Fusarium casha sp. nov. and F. curculicola sp. nov. in the Fusarium fujikuroi Species Complex Isolated from Amaranthus cruentus and Three Weevil Species in South Africa

Trials are currently being conducted in South Africa to establish Amaranthus cruentus as a new pseudocereal crop. During recent surveys, Fusarium species were associated with weevil damage in A. cruentus fields. Preliminary studies showed that some of these Fusarium species grouped into two distinct clades within the F. fujikuroi species complex. The aim of this study was to characterize these isolates based on the morphology and phylogeny of the translation elongation factor 1α (TEF1α) gene region, ß-tubulin 2 (ßT) gene region and RNA polymerase II subunit (RPB2), and to determine if these isolates are pathogenic to A. cruentus. Phylogenetic and morphological studies showed that these two clades represent two novel species described here as F. casha and F. curculicola. Both species were shown to have the potential to be pathogenic to A. cruentus during routine greenhouse inoculation tests. While isolations indicate a possible association between these two species and weevils, further research is needed to understand this association and the role of weevils in disease development involving F. casha and F. curculicola in A. cruentus.


Introduction
Food security worldwide currently relies on only a few major crops to feed the growing world population. The introduction of new crops increases crop diversity and, therefore, lowers the exposure to food shortage due to harvest failures. Amaranthus cruentus is one of three Amaranthus species that can be cultivated for grain and has been used as a food source in Central America since 4000 BC [1]. Trials are currently being conducted to establish A. cruentus as a pseudograin crop in South Africa in addition to major crops such as maize and wheat.
In recent years, several new species have been described in the important Fusarium fujikuroi species complex (FFSC) [2][3][4][5][6][7][8][9]. The FFSC includes human and plant pathogens, and several species are known to produce mycotoxins such as beauvericin, fumonisin and moniliformin which can contaminate food sources for human and animal consumption, leading to serious mycotoxicoses [10][11][12]. Novel species descriptions are important because they increase the knowledge on the diversity, geographic distribution, host range, evolution and global movement of this species complex. Currently, more than 50 phylogenetic clades are known from this complex, with a number of undescribed species still awaiting description [2,3,5,6,8,9,13].
Species belonging to the FFSC group into three distinct clades, namely, the African, American and Asian clades [14][15][16]. The American clade represents most of the species in the FFSC [3], followed closely by the African clade [4,11,17]. The geographic topology of Single conidial isolates (129) grouping into two novel clades in the FFSC from a previous study [22] were included in the present study (Appendix Table A1). Isolates originated from plants collected from agricultural plots of A. cruentus in Potchefstroom and Taung, North West Province, central South Africa (January 2013 and March 2013). They were isolated (Table 1, Appendix Table A1) from emergence holes and lesions associated with the weevils Athesapeuta dodonis and Baris amaranti, discoloration in tunnels of H. haerens and larvae from the larval tunnels of H. haerens and adults of Ath. dodonis and B. amaranti (March 2013) from plants with insect damage [22]. An isolate obtained from A. hybridus plants [23] was also included in this study. Selected cultures were stored in 15% glycerol and sterile distilled water and deposited in the culture collection of the National Collection of Fungi, Biosystematics Division, Plant Protection Research Institute, Pretoria, South Africa (PPRI) ( Table 1). Dried cultures of novel species were prepared according to Leslie and Summerell [10] and were deposited in the herbarium of the Agricultural Research Council (ARC), Pretoria, South Africa (PREM) ( Table 1).

DNA Sequence Comparisons
DNA was extracted from 1-week-old cultures grown on potato dextrose agar (PDA, 20%, Biolab, Johannesberg, South Africa). Mycelium was scraped from the surface of PDA plates, freeze dried and ground to a fine powder with 2 mm-diameter metal beads in a Qiagen TissueLyser II cell disrupter (Whitehead Scientific, Cape Town, South Africa). DNA was extracted [24], with DNA concentrations determined on a NanoDrop 2000 (Thermo Fisher Scientific, NanoDrop products, Wilmington, DE, USA) and standardized to 10 ng/µL. Three gene regions shown to be sufficient for species delimitation in the FFSC [3,8,9] were amplified on a T100 TM Thermal Cycler (Bio-Rad, Johannesburg, South Africa). The TEF1α region was amplified using primers EF1 and EF2 [25]; the ß-tubulin 2 region of the ß-tubulin (ßT) gene was amplified using primers T1 and T2 [26]; and two non-contiguous fragments of the RNA polymerase II subunit (RPB2) were amplified using primer sets RPB2-5f2/7cr and RPB2-7cf/11ar, respectively [27,28]. In order to characterize the 129 isolates suspected to represent the two species (Appendix Table A1), only the TEF1α gene was first sequenced [25] to group them into two distinct clades named Clade A and B (data not shown). Three isolates were then randomly selected to represent each clade for further sequencing of the ßT and RPB2 genes. The isolate from Blodgett et al. [23] was also included in the multigene analyses.
Reactions were performed in a total volume of 25 µL, consisting of 40 ng DNA template, 0.25 µL (1 µM) of each primer, 12.5 µL EconoTaq PLUS GREEN 2× Master Mix (1×) (Lucigen corporation, Middleton, WI, USA) and 8 µL nuclease-free water (WhiteSci, Cape Town, South Africa). The PCR program included an initial denaturation step of 95 • C for 3 min followed by 40 amplification cycles consisting of 95 • C for 30 s, 61.3 • C (TEF1α), 52 • C (ßT) or 62.3 • C (RPB2-5f2/7cr) and 62.3 • C (RPB2-7cf/11ar) for 30 s, and 72 • C for 50 s, followed by a final extension step at 72 • C for 5 min. PCR products were viewed with UV light on 2% agarose gels with GelRed TM (Anatech, Johannesburg, South Africa). PCR products were cleaned enzymatically by adding 10 µL of PCR product to 0.5 µL (10 u) of Exonuclease I (Fermentas, Nunningen, Germany), and 2 µL (2 u) of FastAP™ Thermosensitive Alkaline Phosphatase (Fermentas, Germany) at 37 • C for 15 min. The enzymes were inactivated at 85 • C for 15 min and cooled to 4 • C. The same primers as those used in the PCR amplification were used for ßT and RPB2, but for the TEF1α region, internal primers EF3 and EF22T were used [25]. An ABI Prism BigDye ® terminator v. 3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA, USA) was used according to the manufacturer's instructions on an ABI Prism TM 3100 DNA sequencer (Applied Biosystems). Sequences obtained were viewed and edited, if required, with GENEIOUS 7.1.9 (Biomatters Limited, Auckland, New Zealand).
Phylogenetic analyses were performed with the software package PAUP* 4.01b10. Phylogenetic analyses were conducted for each gene region separately, as well as combined, as previously described [3,[14][15][16]33]. Analyses were conducted with maximum parsimony (MP) (heuristic search with 100 random sequence additions). A 1000 bootstrap replication test was performed to determine the support of branches [34] of the most parsimonious tree for the various dataset combinations. This was conducted after the exclusion of uninformative sites, with the heuristic search with 100 random sequence additions and tree bisection reconnection (TBR) branch swapping selected, and MAXTREES set to 1000 to allow for the completion of analysis.
Additional phylogenetic analyses were conducted based on maximum likelihood (ML) and Bayesian inference (BI). The correct models for the datasets were identified using jModeltest 0.0.1 [35]. The TrN + G model [36] was shown to be appropriate for the TEF1α dataset, the TIM2 +G model for the ßT and RPB2 datasets [35] and the TIM2ef+G model for the combined dataset [37] (Table 2). Maximum likelihood analyses were performed with PhyML 3. A 1000-replicate bootstrap analysis was conducted to assess the confidence of the branch nodes in the phylogenetic trees. Bayesian analyses were performed with the MrBayes plug-in for GENEIOUS 7.1.9 using the same evolutionary model, a chain length of 1,100,000 for four heated chains and a burn-in of 20,000.

Morphology
To study the morphology of the two new species, the three single conidial isolates ( Table 1) selected as representative isolates for each of the two novel clades observed (PPRI 21883, PPRI 20468 and PPRI 20462, and PPRI 20458, PPRI 20386 and PPRI 20464, respectively) were grown on synthetic low-nutrient agar (SNA) [38] and carnation leaf agar (CLA) [39], unwrapped, for 7 to 14 d at 25 • C under near-ultraviolet light [6,40]. Fungal structures formed were mounted on microscope slides in 85% (vol/vol) lactic acid (Sigma-Aldrich, St Louis, MI, USA) or were examined directly on agar by cutting out a small block of agar and placing it on a microscope slide. Fifty measurements were recorded for each character. Measurements and photographs were captured with an Olympus BX53 light microscope and DP75 camera (Olympus, South Africa). Illustrations were prepared by free hand by observing morphology on SNA and CLA carefully.
Colony characteristics and growth rates of the unknown clades were determined for the same three representative single conidial isolates of each clade as used for the microscopic study (Table 1). Discs were taken from the edges of actively growing cultures on PDA and transferred to the centers of 90 mm Petri dishes containing PDA. Five plates per isolate were placed in the dark and incubated at 15-35 • C at five degree intervals. The average growth rate per day was calculated for each isolate by performing two perpendicular measurements of the growth diameter, each day, until the fastest-growing culture had covered the surface of the plate. The trial was repeated once. Analysis of variance (ANOVA; α = 0.05) of the data was performed using base R functions to test if results from the two trials could be pooled, and no significant difference was observed between the trails. Pooled data were analyzed with means separated using an adjusted Bonferroni test function from the 'agricolae' package, in order to account for the potential family-wise error rate [41], as more than three groups were being compared. Colony characters were also recorded under near-UV light at 25 • C. Colony colors were described using the charts of Rayner [42].

Pathogenicity
Pathogenicity experiments were conducted with the same isolates used for the phylogenetic and morphological studies (Table 1). Amaranthus cruentus plants were established in a greenhouse at 27/21 • C day/night conditions, which are optimal for the growth and germination of the plant [43]. Seeds were sown in 2 L pots lined with plastic bags, with a soil compost mix (80:20 vol/vol), and watered to field capacity. Plastic bags were sealed and opened at seedling emergence (3-5 d). At 6 weeks, plants were thinned to three plants per pot. Pots were watered daily to field capacity and were fertilized at 10 weeks with Nutrifeed (Starke Ayres, Bredell, South Africa) (65 g/kg N, 27 g/kg P, 130 g/kg K, 70 g/kg Ca, 22 g/kg Mg, 75 mg/kg S, 1500 mg/kg Fe, 240 mg/kg Mn, 240 mg/kg B, 20 mg/kg Cu and 10 mg/kg Mo) as per the instructions of the supplier. Plants were cultivated for 12 weeks.
Ten plants were inoculated with each isolate after 12 weeks. A small section (5 × 5 mm) of the epidermis of the stem was removed with a scalpel, and an agar plug cut from an actively growing culture was placed mycelium-side-down into the wound [23,44]. Non-colonized agar Diversity 2021, 13, 472 7 of 25 plugs were applied as controls. To prevent desiccation of agar plugs, wounds were sealed with a strip of Parafilm (Sigma-Aldrich, Johannesburg, South Africa) [23,44]. The length of lesions formed was measured (if present) after four weeks. The trial was repeated once. Pieces were cut from the margins of necrotic tissues and placed on CLA and PDA to re-isolate the inoculated fungus. The identities of the re-isolated cultures were confirmed by studying the morphology and TEF1α sequencing.
Analysis of variance (ANOVA; α = 0.05) of the data was performed using base R functions. ANOVA was applied to determine if experimental replicates differed significantly. Trials differed significantly and were thus not pooled for further analyses. Means were separated using an adjusted Bonferroni test function from the 'agricolae' package [41], as there were more than three groups being compared. Data processing and analyses were performed with R version 4.0.2 [45] within R Studio version 1.2.5042 [46]. Data exploration, wrangling and visualization were conducted using the 'Tidyverse' package [47].

Fungal Isolates
A total of 129 isolates were obtained that grouped into two novel clades in the FFSC based on preliminary TEF1α sequence identities (Appendix Table A1). This included two isolates from Blodgett et al. [23] from galleries in A. hybridus associated with H. haerens larvae, and cankered stems of A. hybridus from Bloemfontein. The remaining 127 isolates [22] included 66 isolates from lesions below the emergence holes of Ath. dodonis

DNA Sequence Comparisons
Seven isolates representing the two novel clades were included for final phylogenetic analysis (Table 1, Figure 1). These included one isolate from Blodgett et al. [23], and six representative isolates [22] based on the preliminary phylogenetic analyses of TEF1α for all isolates. The datasets for TEF1α, ßT and RPB2 and the combined dataset consisted of 53 taxa ( Table 2). The alignment lengths of TEF1α, ßT and RPB2 were 668, 505 and 1637 bp in length, respectively, and the combined alignment was 2810 bp long ( Table 2).
Maximum parsimony, ML and BI analyses of the three datasets (data not shown) generated trees with similar topologies and resembled the phylogenetic positions of those previously published for the TEF1α, ßT, RPB2 and combined datasets [3,8,9,11,14,16,33]. The three distinct clades known in the FFSC, namely, the African, American and Asian clades, were observed. Isolates from this study all grouped into the African clade.
Etymology: The species name 'casha' denotes the Zulu for 'hiding' since it was first isolated from cankered stems and Hypolyxis haerens larval galleries in Amaranthus hybridus by Blodgett et al. [23]. At that time, it was only identified based on morphology, and this new species has, therefore, been 'hiding' and remained undescribed since 2004.
Macroconidia on CLA abundant to absent, slender, straight to slightly curved, 20.    Figure 2G-L and Figure 4.   Etymology: Curculi-(Latin) refers to weevils, indicating its close association implied by the direct isolation from the insect and related damage, and -cola means dweller or inhabiting.

Pathogenicity
Lesion lengths from the first and second trial were significantly different (p > 0.05), and data were, therefore, analyzed separately (Table 4). For the first trial, two of the three isolates for F. casha (Clade B; PPRI 21883 and PPRI 2046) and one of the three isolates for F. curculicola (Clade A; PPRI 20386) formed lesions that were significantly larger (p < 0.05) than the measurements for the negative control (Table 4; Control). In the second trial, the same two of the three isolates for F. casha (Clade B) and two of the three for F. curculicola (Clade A; PPRI 20464 and PPRI 20458) formed lesions that were significantly larger (p < 0.05) than the measurements for the negative control (Table 4; Control). For wounds of control plants, there was a degree of callus and l formation in some of the inoculations ( Figure 5A). For both the first and second trials, lesions made by F. casha isolates were not significantly different in size from those made by F. curculicola isolates ( Table 4). All fungi were reisolated from lesion margins of the tested A. cruentus plants, and their identities were confirmed, fulfilling Koch's postulates.

Discussion
This is the first study to identify Fusarium species occurring in Amaranthus species based on both phylogeny and morphology instead of only morphology. This approach led to the description of two novel species in the FFSC, one of which is a representative culture that was already isolated in 2004 from A. hybridus and misidentified as F. subglutinans at the time due to a lack of molecular data [23]. Both of these species showed the potential to be pathogenic to A. cruentus through routine pathogenicity techniques. These species should, therefore, be monitored due to their prominence and potential to cause disease in this crop.
Species in the FFSC are difficult to identify based solely on morphology due to the overall shortage of diagnostic morphological characters and the similarity of these characters between some species [14,15]. However, the two new species, F. curculicola and F. casha, have differentiating morphological characteristics from each other and from morphologically similar species. The microconidia of F. casha are quite variable, being oval, short clavate, ellipsoid, curved, c-shaped to obpyriform and arranged in false heads, while in F. curculicola, the microconidia are oval to obovoid, ellipsoid, short clavate, fusiform

Discussion
This is the first study to identify Fusarium species occurring in Amaranthus species based on both phylogeny and morphology instead of only morphology. This approach led to the description of two novel species in the FFSC, one of which is a representative culture that was already isolated in 2004 from A. hybridus and misidentified as F. subglutinans at the time due to a lack of molecular data [23]. Both of these species showed the potential to be pathogenic to A. cruentus through routine pathogenicity techniques. These species should, therefore, be monitored due to their prominence and potential to cause disease in this crop.
Species in the FFSC are difficult to identify based solely on morphology due to the overall shortage of diagnostic morphological characters and the similarity of these characters between some species [14,15]. However, the two new species, F. curculicola and F. casha, have differentiating morphological characteristics from each other and from morphologically similar species. The microconidia of F. casha are quite variable, being oval, short clavate, ellipsoid, curved, c-shaped to obpyriform and arranged in false heads, while in F. curculicola, the microconidia are oval to obovoid, ellipsoid, short clavate, fusiform and arranged in false heads and long chains. Fusarium casha forms macroconidia that have distinctly notched basal cells and that are 0-5 septate, while those of F. curculicola are 0-3 septate and shorter than those of F. casha. No chlamydospores were observed in both species. F. curculicola also formed intercalary phialides.
Interesting features of F. casha and F. curculicola are the occurrence of pleurophialides with phialidic openings on the surface of the running hyphae. The conventional phialides normally associated with Fusarium can then be referred to as orthophialides. These phialide morphologies were used in the sense of Gams [51]. It is unclear if other species in the FFSC have these characteristics since the morphology of conidiogenous cells forming macroconidia is poorly published overall. Such details are also usually not noted for microconidial conidiogenous cells. Species descriptions including such morphology [3,5,7,8,52,53] together with descriptions from this study indicated that a conidiogenous cell morphology can prove diagnostic. For example, they indicated if conidiogenous cells are being formed in sporodochia or singly, whether they are found mingled with those bearing microconidia, whether phialides have different morphologies and the morphology of the conidiophore. Such details will enable more in-depth morphological identifications of species than what is currently possible.
No sterile, coiled or circinate hyphae were observed for F. casha and F. curculicola. However, vegetative or running hyphae bearing conidiogenous cells often formed coiling circles consisting of numerous loops for both species. This can be confused with true circinate or coiled hyphae, as introduced for F. circinatum [49] and also present in F. pseudocircinatum, F. sterilihyphosum, F. parvisorum, F. mexicanum and F. tupiense as their diagnostic feature [3,49,[54][55][56]. True circinate hyphae are thick, coiling, sterile and septate that can branch themselves and that are borne on running hyphae. The hyphal coilings of the running hyphae observed for F. casha and F. curcuricola are found in other Fusarium species, but their presence has not been noted well and is not known among the morphological characteristics usually considered when observing the morphology of FFSC species. Such a feature has, however, been noted in the recent species description of F. ficicrescens [13].
The present study increases our knowledge on the diversity and host range of the FFSC. The two novel species both group in the African clade of the FFSC. Based on the biogeographic hypotheses, the African, American and Asian clades in the FFSC are associated with the origin of the respective plant hosts from which the species of the FFSC were isolated [14]. The hosts of F. curculicola (A. cruentus) and F. casha (A. cruentus and A. hybridus) are, however, originally South American and not African. The lack of association can be ascribed to the fact that these two species have thus far not been found in South America and were introduced into Africa. Alternatively, these species are proposed to have an alternate native host in Africa and could then have the ability to infect A. cruentus and A. hybridus when these species were introduced to Africa. This is also possibly true for other species in the African clade. Fusarium coicis, F. verticillioides and F. musae, which are closely related to F. curculicola, have hosts that originate from Australia, Central America and Asia, respectively [6,50,57]. Fusarium volatile, also closely related to F. curculicola, is a human pathogen first described from a patient in French Guiana and not Africa [9]. Fusarium napiforme, F. ramigenum and F. tjaetaba are closely related to F. casha and have hosts that originate from Australia, the Middle East and Africa, respectively [6,9,49]. It is thus clear that the FFSC is under-represented in terms of reported host and geographical occurrences for species, and that more surveys on various hosts and in different geographic areas will possibly improve the phylogenetic resolution that will support or disprove the biogeographic origin hypothesis.
Not all isolates of F. casha and F. curcicola caused lesions in the pathogenicity trials on A. cruentus. These basic inoculations indicated that isolates of F. casha and F. curculicola have the potential to cause stem lesions, and that they could contribute to the disease symptoms originally observed in the field. However, more extensive pathogenicity trials may be needed to define this more closely, as indicated by the variation observed by some isolates in causing or not causing lesions. Furthermore, comparisons of F. casha and F. curcicola with Fusarium species shown not to cause disease symptoms on A. cruentus, or consistent lesions, will enable better comparisons.
Fusarium curculicola and F. casha were both isolated from weevils and their associated damage to amaranth. The presence of Fusarium species in association with weevil damage could either be due to the weevils vectoring the fungi, the fungi possibly serving as a food source or, alternatively, Fusarium species occurring as endophytes in amaranth and possibly being stimulated to become pathogenic due to the feeding activity of the weevils. This has been shown for a number of insect-associated Fusarium species [58]. Pathogenicity tests showed that both species newly described in this study have the potential to be pathogenic to A. cruentus without interaction with weevils. Further research is, however, needed to establish the role that these weevils and their feeding activity play in disease development caused by F. casha and F. curculicola in A. cruentus, and the type of association that exists between these fungi and the weevils. This study did not explore this further or attempt various inoculation techniques to study these interactions or the effect of the weevils alone. Such studies will have important implications for the control of both the weevils and their associated fungal pathogens should they be dependent on each other.
Fusarium casha and F. curculicola were prevalent in A. cruentus stems. Together with damage caused by weevils in amaranth stems [22], this can prevent the movement of nutrients, resulting in reduced yield. Amaranthus cruentus currently planted in central South Africa is not clonal and, therefore, has genetic diversity that can account for variability in disease development. It is, therefore, important to consider F. casha and F. curculicola in future breeding and integrated pest management programs for A. cruentus. Since our study showed that F. casha can also infect another species, namely, A. hybridus, such programs most likely could be applied to other species and cultivars of Amaranthus.  Table 1.

Conflicts of Interest:
The authors declare no conflict of interest. All the experiments undertaken in this study comply with the current laws of the country where they were performed.