MAT1-1 amino acid sequence of A. bertholletius was used as the reference sequence to compare with Mat1-1 genes from other examined taxa. Basal taxa (A. bertholletius, A. nomius, A. caelatus, A. pseudocaelatus, A. tamarii and A. pseudotamarii) presented alanine, asparagine, histidine and threonine at position 36, 46, 61 and 65, respectively, which changed in derived species into serine, lysine, asparagine and asparagine ( Figure 2). The A. tamarii clade presented four apomorphies that are specific to the clade, and one apomorphy that was fixed in the derived species. The A. parasiticus and A. flavus clades exhibited a highly conserved MAT1-1 amino acid sequence, except for one amino acid substitution in A. parasiticus at position 101, and two substitutions in both A. flavus and A. oryzae at positions 49 and 75. Although haplotypes of A. minisclerotigenes, A. parvisclerotigenus and A. korhogoensis sp. nov. shared identity for their respective MAT1-1 amino acid sequences (Figure 2), there were single nucleotide polymorphism (SNP) differences that did not result in an amino acid replacement. Aspergillus minisclerotigenes exhibited two apomorphies, whereas A. parvislerotigenus and A. korhogoensis sp. nov. each exhibited only one (except MACI264, which exhibited the conserved ancestral state).
In the case of Mat1-2 gene, amino acid sequences were identical for A. nomius, A. pseudonomius, A. sergii, A. transmontanensis, A. arachidicola, A. parasiticus, A. parvisclerotigenus and A. minisclerotigenes. There was one substitution in the amino acid sequence for A. korhogoensis sp. nov. (S168P), and one for A. flavus (E181K) (Figure 3). Basal taxa, A. avenaceus and A. alliaceus, exhibited several differences in their amino acid sequences.

Secondary Metabolism Characterization
An analysis of secondary metabolites produced by the four A. korhogoensis strains was performed and the results are summarized in Table 1. Metabolites were identified according to the Metabolomics Standard Initiative level definitions [24]. Metabolites were identified at level 1 when they displayed the same retention time, and UV and MS/MS spectra as the authentic standard. They were identified at level 2 when the metabolites shared the same UV spectrum and/or the same MS/MS fragmentation pattern in accordance with the literature.    Table  S2.   Table S2 except for Aspergillus parvisclerotigenus MACI5 (MF966968). Aspergillus parvisclerotigenus MACI5 species identification was based on genomic sequences from ITS (KY689161), benA (KY628772) and cmdA (KY661269). The A. sergii sequence available from GenBank is shorter.   Table S2 except for Aspergillus parvisclerotigenus MACI5 (MF966968). Aspergillus parvisclerotigenus MACI5 species identification was based on genomic sequences from ITS (KY689161), benA (KY628772) and cmdA (KY661269). The A. sergii sequence available from GenBank is shorter. All strains produced aflatoxins B 1 , B 2 , G 1 , and G 2 , as well as several aflatoxin biosynthetic intermediates including 3-O-methylsterigmatocystin, sterigmatocystin and versicolorins A and B. Cyclopiazonic acid (CPA) and different other members of the CPA family were detected in the whole fungal extract of each strain. This new species also produced kojic acid, aflatrem, and its precursors or related compounds (paspaline, paspalinine, paxilline, and 13-desoxypaxilline). Aflatrem and paspaline were also present in the sclerotium extracts, as well as another related compound that appears to be an uncharacterized hydroxylated form of aflatrem ([M + H] + /z = 518.28992; deviation = −0.347 ppm) previously evoked by Nicholson et al. [33]. Leporin B and its precursor were previously detected in culture extracts [25].
The HPLC-DAD analysis of sclerotium extracts revealed the presence of members of at least three families. First, five compounds showed a typical anthraquinone UV spectrum (nm): 223 (100%), 269sh, 293, 319sh, 455. On the bases of UV and MS/MS fragmentation patterns, this compound was identified as asparasone A [31,34]. Three other asparasone-derived anthraquinones were identified in sclerotium extracts (Table 1). Six compounds displayed an aflavarin UV spectrum (nm): 221 (100%), 238sh, 291, 310, 322sh [35]. The LC-HRMS allowed the identification of aflavarin and four aflavarin-associated compounds previously reported by Cary et al. [32]. The last chemical family identified in A. korhogoensis sclerotia were aflavinines [36,37]. Indeed, three compounds with a typical aflavinine UV spectrum (nm): 224 (100%), 283, and 291, were present in the sclerotium extracts of each tested strain. The comparison with the 20-hydroxyaflavinine standard confirmed the presence of this metabolite. All strains produced aflatoxins B1, B2, G1, and G2, as well as several aflatoxin biosynthetic intermediates including 3-O-methylsterigmatocystin, sterigmatocystin and versicolorins A and B. Cyclopiazonic acid (CPA) and different other members of the CPA family were detected in the whole fungal extract of each strain. This new species also produced kojic acid, aflatrem, and its precursors or related compounds (paspaline, paspalinine, paxilline, and 13-desoxypaxilline). Aflatrem and paspaline were also present in the sclerotium extracts, as well as another related compound that appears to be an uncharacterized hydroxylated form of aflatrem ([M + H] + /z = 518.28992; deviation = −0.347 ppm) previously evoked by Nicholson et al. [33]. Leporin B and its precursor were previously detected in culture extracts [25].

Aspergillus korhogoensis
The HPLC-DAD analysis of sclerotium extracts revealed the presence of members of at least three families. First, five compounds showed a typical anthraquinone UV spectrum (nm): 223 (100%), 269sh, 293, 319sh, 455. On the bases of UV and MS/MS fragmentation patterns, this compound was identified as asparasone A [31,34]. Three other asparasone-derived anthraquinones were identified in sclerotium extracts (Table 1). Six compounds displayed an aflavarin UV spectrum (nm): 221 (100%), 238sh, 291, 310, 322sh [35]. The LC-HRMS allowed the identification of aflavarin and four aflavarinassociated compounds previously reported by Cary et al. [32]. The last chemical family identified in A. korhogoensis sclerotia were aflavinines [36,37]. Indeed, three compounds with a typical aflavinine UV spectrum (nm): 224 (100%), 283, and 291, were present in the sclerotium extracts of each tested strain. The comparison with the 20-hydroxyaflavinine standard confirmed the presence of this metabolite.      Etymology: The specific epithet "korhogoensis" is a noun in the genitive case and refers to the Korhogo region located in the Côte d'Ivoire, from where the new species was isolated.

Aspergillus korhogoensis
Diagnosis: Colonies on MEA deeply floccose with a dominant white aerial mycelium. Sporulation dull yellowish green. Abundant sclerotia (especially on MEA and CYA), mostly at the colony surface, small size (<400 µ m), dark brown at mature state; conspicuous amber exudate produced by sclerotia. Reverse orange to brownish orange, more conspicuous on MS, and on MEA    Etymology: The specific epithet "korhogoensis" is a noun in the genitive case and refers to the Korhogo region located in the Côte d'Ivoire, from where the new species was isolated.
Diagnosis: Colonies on MEA deeply floccose with a dominant white aerial mycelium. Sporulation dull yellowish green. Abundant sclerotia (especially on MEA and CYA), mostly at the colony surface, small size (<400 µ m), dark brown at mature state; conspicuous amber exudate produced by sclerotia. Reverse orange to brownish orange, more conspicuous on MS, and on MEA    Etymology: The specific epithet "korhogoensis" is a noun in the genitive case and refers to the Korhogo region located in the Côte d'Ivoire, from where the new species was isolated. Diagnosis: Colonies on MEA deeply floccose with a dominant white aerial mycelium. Sporulation dull yellowish green. Abundant sclerotia (especially on MEA and CYA), mostly at the colony surface, small size (<400 µm), dark brown at mature state; conspicuous amber exudate produced by sclerotia. Reverse orange to brownish orange, more conspicuous on MS, and on MEA and CYA presence of concentric rings on orange shades. Conidial heads typically radiate, fertile upper 75% of their surface and splitting, less frequent narrow and long columnar to short columnar, rarely micro-heads. Conidiophores of radiate heads are hyaline, long, large and slightly roughened, whereas conidiophores of columnar heads and micro-heads are short, narrow and smooth. Conidial heads biseriate for radiate heads, and uniseriate for the others. Vesicles oblong to spatulate, 25-47 µm in diam; metulae 6.7-11.2 µm X 4-5.5 µm; phialides 7-10 µm X 3-5.7 µm; conidia yellowish green to green, oblong and smooth or slightly rough, 3-5 µm diam. Physiological studies: All strains analyzed on AFPA showed a bright orange reverse, a sign of aspergillic acid production. The colonies did not sporulate and presented reduced aerial mycelia. On CREA, the strains showed a positive production of organic acids, except for MACI219.
Aspergillus korhogoensis sp. nov. exhibited phenotypic characters that place it within the A. flavus clade, such as conidial heads typically radiate that split into several columns in green shades. Phenotypically, the new species resembled A. parvisclerotigenus. Both species shared several common traits, making difficult to distinguish between them; however, some subtle differences were observed. The new species grew faster on MEA and CYA at 25 • C than A. parvisclerotigenus, and the reverse coloration on MEA and MS was orange for A. korhogoensis and cream for A. parvisclerotigenus. The size of sclerotia was also comparatively smaller in A. korhogoensis. On MEA at 42 • C, A. parvisclerotigenus grew faster (15.5-20 mm) than A. korhogoensis, for which growth was reduced (7.5-12 mm) or inexistent in strain MACI219. On AFPA, A. parvisclerotigenus colonies were mildly to highly floccose, produced profuse sclerotia and conidia in yellowish shades, whereas A. korhogoensis sp. nov. colonies exhibited sparse aerial mycelium and sclerotia, and conidia were almost non-existent.
Holotype: Isolated from Gbandokaha. Deposited in the NRRL collection.

Discussion
The number of species in section Flavi, in direct correlation with the number of species capable of producing aflatoxins, has increased over the last decade [17][18][19]23]. Although unable to produce aflatoxins, a 26th species, A. hancockii sp. nov., was very recently identified and grouped with section Flavi species [38]. The use of a polyphasic approach to characterize a species, based on the unified species concept [39,40], has enabled mycologists to acknowledge cryptic diversity in Aspergillus section Flavi. By using this approach, morphological, physiological, and molecular characters are integrated to understand species relationships [12,[17][18][19][20]23,41]. It is accepted that the use of a single approach will mask the exact relationships among species, not only in Aspergilli, but also in other fungi [12].
In the present study, we included an ensemble of six genomic regions already tested to be informative for section Flavi (ITS, benA, cmdA, mcm7, rpb1 and amdS). Three other genes (ppgA, preA, preB), reportedly involved in sexual development, were added and the set resulted in a concatenated sequence of 4624 bp. These three genes are required in the specific mating recognition. PreA and preB are MAT target genes that encode a-pheromone and α-pheromone receptors, respectively [42]. PpgA encodes the α-pheromone precursor that binds to PreB [43]. To our knowledge, this is the first study to make a phylogenetic inference that includes genes PreA, PreB and PpgA. Some authors have suggested that the accuracy of the phylogenetic approach could be increased by adding molecular data with different evolutive rates, diminishing possible artifacts caused by polymorphic haplotypes [44,45] and providing more robust information to elucidate potential complexity within a clade. The use of nine concatenated genes resulted in a robust phylogenetic tree topology, which includes the most important species of the section in terms of economic and public health impact, as well as species described in the last decade (not basal taxa). The results of the present study were congruent with several studies performed involving species from section Flavi [17][18][19][20]23]. The tree obtained from nine concatenated genes showed a clear partition of A. novoparasiticus strains in two, one subgroup containing South American isolates and another containing African isolates. More A. novoparasiticus strains isolated from both continents would be needed in order to confirm this observation. Additionally, the ensemble allowed to determine a cryptic species, A. korhogoensis sp. nov.
Most species in section Flavi are considered heterothallic, containing either the MAT1-1 or MAT1-2 idiomorph [15,18,46], and from the present study so is A. korhogoensis. Thus far, A. alliaceus is the only homothallic species in section Flavi [46]. Reportedly, strains of A. nomius may contain both idiomorphs, but only one is functional [11]. Diversity from sexual reproduction in these fungi is expected to arise from out-crossing of heterothallic species, between complementary strains that are able to produce sclerotia [47,48]. Laboratory crosses between sexually compatible strains showed that sexual reproduction was possible in A. parasiticus, A. flavus, and A. nomius [11,49,50]. Inter-specific hybridization was shown to be a possibility via laboratory crosses that resulted in viable ascospores, including recombinant offspring, being produced [51]. Moreover, sexual reproduction is more likely to occur within populations having a 1:1 ratio of both idiomorphs [11,42,48], although asexual reproduction is still a large component to the life cycle of micro-fungi such as the Aspergilli [52]. Presence of both idiomorphs in A. parvisclerotigenus, and A. korhogoensis suggests that cryptic sexuality might occur in natural populations, yet laboratory mating experiments involving these species is necessary to yield conclusions. As well, more population-scale field sampling of strains from these species are necessary to determine if they have a history of recombination as observed in A. flavus and A. parasiticus populations [53,54]. Indeed, the ratio 1:1 is not discernable due to the few strains isolated and curated in different collections.
The present study increases the number of species in the A. flavus clade, which is comprised of many heterothallic species that share common morphological characters, such as biseriate heads, greenish to brownish colony coloration, ability to produce sclerotia, among others [17]. Likewise, species in this clade are capable of producing aflatoxins, aspergillic acid, CPA, kojic acid, versicolorins, aflatrem, etc. [20,55]. Within the clade, the main difference between A. flavus and the remaining species (A. minisclerotigenes, A. parvisclerotigenus and A. korhogoensis sp. nov.), is that A. flavus has lost the ability to produce G-aflatoxins [22]. Another important difference is that A. flavus is comprised of two morphotypes: small sclerotium producers and those that are able to produce larger sclerotia than the other three species. Aspergillus flavus is a ubiquitous fungus, being readily sampled across the globe, but A. minisclerotigenes, A. parvisclerotigenus, and A. korhogoensis sp. nov. have smaller geographic distributions. For example, A. minisclerotigenes has been isolated from Africa, South and North America, Europe, and Australia, whereas A. parvisclerotigenus has been isolated from Guinea Gulf [16], and A. korhogoensis has only been found in the Côte d'Ivoire.
Herein, we proposed that Aspergillus korhogoensis sp. nov. is the sister taxon of A. parvisclerotigenus, based on secondary metabolite analyses, morphology and molecular evidence. Both species share a similar secondary metabolic profile according to Frisvad et al. [9]. However, there are some differences in their secondary metabolite production. Unlike A. parvisclerotigenus, the production of A-30461 (aspirochlorin) was not observed in any A. korhogoensis extracts. On the other hand, A. korhogoensis produced aflavinines, asparasones and leporin B. A pattern close to each other could be appreciated while comparing morphological characters, though subtle differences were observed between strains of both species. On AFPA, A. parvisclerotigenus had a trend to produce yellowish spores and highly floccose colonies, whereas A. korhogoensis sp. nov. tended to have flatter colonies and reduced sporulation. At the molecular level, the concatenation of nine different loci strongly suggested that they were two different species. The inclusion of strains from different populations of A. parvisclerotigenus (Benin, Nigeria and Côte d'Ivoire) reduced the possibility of any artifact linked with polymorphisms.
Sub-Saharan West Africa, and especially the Guinea Gulf, displays an interesting diversity of species from section Flavi, including several cryptic species, although A. flavus continues to be the most frequent species sampled [56]. Despite the prevalence of A. flavus, it is noteworthy that other S-strain species are present at lower rates, are usually G-aflatoxin producers, and their production of aflatoxins is usually higher than that of A. flavus sensu stricto [57,58]. Some strains, previously characterized as A. flavus S BG , are nowadays being classified as A. minisclerotigenes, A. parvisclerotigenus, and in this study as A. korhogoensis. In different countries of the Guinea Gulf, strains exhibiting the S BG chemotype have been associated with drier agroecological zones bordering the Sahara desert [14,59,60]. Cadwell and Cotty [59] suggested that production of G-aflatoxins in Northern Benin could be mainly due to S BG strains. Likewise, in this area, the environmental conditions could allow the presence of species that could be more sensitive to climate changes, resulting in a shift of the frequency of these strains [59]. Inter-specific sex is possible for these fungi, which has been shown via laboratory crosses [51]. All that is required for these strains to override heterokaryon incompatibility, or even species boundaries, is the need to circumvent an unfavorable environmental situation [61]. It may eventually be determined that many of the recently characterized novel species, with such similar morphological, genetic, metabolic and physiological characteristics, are hybrids resulting from cryptic inter-specific sex that comprise a species complex. However, much more research is required before this can be proven or refuted. Moore and co-workers [62][63][64] are sequencing the genomes of aflatoxigenic fungi in an effort to determine the relatedness of these fungi and to elucidate the evolution of aflatoxin production. The comparison of the genomes of species close to A. flavus such as A. minisclerotigenes, A. parvisclerotigenus and A. korhogoensis will also help to understand the genetic determinants of the A. flavus success.

Chemicals
Solvents (phenol, chloroform, ethanol, ethyl acetate, methanol, and acetonitrile) of analytical grade used in the extraction and high-performance liquid chromatography (HPLC) were obtained from ThermoFisher Scientific (Illkirch, France). Ultrapure water used for HPLC with Diode Array Detector (DAD), LC/MS analyses, and for molecular biology experiments was purified from a MilliQ purification system (Millipore, Billerica, MA, USA). Unless otherwise specified, chemicals were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France).

Fungal Isolates and Culture Conditions
Seven atypical S BG A. flavus strains were isolated in 2014 from peanuts in the Northern Korhogo region of Côte d'Ivoire. Four other atypical S BG strains isolated from food, decaying leaves, and logs of wood collected in Southwest Nigeria [65] were added. To identify these eleven S BG strains to species level, we compared them against a dataset comprised of strains obtained from different international collections and stored under controlled conditions at Research Center in Food Toxicology TOXALIM, Toulouse. We included at least one strain belonging to most species within the section Flavi (Table 3).
The isolates were cultured on Malt Extract Agar (MEA) (Biokar Diagnostics, Allone, France) at 25 • C for seven days, and stored as spore suspensions on 20% glycerol for further analyses. The A. korhogoensis Type strain, along with three other strains, were deposited at Agricultural Research Service Culture Collection (NRRL) (Peoria, IL, USA).

DNA Extraction and Amplification
A loopful of spores from each of the examined Aspergillus isolates was inoculated on Yeast Extract Sucrose (YES) liquid medium, and kept in agitation in an orbital incubator at 170 rpm at 27 • C for five days. DNA extraction was performed according to Girardin et al. [71] by grinding a portion of mycelium in a 5 mL mortar on ice, followed by the addition of 5.5 mL lysis buffer 2 (5 mL Tris-HCL 1 M, 3.65 g NaCl, 12.5 mL EDTA 0.5 M pH 8, 2.5 g SDS, H 2 O qsp 250 mL). The content was transferred to a 15 mL tube, and 12.5 µL of proteinase K were added, before being incubated from 30 min to 1 h at 37 • C, and then incubated for 10 min at 65 • C. Afterwards, one volume of phenol/chloroform (7:3, v/v) was added and samples were then centrifuged at 3000× g for 1 h. The supernatant was recovered into a new tube, where 6 µL RNAse were added, and it was subsequently incubated for 2-3 h at 37 • C. Next, one volume of chloroform was added and centrifuged at 3000× g for 10 min. The supernatant was recovered into a new tube and one volume of isopropanol was added. At this point, samples were softly shaken for 2 h in a horizontal shaker and kept overnight at 4 • C. The following day, samples were centrifuged at 10,000× g for 30 min. The supernatant was eliminated and the pellet carefully washed with 300 µL of 70% ethanol, and centrifuged at 10,000× g for 15 min, followed by a gentle aspiration of the supernatant. Finally, the pellet was resuspended with 30 µL of pure water. DNA of samples was quantified using a NanoDrop ND-1000 (NanoDrop Technologies, Wilmington, DE, USA).

Amplification and Sequencing of Genomic Loci
Genes were amplified as follows: (1) pre-denaturation at 94 • C for 5 min; (2) denaturation at 94 • C for 45 s; (3) annealing at 55-57.3 • C for 1 min (55 • C = ITS, benA, cmdA, mcm7, preB and preA; 56 • C = ppgA; 57.3 • C = amdS and rpb1); (4) extension at 72 • C for 1 min (Steps 2 to 4 were carried out for 40 cycles); (5) final extension at 72 • C for 10 min; and (6) final temperature hold at 4 • C. Primers used in the study are shown in Table S1. Polymerase Chain Reaction (PCR) amplifications were performed in a C1000 Touch TM thermal cycler (BioRad, Marnes-la-Coquette, France). PCR amplicons were purified with GeneElute TM PCR Clean-Up Kit (Sigma-Aldrich, Saint Quentin Fallavier, France). Sanger sequences were obtained by using the Applied Biosystems Big Dye Terminator v3.1 chemistry (ThermoFisher Scientific, Illkirch, France), they were then purified with the Applied Biosystems Big Dye XTerminator protocol (Thermofisher Scientific, Illkirch, France) and finally processed on the ABI 3130xl Genetic Analyzer (Thermofisher Scientific, Illkirch, France), available on the GeT-Purpan technological facility (Genome and Transcriptome, GenoToul, Toulouse, France). New sequences were deposited in GenBank and accession numbers are reported in Table S2. Sequence data of some isolates, obtained from previously accessioned data in the GenBank database, were included for constructing phylogenetic trees (Table S2).

Alignment, Model Selection, and Molecular Analyses
Data were assembled, aligned and trimmed in BioEdit/ClustalW (http://www.mbio.ncsu. edu/bioedit/bioedit.html). Gene regions with multiple gaps were aligned to minimize indels and optimize nucleotide identities among different strains. Sequences from multiple genomic regions were concatenated using Mesquite v3.2 [72], but the mating type loci were analyzed independently. For concatenated data, the best-fit nucleotide substitution models and partitioning scheme were chosen using PartitionFinder v2.0.0 [73] under BIC. To search for the best-fit scheme, a greedy algorithm with linked branch lengths of alternative partitions was used. Partitions obtained consisted of four subsets that corresponded to a specific model (noted in parentheses): Subset 1 included ppgA, cmdA, benA, rpb1 and mcm7 = 2129 bp (K80+G); Subset 2 included ITS = 778 bp (TRNEF+I); Subset 3 included preB and preA = 1223 bp (HKY+G); and Subset 4 included amdS = 491 bp (K80+G).
Bayesian inference statistical methods were used to obtain tree topologies for concatenated data, using the best-fit substitution models listed above. For Bayesian analyses, MrBayes v3.2 [74] was used, and four independent runs were carried out for 10 7 generations, each with four chains, Markov Chain Monte Carlo, and sampling every 10 3 generations. We confirmed for each analysis that the average standard deviation of split frequencies between chains approached values of ≤0.01, and the potential scale factor reduction factor (PSRF) to 1. For all the analyses, and from the total number of trees per run, 25% were arbitrarily discarded as "burn-in". The remaining trees were used to calculate posterior probabilities (PP) for each bipartition in a 50% majority-rule consensus tree using Tracer v1.6 [75]. Phylogenetic trees were visualized and edited with FigTree v1.4.2 [76].
Cultures on MEA, MS, CZ, CYA, YES, CREA, and AFPA were seeded with three calibrated inoculates of 500 spores, and incubated in the dark at 25 • C for seven and ten days. Macroscopic characters were observed with a stereomicroscope SZX9-X12-120 (Olympus, Rungis, France). Microscopic characters were observed on MEA at 7 and 10 days using a microscope CX41-X400 and X1000 (Olympus, Rungis, France). In addition, growth analyses were calculated from MEA cultures, which were centrally inoculated with 10 3 spores, and incubated at 25 • C, 37 • C and 42 • C for seven days [17,79,80]. . For each medium, three biological replicates were inoculated centrally with 10 µL of calibrated spore suspensions (10 5 spores/mL) prepared from seven-day cultures on 7.5 cm Petri dishes. The samples were incubated in the dark at 27 • C for seven days.
To perform extrolite extractions, culture media were macerated and placed in 50 mL sterilized tubes, and thereafter 35 mL of ethyl acetate was added to each sample. Samples were agitated 48 h in an orbital incubator at 170 rpm at room temperature. Ethyl acetate was filtered through a Whatman 1PS phase separator (GE Healthcare Life Sciences, Vélizy-Villacoublay, France) and evaporated at 60 • C until dry. Samples were then dissolved in 400 µL of methanol. To eliminate possible impurities, each sample was filtered through a 0.45 µm disk filter (ThermoFisher Scientific, Illkirch, France) [81].

Secondary Metabolic Characterization of Sclerotia
The isolates were cultured on MEA whereby a loopful of spores was taken from seven-day cultures and streaked onto 9 cm Petri dishes. The MEA samples were incubated in the dark at 27 • C for eight days. To recover sclerotia from culture media, 10 mL of 0.01% Triton-X solution were added to each Petri dish. Sclerotia were gently scraped and transferred into 15 mL tubes. To remove mycelium and conidium debris, 10 mL of 0.01% Triton-X were added to each tube and homogenized in a vortex. Once the sclerotia precipitated, the supernatant was discarded. This step was carried out 4 to 5 times to eliminate possible residual debris [82].
Sclerotia were transferred into a 5 mL mortar. Then 5 mL of ethyl acetate were gently added while grinding the sclerotia. This step was carried out three times. Next, 5 mL of chloroform were gently added and the same procedure followed. This step was repeated three times. Samples were evaporated at 60 • C until dry, and resuspended in 0.5 mL solution methanol/acetonitrile/H 2 O qsp (30:30:40, v/v/v) [31,82]. To remove possible impurities, samples were filtered through 0.45 µm disk filters (ThermoFisher Scientific, Illkirch, France).
Full MS spectra were accomplished at a resolution of 60,000 with a mass-to-charge ratio (m/z) range 50-800. The MS/MS spectra were generated by collision-induced dissociation (CID) according the following parameters: collision energy = 35 eV, resolution = 7500, isolation width = 1.5 Da, activation Q = 0.250, and activation time = 30 ms.