Phylogenetic diversity, trichothecene potential, and pathogenicity within Fusarium sambucinum species complex

The Fusarium sambucinum species complex (FSAMSC) is one of the most taxonomically challenging groups of fusaria, comprising prominent mycotoxigenic plant pathogens and other species with various lifestyles. Among toxins produced by members of the FSAMSC, trichothecenes pose the most significant threat to public health. Herein a global collection of 171 strains, originating from diverse hosts or substrates, were selected to represent FSAMSC diversity. This strain collection was used to assess their species diversity, evaluate their potential to produce trichothecenes, and cause disease on wheat. Maximum likelihood and Bayesian analyses of a combined 3-gene dataset used to infer evolutionary relationships revealed that the 171 strains originally received as 48 species represent 74 genealogically exclusive phylogenetically distinct species distributed among six strongly supported clades: Brachygibbosum, Graminearum, Longipes, Novel, Sambucinum, and Sporotrichioides. Most of the strains produced trichothecenes in vitro but varied in type, indicating that the six clades correspond to type A, type B, or both types of trichothecene-producing lineages. Furthermore, five strains representing two putative novel species within the Sambucinum Clade produced two newly discovered type A trichothecenes, 15-keto NX-2 and 15-keto NX-3. Strains of the two putatively novel species together with members of the Graminearum Clade were aggressive toward wheat when tested for pathogenicity on heads of the susceptible cultivar Apogee. In planta, the Graminearum Clade strains produced nivalenol or deoxynivalenol and the aggressive Sambucinum Clade strains synthesized NX-3 and 15-keto NX-3. Other strains within the Brachygibbosum, Longipes, Novel, Sambucinum, and Sporotrichioides Clades were nonpathogenic or could infect the inoculated floret without spreading within the head. Moreover, most of these strains did not produce any toxin in the inoculated spikelets. These data highlight aggressiveness toward wheat appears to be influenced by the type of toxin produced and that it is not limited to members of the Graminearum Clade.

Introduction known about the potential of many species in this species complex to synthesize trichothecenes and cause FHB and FCR. Given the threat that these species pose to agricultural biosecurity, the present study was initiated to (i) collect and analyze multilocus DNA sequence data to reassess the phylogenetic diversity of 171 strains within the FSAMSC obtained from the Fusarium Research Center (FRC) at Pennsylvania State University and/or the Agricultural Research Service Culture Collection (NRRL), Peoria, Illinois, U.S.; (ii) test strains for their potential to produce trichothecenes in liquid media and on solid substrates; and (iii) evaluate their ability to induce FHB and synthesize trichothecenes in wheat.

Strains studied
A global collection of FSAMSC strains (n = 171) selected to represent the phylogenetic diversity of this species complex was included in the present study (Table 1, S1 Table). Strains were selected based on analyses of partial translation elongation factor sequence data. Only one strain of the 23 species in the Graminearum Clade was included in the present study because they have been the focus of several recent studies by our research group [6, reviewed in 38]. The strains were recovered from various hosts/substrates and are available from FRC (http:// plantpath.psu.edu/facilities/fusarium-research-center) and/or the ARS Culture Collection (https://nrrl.ncaur.usda.gov/) upon request.

Molecular phylogenetic analysis
To obtain mycelium for DNA extraction, strains were grown in 50 mL disposable polypropylene tubes containing 25 mL yeast-malt broth (3 g yeast extract, 3 g malt extract, 5 g peptone, and 20 g dextrose per L; Difco, Detroit, Michigan) on a rotary shaker set at 200 rpm and 25˚C. After 3-4 days mycelia were harvested on #1 Whatman filter paper discs over a Büchner funnel, lyophilized, and ground to a fine powder. Genomic DNA was then extracted following a cetyl trimethyl-ammonium bromide method [62]. For polymerase chain reaction (PCR) amplification, total genomic DNA was diluted 1:50 in sterile deionized water in 96-well plates and stored at -20˚C when not in use. To assess phylogenetic diversity of the 171 strains, portions of DNA-directed RNA polymerase II largest subunit (RPB1), RNA polymerase secondlargest subunit (RPB2), and translation elongation factor (TEF1) were amplified and then sequenced as previously described [63]. PCR amplifications were performed using Platinum Taq DNA polymerase (Invitrogen Life Technologies, Carlsbad, California) in an Applied Biosystems ProFlex thermocycler (ABI, Emeryville, California). Amplicons were purified using ExoSAP-IT (ABI) and Sanger sequenced using ABI BigDye v3.1 Terminator reaction mix in a ProFlex thermocycler. Sequencing reactions were then purified using ABI BigDye XTerminator and sequences were obtained using an ABI 3730 48-capillary DNA analyzer. Sequence chromatograms were edited using Sequencher 5.2.4 (Gene Codes Corp., Ann Arbor, Michigan). Sequences were then aligned with MUSCLE [64] in SeaView 4.7 [65] and individual trees were inferred from the RPB1, RPB2, and TEF1 data sets using maximum likelihood (ML) as implemented in IQ-TREE 1.6.12 [66]. Sequences of F. nelsonii NRRL 13338 from the F. chlamydosporum species complex (FCSC) were selected as the outgroup based on more inclusive analyses [67]. Using the Bayesian information criterion scores, ModelFinder [68] identified TIMe+I+G4, TNe+G4, and TIM2e+G4, respectively, as the best-fit models of molecular evolution for the RPB1, RPB2, and TEF1 partitions [69]. Clade support assessed by 5000 bootstrap (BS) pseudoreplicates of the data did not reveal any conflict between strongly supported nodes. Therefore, the RPB1, RPB2, and TEF1 partitions were analyzed as a combined data set with IQ-TREE. Clade support was also assessed by Bayesian posterior probabilities (BPP). Bayesian inference (BI) analyses of the individual partitions and combined 3-locus data set were performed using MrBayes 3.2.7a [70,71] on the CIPRES Scientific Gateway (https://www. phylo.org). Generalized Time Reversible (GTR) with a gamma distribution was selected as the model for nucleotide substitution. Two runs of four MCMC (Markov Chain Monte Carlo) chains were run simultaneously for 10 7 generations with trees sampled every 1000 generations. For each run, the first 25% trees were discarded as the burn-in phase and BPP were determined from the remaining trees. Lineages that were represented by two or more strains were considered to be phylogenetically distinct if they were supported as genealogically exclusive by a BPP � 0.99 and ML bootstrap (ML-BS) � 95% from one or more of the individual partitions and/or the combined data set, and none of the individual partitions contradicted their monophyly. The RPB1, RPB2, and TEF1 sequences were deposited in GenBank NCBI (accession numbers MW233055-MW233569).

Trichothecene production in broth, rice, and corn substrates
Two liquid culture media were used to test the ability of 171 strains to produce trichothecenes. Strains were initially grown on V8 agar [72] for 7 days at 25˚C. Two 0.5 cm 2 mycelial plugs cut from V8 cultures were inoculated into 20 mL of agmatine broth [73] and 20 mL of yeast extract peptone dextrose medium (YEPD: 1 g yeast extract, 1 g peptone, and 50 g dextrose per 1 L distilled water) in 50 mL flasks. Flasks were incubated at 28˚C in the dark on a rotary shaker at 200 rpm. After 7 days the cultures were extracted with 8 mL of ethyl acetate. The extracts were dried under a nitrogen stream in 1-dram vials and then resuspended in 1 mL ethyl acetate. Strains that failed to synthesize trichothecenes in agmatine and YEPD broth, and some isolates of special interest, were cultured on autoclaved rice (4.4 g rice grain + 1.8 mL water). After 15 days incubation in the dark at 25˚C, each rice culture was extracted with 10 mL of ethyl acetate, dried under nitrogen, and then resuspended in 1 mL ethyl acetate. Strains that did not produce detectable levels of trichothecenes in rice cultures were grown on autoclaved cornmeal (4.4 g commercial Quaker 1 cornmeal + 1.8 mL water) for 15 days in the dark at 25˚C. Toxins were extracted following the protocol described above. Gas chromatographymass spectrometry (GC-MS) analyses of the ethyl acetate extracts were performed on an Agilent 6890 gas chromatograph fitted with a HP-5MS column (30 m, 0.25 mm, 0.25 μm) coupled to a 5973-mass spectrometer (Agilent, Santa Clara, California). Helium was used as the carrier gas with a 20:1 split ratio and a 20 mL/min split flow. The column was held at 150˚C for 1 min following injection, heated to 280˚C at 30˚C/min where it was held for 7.7 min. Compound identification was based on comparisons of their retention times and mass spectra with a library of mycotoxin standards and a NIST reference library.

Pathogenicity test on wheat
Pathogenicity of 75 strains chosen to represent the full range of FSAMSC diversity was tested on the full-dwarf hard red spring wheat cultivar Apogee (Triticum aestivum) [74]. This cultivar is characterized by high susceptibility to FHB [75]. Five seeds of Apogee were planted in 2.5 L plastic pots filled with Sun Gro1 Horticulture potting mix (Agawam, Massachusetts) supplemented with 2 g of Micromax, and 13 g Osmocote 14-14-14 controlled-release fertilizer (ICL Specialty Fertilizers, Summerville, South Carolina). Pots were arranged randomly and maintained in a controlled growth room under the following conditions: 50-60% relative humidity, 14 h day period, and temperature set at 23˚C day and 20˚C night. Every two weeks each pot was fertilized with 500 mL of a solution containing 0.5 g per L of Peters 20-20-20 (Grace-Sierra Horticultural Products, Milpitas, California). At anthesis, five wheat heads representing biological replicates were inoculated with one of the 75 strains by injecting a spikelet with 10 μL of 0.04% sterile Tween 20 containing 10 5 conidia. The negative control was inoculated with 10 μL of 0.04% sterile Tween 20. Fusarium graminearum NRRL 31084 (= PH1), which is a well-characterized FHB pathogen [11], was used as the positive control. The number of symptomatic florets displaying premature whitening or necrosis, typical symptoms of FHB, was scored at 7, 10, 14, 17, and 21 days after inoculation. Isolates that only induced FHB symptoms in the inoculated spikelet were considered as pathogenic. Strains that overcame the plant's defenses and spread to other florets were scored as aggressive. Disease spread was then assessed by calculating the total percentage of diseased florets 21 days post inoculation.

Trichothecene production in planta
To evaluate trichothecene production in planta, independent of disease spread, the five wheat heads inoculated with one of the 75 strains were harvested 21 days post inoculation, combined, and frozen at -80˚C. These samples were freeze-dried for 48 hr and then ground to a fine powder using a Geno/Grinder 2010 automated tissue homogenizer (OPS Diagnostics, Lebanon, New Jersey). One g of each sample was extracted with 10 mL of acetonitrile-water (86:14, v/v) by shaking for 15 min. After centrifugation 5 mL of the extract were cleaned using a MycoSep 225 Trich column (Romer Labs, Union, Missouri). Two mL of the purified extract were transferred to a 1-dram vial, dried under a nitrogen stream, and then derivatized with 100 μL of a 100:1 freshly prepared mixture of N-trimethylsilylimadazole/trimethylchlorosilane (Sigma-Aldrich, St. Louis, Missouri). After 20 min, 900 μL of isooctane were added to the vial, followed by 1 mL of water. The samples were vortexed until the top organic layer became clear and then it was analyzed by GC-MS as described above.

Molecular phylogenetic analysis
Maximum likelihood bootstrap (ML-BS) analyses of the individual RPB1, RPB2, and TEF1 partitions were conducted to assess whether they could be analyzed as a combined 3-gene data set. Because trees inferred from the individual genes did not reveal any conflict between strongly supported nodes (S1-S3 Figs), the data sets were combined and analyzed by Bayesian inference (BI) and ML bootstrapping (ML-BS) to assess species boundaries and to gain insight into evolutionary relationships of the 171 strains. The concatenated 3-gene data set included partial RPB1 (1774 bp alignment, 492 parsimony informative characters = PICs), RPB2 (1786 bp alignment, 489 PICs), and TEF1 (640 bp alignment, 276 PICs) gene sequences. Sequences of NRRL 13338 F. nelsonii from the FCSC, which is sister to the FSAMSC, were selected as the outgroup for rooting the phylogeny. BI and ML analyses of the 3-locus data set yielded similar trees and resolved the 171 strains as 74 phylogenetically distinct species (Fig 1).  Fig 1). Analyses of the combined data set revealed that the 39 lineages represented by two or more strains were strongly supported as phylogenetically distinct (BPP = 1; ML-BS = 96-100%, Fig 1). These analyses also resolved the FSAMSC as six strongly supported monophyletic clades (BPP = 1; ML-BS = 100%, Table 1, Fig 1) and evolutionary relationships among them (BPP = 1; ML-BS = 89-100%, Fig 1). The analyses provided moderate to strong support for most nodes (BPP = 0.84-1; ML-BS = 80-100%, Fig 1); however, eight internodes in the interior of the ML phylogeny received bootstrap values < 70% (see Brachygibbosum, Graminearum, and Sambucinum Clades). Four of these internodes, however, were strongly supported by the Bayesian posterior probabilities (BPP = 0.99-1, Fig 1). In addition, seven of the nine internodes that received a BPP < 0.7 (see Brachygibbosum, Graminearum, Longipes, Sambucinum, and Sporotrichioides Clades) were supported by ML-BS values � 70-95%. The six multispecies lineages identified within the FSAMSC are (i) the Sporotrichioides Clade, which included 32 isolates accounting for 7 described and four novel phylogenetically distinct species closely related to F. armeniacum (F. sp. nov.-1 to -4; Fig 1). Each unnamed putatively novel species was represented by two strains from different geographic origins and isolated from cereals or other hosts/substrates (Table 1). Evolutionary relationships among the four putatively novel species and F. armeniacum were fully resolved (BPP = 0.99-1; ML-BS = 96-100%, Fig 1); (ii) the Graminearum Clade contained one unnamed species (NRRL 34461 F. sp. nov.-5) and 22 formally described species, each represented by a single strain (Fig 1). NRRL 34461 F. sp. nov.-5, recovered from soil in South Africa, formed a strongly supported cluster with F. acaciae-mearnsii, F. vorosii, and F. ussurianum within what was previously named the F. graminearum species complex (BPP = 1; ML-BS = 100%, Fig 1). To facilitate communication, given the recent advances in Fusarium systematics, we propose abandoning the informal designation F. graminearum species complex and to refer to this group and closely related species as the Graminearum Clade, to emphasize its status as a monophyletic multispecies lineage within the FSAMSC. The Graminearum and Sporotrichioides Clades were strongly supported as sisters (BPP = 1; ML-BS = 99%, Fig 1); (iii) a Novel Clade comprised two phylogenetically distinct species, F. sp. nov.-6 and -7 (BPP = 1; ML-BS = 100%, Fig 1). Fusarium sp. nov.-6 is represented by three strains: NRRL 54640 and 54683 from Papua New Guinean sweet potato received as F. lateritium and NRRL 28066 isolated from corn in Japan accessioned as F. graminearum (S1 Table). Fusarium sp. nov.-7 comprised two Ethiopian isolates, NRRL 46743 from wheat seed and NRRL 66736 from soybean plants exhibiting root rot symptoms (S1 Table). BPP and ML bootstrapping resolved the Novel Clade as sister to the Sporotrichioides + Graminearum Clades (BPP = 1; ML-BS = 89%, Fig 1); (iv) the Sambucinum Clade included 42 isolates comprising 6 described and 10 putatively novel phylogenetically distinct species (F. sp. nov.-8 to -17; Fig 1). The latter were represented by 2 or 3 strains except for F. sp. nov.-8, -11, -14, and -17, which are single species lineages (Fig 1). Monophyly of the 6 putatively novel and 4 described species represented by 2 or more strains was strongly supported (BPP = 1; ML-BS = 100%, Fig 1). BPP and ML bootstrapping resolved the Sambucinum Clade as a basal sister to the Sporotrichioides, Graminearum, and Novel Clades (BPP = 1; ML-BS = 100%, Fig 1). BI and ML and analyses of the 3-locus data set indicated that the three strains received as F. tumidum [76] from gorse, lupine, and Scotch broom in New Zealand are F. sambucinum (S1 Table, Fig 1); (v) the Longipes Clade contained 19 isolates that were resolved as 9 phylospecies (BPP = 1; ML-BS = 100% , Fig 1), 5 of which were represented by single strains. Seventeen of the 19 isolates were received as F. longipes and these included  [71]. ML-BS values were determined using IQ-TREE 1.6.12 [66]. The FSAMSC was rooted on sequences of NRRL 13338 F. nelsonii from its sister group, the F. chlamydosporum species complex. The following six clades (from top to bottom defined by thickened internodes) were strongly supported as monophyletic ( representatives of all 9 phylospecies (S1 Table). Following a previously used informal ad hoc nomenclature, four of the species were distinguished as F. longipes-1 to -4, and the other 5 as F. sp. nov.-18 to -22 (Fig 1). Seven of the 9 species and 15/19 isolates were recovered from soil ( Table 1). The Longipes Clade was resolved as a basal sister to the Sporotrichioides, Graminearum, Novel, and Sambucinum Clades (BPP = 0.91; ML-BS = 94%, Fig 1); and (vi) the Brachygibbosum Clade, the earliest diverging clade within the FSAMSC, accounted for 50 of the 171 strains and these included 2 described species (i.e., F. brachygibbosum and F. transvaalense) and 11 putatively novel phylogenetically distinct species distinguished as F. sp. nov.-23 to -33 (Fig 1). Representatives of 10/13 species and 28/50 of the Brachygibbosum Clade isolates were received as F. compactum (S1 Table). Of the 13 species in this clade, NRRL 66919 F. sp. nov.-31 from Ghanaian soybean exhibiting root rot was the only species represented by a single strain (Fig 1). Monophyly of the 12 other species was strongly supported (BPP = 1; ML-BS = 95-100%, Fig 1). Evolutionary relationships within this clade were mostly wellresolved by BPP and ML-BS. However, two pairs of putative sisters were only resolved by the Bayesian analysis, and these are

Trichothecene production in broth, rice, and corn substrates
Structurally diverse toxins were produced by the FSAMSC strains in the current study. Of the 171 strains tested 38 failed to synthesize any toxin in the liquid media and solid substates we used (S2 Table). The type A trichothecenes DAS, NEO, and T-2 and type B trichothecenes NIV, 3ADON, and 15ADON were among the most prevalent toxins detected. The six clades resolved by BI and ML bootstrap analyses of the 3-locus data set corresponded to six trichothecene-producing lineages represented by strains that synthesized type A trichothecenes, type B, or both types (S2 Table, Fig 1).  Table, Fig 1).
Sambucinum Clade. Fourteen of the 16 species within Sambucinum Clade only produced type A trichothecenes; however, NRRL 6491 and 3509 F. kyushuense from Japan synthesized NIV and DAS, which are type B and type A trichothecenes respectively. NRRL 22192 F. sp. nov.-11 (= BBA 64918), represented by a single strain from an Indonesian palm tree, was the only species within this clade that failed to produce toxins (Fig 1). In addition to F. kyushuense, DAS was detected in 9/16 species, T-2 in 7, NEO in 5, CUL in 3, butenolide (BUT) in 3, CAL in 2, and 3OH in 2 (Table 1). Of special note, two putative novel sister species, F. sp. nov.-15 and -16, which were received as F. longipes isolated from South African wheat straw or soil, produced CUL, 7-hydroxy isotrichodermol (3,7-diOH), 15-decalonectrin (15-decal), and two novel type A trichothecenes, 15-keto NX-2 and 15-keto NX-3 in agmatine broth (Fig 2A). The latter two toxins are similar to NX-2 and its deacetylated form NX-3 but differ in that they possess a keto or aldehyde group, respectively, at the C-15 position on the trichothecene ring. When cultured on rice, F. sp. nov.-15 and -16 produced the same two novel toxins with 15-hydroxy culmorin (15-OHCUL) (Fig 2B).
Longipes Clade. Of the 19 isolates comprising 9 phylospecies in the Longipes Clade tested, DAS was only produced on rice by one isolate of four different species. The other 15 strains failed to synthesize any toxin in liquid media and on solid substrates (S2 Table, Fig 1).
Brachygibbosum Clade. The Brachygibbosum Clade was represented by 50 isolates comprising 13 phylogenetically distinct species; however, only F. brachygibbosum and F. transvaalense possess binomials. Eight of the species synthesized one or more type A trichothecenes (i.e., DAS, NEO, and CAL). Of the 5 remaining species, four produced DAS + NIV, and one only NIV.
In addition to type A and B trichothecenes, several other biologically active secondary metabolites were detected in culture extracts of some isolates in four clades (Table 1, Fig 1). The sesquiterpene CUL was detected in cultures of most strains within the Graminearum Clade (14/23), 2 strains in the Sporotrichioides Clade, and 6 strains representing 3 novel type A trichothecene-producing species in the Sambucinum Clade. In addition, BUT was produced in rice and cornmeal cultures by some isolates of F. venenatum, NRRL 39685 F. sp. nov.-13, and NRRL 26795 F. sp. nov.-17 in the Sambucinum Clade, and FRC R-4478 F. transvaalense in the Brachygibbosum Clade (S2 Table, Fig 1).

Pathogenicity test on wheat
The pathogenic potential of 75 strains representing the six clades of the FSAMSC (Brachygibbosum = 17, Graminearum = 12, Longipes = 10, Novel = 4, Sambucinum = 20, and Sporotrichioides = 12) was determined by their ability to cause FHB on the highly susceptible wheat cv. Apogee and to spread within the wheat head from a single floret inoculation. Disease spread was evaluated by assessing the percentage of florets exhibiting premature whitening or necrosis at different timepoints, and when the experiment was terminated 21-days post inoculation. Of the 75 strains evaluated, only 17 were nonpathogenic, and heads inoculated with these were asymptomatic and indistinguishable from the negative controls. These nonpathogenic strains were members of the Brachygibbosum, Longipes, Novel, Sambucinum, and Sporotrichioides Clades. The remaining 58 strains induced FHB symptoms, but the majority (41/58) did not spread beyond the inoculated floret and were classified as nonaggressive (S3 Table). Phenotypic variation in symptoms was noted on wheat heads inoculated with the nonaggressive strains. Some strains produced necrotic lesions or black spots on the inoculated spikelet, while others caused premature bleaching, a typical symptom of FHB. However, bleaching did not spread to neighboring florets. The remaining strains, which belong to the Graminearum (n = 12) and Sambucinum (n = 5) Clades were able to spread and colonize nearby florets or the entire wheat head. Some of these strains exhibited levels of aggressiveness like the highly aggressive positive control NRRL 31084 F. graminearum (S3 Table, Fig 3). Strains of F. sp. nov.-15 and -16, the two novel species within the Sambucinum Clade that produced 15-keto NX-2 and 15-keto NX-3 along with other toxins in culture, induced bleaching on Apogee and spread, causing considerable disease above and below the inoculation point (S4 Table, Fig 3). Variation in aggressiveness was noted among strains tested within the Graminearum Clade (S4 Table, Fig 3). NIV-producing species within this Clade (F. austroamericanum, F. cerealis, F. cortaderiae, F. meridionale, and F. praegraminearum) displayed moderate (43-64%, Fig 3)  to high (95-100%, Fig 3) aggressiveness toward wheat cv. Apogee (Fig 3). Strains of F. cortaderiae, F. meridionale, and F. praegraminearum infected approximately half of the wheat head by the end of the assay, which is modest compared to the more aggressive species F. austroamericanum and F. cerealis that were able to spread through the whole head 21 days post inoculation (S4 Table, Fig 3). The 3ADON/15ADON-producing species (F. graminearum, F. pseudograminearum, F. vorosii, F. boothii, and NRRL 34461 F. sp. nov.-5) as well as F. asiaticum and F. culmorum, which did not produce toxins in vitro but made NIV and DON, respectively, in planta (see below), were able to spread and induce severe head blight symptoms on wheat cv. Apogee where they colonized the entire wheat head 17 to 21 days following single floret inoculation (S4 Table, Fig 3).

Trichothecene production in planta
Analyses of trichothecene production in the inoculated wheat heads showed that two of the 17 nonpathogenic strains (FRC R-738 F. sambucinum and FRC R-7107 F. sp. nov.  Table). The other nonaggressive strains in the Brachygibbosum, Longipes, Novel, Sambucinum, and Sporotrichioides Clades failed to produce any toxin in the inoculated heads (S3 Table). By contrast, the five strains representing F. sp. nov.-15 and -16 from the Sambucinum Clade, which synthesized 15-keto NX-2 and 15-keto NX-3 along with other toxins in vitro, made NX-3, 15-keto NX-3, and hydroxy culmorin (OHCUL) in planta (S3 Table). Analyses of wheat heads inoculated with members of the Graminearum Clade revealed that they were all toxigenic in planta (S3 Table). Strains of F. austroamericanum, F. cerealis, F. cortaderiae, F. meridionale, F. praegraminearum, and F. asiaticum, which failed to produce any toxin when tested in vitro, made Percentage of symptomatic spikelets (± SEM) was calculated from the average of percent diseased florets in the five inoculated wheat heads 21 days post inoculation. Letters above bars indicate significant differences as determined by ANOVA analysis followed by Fisher's protected least significant difference (LSD) test for individual comparisons (P < 0.05). Strains represented by four-and five-digit numbers were obtained, respectively, from the FRC and NRRL culture collections (strain identity and history can be found in Table 1 NIV in planta. Some of these species also produced CUL and OHCUL (S3 Table). Similarly, CUL and OHCUL were synthesized by F. sp. nov.-5, F. culmorum, and F. pseudograminearum in the inoculated heads along with deoxynivalenol (DON). The acetyl-ester derivative of DON, 3ADON was also detected in heads inoculated with F. pseudograminearum. DON was also produced in planta by F. boothii, F. graminearum, and F. vorosii.

Phylogenetic diversity and trichothecene potential of FSAMSC strains
The present study was initiated to investigate species diversity, trichothecene toxin potential, and pathogenicity on wheat within the FSAMSC. Toward this end, 171 FSAMSC strains were selected from the FRC and NRRL culture collections, based on analyses of partial TEF1 sequence data, to represent the phylogenetic diversity of this species complex. GCPSR-based analysis of combined partial RPB1, RPB2, and TEF1 sequence data, together with prior extensive studies on the Graminearum Clade [5,6, reviewed in 38], resolved the 171 strains as 74 genealogically exclusive phylospecies distributed among six strongly supported trichothecene toxin-producing clades. This finding is consistent with published genomic data that all FSAMSC species examined to date possess intact trichothecene biosynthetic genes [22,77,78]. However, previous phylogenetic studies have found that the evolutionary history of several TRI cluster genes does not mirror the species phylogeny [22,26,78]. As previously reported, results of the current study showed clades within the FSAMSC generally correlate with the type of trichothecene mycotoxin produced [29,49,79,80].
Sporotrichioides Clade. Consistent with what is known about trichothecene production within this clade (Table 1, [29,48,57,58,60]), the 11 species analyzed, including 4 novel undescribed species closely related to F. armeniacum, produced DAS, NEO + T-2, or DAS, NEO, and T-2 type A trichothecenes. Toxin production by the 4 newly discovered phylospecies (F. sp. nov.-1 to -4) and F. nodosum is reported here for the first time. The two strains of F. nodosum studied herein, NRRL 13431 (= FRC T-384) from a tomato stem in Turkey and NRRL 36351 (= CBS 201.63) from stored peanuts in Portugal, were received as F. chlamydosporum. Fusarium nodosum has also been isolated from Arundo donax in France and wheat in Iran [8].
Graminearum Clade. The Graminearum Clade isolates investigated here have been the subject of several detailed phylogenetic studies [6,9,10,38,81]. Nineteen of the 23 species in this clade were discovered in prior GCPSR studies over the past two decades and all have been formally described except for NRRL 34461 F. sp. nov.-5. The latter species was initially reported as F. acaciae-mearnsii [35], but subsequently as a putatively novel sister to F. acaciae-mearnsii [43]. In contrast to F. sp. nov.-5 from soil in South Africa, the few characterized strains of F. acaciae-mearnsii were recovered from Acacia mearnsii and wheat in South Africa and soil in Australia [10,35,82]. Prior to the recent discovery of type A trichothecene NX-2 toxin-producing F. graminearum strains in North America [9,27,28], members of this clade were only known to produce type B trichothecenes. We found that 19/23 strains, each representing a unique phylospecies in this clade, synthesized type B trichothecenes in vitro and/or in planta. However, F. brasilicum made NIV and the type A trichothecene CAL in agmatine broth. Although F. mesoamericanum, F. louisianense, and F. dactylidis failed to synthesize type B trichothecenes here, they are all known to produce them [10,41,43]. Even though NRRL 13393 F. lunulosporum possesses an intact trichothecene toxin gene cluster [77], it failed to synthesize any trichothecene. However, it did produce zearalenone as previously reported [10]. Global surveys have revealed that F. asiaticum, F. culmorum, and F. graminearum are segregating for 15ADON, 3ADON, and NIV worldwide [26,32,82] whereas the available data indicate the 15ADON chemotype is fixed in F. boothii [26,82,83]. Fusarium ussurianum, by contrast, appears to be fixed for the 3ADON chemotype in the Far East of Russia [44]. Some reports suggest that 3ADON toxin-producing FHB pathogens are more aggressive than those that produce 15ADON on wheat [84]. In North America, an introduced 3ADON population of F. graminearum has increased in frequency compared to the native 15ADON population [81,84]. These observations raise concerns about the sustainability of FHB-resistant fungicides and emphasize the need for continued monitoring of their population dynamics.
Novel Clade. The current study also resulted in the discovery of a novel type A T-2, DAS, and NEO-producing trichothecene lineage comprising two novel phylogenetically distinct sister species, F. sp. nov.-6 and -7. The former species, F. sp. nov.-6, has an Asian-Pacific distribution with isolates from sweet potato in Papua New Guinea and corn in Japan ( Table 1). The latter species (i.e., F. sp. nov.-7) was recovered from soybean roots exhibiting root rot, a symptom of sudden death syndrome [34], and wheat seed in Ethiopia (Table 1).
Sambucinum Clade. The ten novel species discovered in this clade were received incorrectly identified as 8 different species (S1 Table). Of the 16 Sambucinum Clade species we tested, F. kyushuense was the only one capable of producing type B NIV and type A DAS trichothecenes, as previously reported [reviewed in 85,86]. Nivalenol and fusarenon-X were first characterized in this species [46,47]. Consistent with previous reports, F. musarum, F. venenatum, and F. sambucinum, including strains of the latter received as F. tumidum [29,49,87], produced the type A trichothecenes T-2, DAS, and NEO. The same toxins were also synthesized by all but one of the 10 novel species. In addition, two strains of F. venenatum and NRRL 26795 F. sp. nov.-17 (= BBA 70355 received as F. torulosum from U.S. soil) synthesized CAL. Production of the type A trichothecene 3OH was also detected in cultures of NRRL 22198 F. venenatum and FRC R-7122 F. sp. nov.-12. We found that 2/3 strains of F. poae synthesized the type A toxin DAS; however, this species has been reported to produce type A and type B trichothecenes, including DAS, NEO, T-2 toxin, and NIV (Table 1, [48]). Proctor et al. [78] discovered that F. poae lacks the TRI16 gene, which is required for synthesis of T-2 toxin. Prior to its formal description, the T-2 and HT-2 producer F. langsethiae was mistakenly thought to be F. poae [59]. These findings may explain the incorrect reports of T-2 toxin production by F. poae. Fusarium venenatum, the species used to produce Quorn mycoprotein for human consumption, was found to synthesize the select agent DAS when grown in rice culture [29,55]. Consistent with this finding, the isolates of F. venenatum analyzed here all produced DAS, but some also made CAL, 3OH, and BUT in agmatine broth and on rice. The Quorn production strain NRRL 26139 (= IMI 145425 = ATCC 20334 = A 3/5; [88]), however, failed to produce DAS and any other toxin in agmatine and rice cultures [McCormick,unpubl,14,55]. Given that a comparative genomic analysis, which included the Quorn production strain ATCC 20334 revealed it possesses the genes required for type A trichothecene biosynthesis [89], the open question is: Why aren't they produced in agmatine medium? Agmatine is a strong inducer of TRI5, the gene that encodes the first enzymatic step in the trichothecene biosynthesis pathway [73]. When the Quorn production strain A3/5 (= ATCC 20334) was originally isolated in 1968 [88] it was identified as F. graminearum [90]; however, this species was not formally recognized as F. venenatum until 1995 [91]. The present study failed to recover additional strains of F. musarum [92], the earliest diverging lineage within the Sambucinum Clade. Similarly, F. robustum NRRL 13392 recovered from a rotten stem of Paraná pine (Araucaria angustifolia) in Argentina [93] is only represented by the ex-type strain.
Global surveys indicated that NX-2 toxin-producing strains of F. graminearum are restricted to Canada and the upper Midwest and Northeast U.S., suggesting that this chemotype might be indigenous to this region [27,28,94]. Thus, the discovery of two species within the Sambucinum Clade from South Africa that produced novel type A trichothecenes in vitro similar to NX-2 and NX-3, but with a keto or aldehyde moiety at the C-15 position, and NX-3, the deacetylated form of NX-2 in planta is one of the most important findings of the present study. This discovery is the first to document that NX-2 toxin-producing fusaria are present outside of North America. Although these species were recovered from soil and debris, future studies are needed to determine whether these novel toxins contaminate food and feed and pose a threat to food safety and human health.
Longipes Clade. Previous phylogenetic studies have showed that the morphospecies F. longipes encompasses at least four phylospecies [67,95]. Our analyses of 19 strains from the FRC and NRRL culture collections indicate this clade comprises at least 9 phylogenetically distinct species. The fact that only 4/19 strains representing four species produced DAS in rice, but not in agmatine, YEPD, and cornmeal media, suggests the experimental conditions used may have been suboptimal, or these soil inhabiting fusaria may have lost the ability to synthesize trichothecenes. To our knowledge beauvericin production by NRRL 13368 F. longipes-1 is the only toxin report from this clade [45]. However, a recent phylogenomic study revealed the presence of an intact TRI cluster in the genome of NRRL 20695 F. longipes-4; but trichothecene production by this strain was not investigated [77].
Several strains in each of the six clades failed to produce trichothecenes in vitro under the experimental conditions used. For example, strains of F. asiaticum, F. louisianense, and F. dactylidis tested failed to synthesize any trichothecene in the current study, but they have been shown to synthesize NIV [11,41,43]. Therefore, this failure to produce trichothecenes might be because some strains have lost their toxigenic potential because they have been in culture for decades or the growth conditions we employed were suboptimal for trichothecene production for some strains. The latter hypothesis is supported by studies that found trichothecene production on agmatine was influenced by strain diversity [29,73]. Further research is required to identify optimal nutrient profiling conditions to stimulate trichothecene production in vitro and/or in planta. This is supported by the fact that some strains that failed to produce toxins in agmatine broth, which is a strong inductor of trichothecene synthesis [73], made trichothecenes when grown on rice or cornmeal (S2 Table).
The sesquiterpene diol culmorin was produced by most of the species within the Graminearum Clade, three novel species within the Sambucinum Clade, and by F. langsethiae in the Sporotrichioides Clade. Biosynthesis of this toxin requires a terpene synthase (TS) and a cytochrome P450 monooxygenase encoded by CLM1 and CLM2, respectively [97,98]. A comparative phylogenomic analysis revealed that the two genes were conserved and were present in the genomes of TRI cluster-containing species closely related to F. graminearum within the FSAMSC [77]. By contrast, one or both genes were absent in the more distantly related species F. sambucinum, F. poae, F. sporotrichioides, F. longipes, and F. langsethiae [77]. Thus, the fact that CUL was detected in F. langsethiae herein and in previous studies [48,59] suggests the genomes should be screened for the presence of these two genes, given that intraspecific variation in the presence and absence of certain secondary metabolite gene clusters was previously noted [99].
Overall, the BI and ML analyses revealed 101 of the 171 strains included in the present study were received misidentified, in some instances under names that are not accepted today (i.e., F. trichothecioides and F. sulphureum = F. sambucinum). It is worth mentioning that most of the species within the Graminearum Clade (n = 19) were also received misidentified, mostly as F. graminearum, prior to GCPSR studies of this clade [10,38]. The present phylogenetic analysis also revealed that close to half of the genealogically exclusive lineages (33/74 = 44.6%) represented novel phylospecies. In addition to the 23 Graminearum Clade strains, each representing a unique phylospecies, the other 148 strains were resolved as 19 previously described and 32 novel phylospecies in the following five FSAMSC clades: Brachygibbosum (n = 50 strains, 11/13 novel species), Sambucinum (n = 42 strains, 10/16 novel species), Sporotrichioides (n = 32 strains, 4/11 novel species), Longipes (n = 19 strains, 5/9 novel species), and a Novel Clade (n = 5 strains, 2/2 novel species). The current study highlights the importance of multilocus DNA sequence typing (MLST) as an essential tool for accurately identifying fusaria. To illustrate their impact, studies over the past two decades indicate that Fusarium comprises several fold more species than included in the traditional morphology-based taxonomic treatments [100]. For example, in a reassessment of fusaria included in Marasas et al. [51], a MLST analysis [29] found that the 17 morphospecies included in this compendium comprised 46 phylospecies, including 10 that were putatively novel.

Pathogenicity on wheat and trichothecene production in planta
We evaluated pathogenicity, aggressiveness, and trichothecene production of 75 strains representing 57 species spanning six clades of the FSAMSC on wheat (cv. Apogee). Our data showed that only members of the Graminearum Clade and two novel species within the Sambucinum Clade were aggressive. Strains of the other species were either not pathogenic to wheat or caused head blight symptoms in the inoculated floret without spreading to nearby spikelets. Furthermore, most of these strains failed to produce detectable levels of mycotoxins in the inoculated wheat heads. The failure of these strains to cause FHB and to spread may be because they did not produce trichothecenes, which are required for fungal spread and FHB development in some Triticeae [17,19]. The nonaggressive strains belong to five clades other than Graminearum and most were isolated from non-gramineous hosts or other substrates. This finding suggests that the FSAMSC may contain nonpathogenic species that possess an endophytic or saprophytic lifestyle. For example, while F. poae and F. sporotrichioides are widely recovered from cereals, they are less aggressive secondary invaders of heads infected by the major FHB pathogen F. graminearum [101]. The available data also suggest that F. armeniacum, F. brachygibbosum, F. longipes, and F. venenatum have evolved principally as common saprophytes of diverse crops, natural grasses, soil, and other substrates [102]. All strains within the Longipes Clade and more than half in the Brachygibbosum Clade were recovered from soil and debris. By contrast, Sporotrichioides and Sambucinum Clade strains were recovered from a wide range of hosts including cereals, wild grasses, soil, and debris.
A detailed comparative genomics study of F. venenatum and F. graminearum revealed a high degree of synteny between the two genomes, including genes known to be involved in virulence [89]. Fusarium venenatum may be nonpathogenic to wheat because it is missing two transcription factors previously shown to be required for F. graminearum virulence on wheat and five F. graminearum specific secondary metabolite gene clusters that are typically upregulated during infection of wheat by F. graminearum. The combined loss of these genes, and the toxigenic difference between DON-producing F. graminearum and DAS-producing F. venenatum, are thought to contribute to the lack of aggressiveness toward wheat in F. venenatum and the divergent lifestyles of these fungi [89]. These findings also highlight the need for further research on the evolution of pathogenesis in other species within the FSAMSC to better understand what factors contribute to FHB pathogen aggressiveness toward cereals.
In that regard, variation in aggressiveness was observed among the Graminearum Clade strains tested in the current study (Fig 3). The NIV-producing species F. austroamericanum, F. asiaticum, and F. cerealis were highly aggressive on wheat (cv. Apogee) and spread throughout most of the head at 21-days post inoculation. However, other NIV-producing species were less aggressive (F. cortaderiae, F. meridionale, and F. praegraminearum). Our data are consistent with other reports that evaluated these species on other wheat cultivars including cv. Norm [6,11,41,83]. Even so, toxin analyses of the inoculated wheat heads revealed that these species were all toxigenic and made NIV in planta. Graminearum Clade strains that produced 3ADON or 15ADON in vitro displayed high aggressiveness on wheat and were able to spread throughout the entire wheat heads. Tracking disease progress at different timepoints revealed that these species rapidly spread and colonized the entire heads 17 days post inoculation. Consistent with their ability to spread throughout the wheat head, GC-MS analyses showed they produced DON in planta. Overall, these findings match previous reports that indicate DON functions as a virulence factor allowing the fungus to overcome wheat defenses and spread throughout wheat heads [17,19].
In addition to producing type B trichothecenes, most of the Graminearum Clade strains tested produced CUL and/or its derivative OHCUL along with DON or NIV in planta. Culmorin is considered an "emerging mycotoxin" that is frequently found in naturally contaminated grain and axenic cultures of the most important FHB causal agents where it co-occurs with DON and other trichothecene analogs [103]. However, information on the toxicological relevance of CUL is limited. Nevertheless, a recent study demonstrated that CUL and DON have a synergistic phytotoxic effect on cereals when the concentration of CUL exceeded that of DON [77]. In addition, these authors reported that the severity of FHB induced by F. graminearum on wheat was positively correlated with the sum of CUL and DON. Given that CUL inhibits the activity of UDP-glucosyltransferases [104], their working hypothesis is CUL suppresses glycosylation of DON by serving as an alternate substrate for UDP-glucosyltransferases [77]. While 15 of the 20 Sambucinum Clade isolates tested for pathogenicity on wheat were not aggressive, strains of the novel species F. sp. nov.-15 and -16 induced FHB and showed levels of aggressiveness on cv. Apogee comparable to members of the Graminearum Clade (Fig 3). Mycotoxin analyses revealed that these newly discovered sister species produced the novel toxins 15-keto NX-2 and 15-keto NX-3 with other toxins in vitro and 15-keto NX-3, NX-3, and the sesquiterpene OHCUL in planta. Future research is required to better understand the pathophysiological impact of these novel toxins and to determine whether they function as virulence factors against wheat and other small grain cereals. Results of the pathogenicity assay provide evidence that the potential to induce FHB and aggressiveness are clearly influenced by the type of toxin produced and this appears to be a strain rather than a clade-specific characteristic. In planta toxin data showed that type B trichothecene NIV-and DON-producing strains from the Graminearum Clade and the type A trichothecene NX-3 and 15-keto NX-3 producers from the Sambucinum Clade were aggressive toward wheat. By contrast, NIV producers from the Sambucinum and Brachygibbosum Clades and strains from other clades, which synthesized toxins such as DAS and T-2TE toxin, were unable to spread in the inoculated wheat heads. While the disease assays herein were only conducted on cv. Apogee, selected for its short life cycle that made it feasible to screen such a large strain subset, the data obtained provide robust genotype-pathogenic phenotype correlations between diverse strains within the FSAMSC. Furthermore, these data provide a framework for subsequent investigations on other cultivars with different degrees of susceptibility to FHB, which will be needed to further elucidate the role of toxins on pathogen aggressiveness, including the novel trichothecenes.
To summarize, phylogenetic, biochemical, and pathogenicity analyses have provided insights into the phylogenetic diversity, trichothecene toxin potential, and pathogenicity to wheat within the FSAMSC. The robust MLST scheme employed, combined with toxin analyses and pathogenicity assays led to several important discoveries (i) the FSAMSC comprises at least 74 genealogically exclusive phylospecies, 33 of which are novel, distributed among six trichothecene toxin-producing clades; (ii) a Novel type A trichothecene-producing clade comprising two sister species; (iii) two putatively novel species within the Sambucinum Clade from South Africa produced NX-3 and two novel type A trichothecenes similar to NX-2 and NX-3 but with a keto or aldehyde group, respectively, at the C-15 position on the trichothecene ring; and (iv) only members of the Graminearum Clade and F. sp. nov.-15 and -16 within the Sambucinum Clade were aggressive on cv. Apogee, which suggests that aggressiveness is likely determined by the type of toxin produced. These results clearly illustrate that robust molecular systematic data are essential for accurately identifying species within the FSAMSC and they provide a firm foundation for ongoing comparative phylogenomic analyses of these agronomically important fusaria, which are focused on advancing global agricultural biosecurity, plant health and food safety.