From April 2018 until December 2019, several field trips were carried out focusing on the presence of banana Fusarium wilt in northern Viet Nam. During these surveys, banana Fusarium wilt samples were collected at 19 locations in three large geographic regions; North-eastern region, North Central region and Red River Delta, with most specimens collected in the latter region (Table 1, Fig. 1). Fusarium infected banana plants were identified by following a set of diagnostic characters in which (mostly older) leaves were clearly yellow (initiated from the leaf margin) or even completely collapsed, halfway along the petiole forming a ring of dead leaves around a dying plant, combined with brown discoloration and longitudinally fissuring of the pseudostems leaf sheaths (Fig. 2). From symptomatic plants observed in the field, discoloured brownish vascular tissue was collected from pseudostems and roots. Subsequent to the collection, infected tissue samples were stored in paper bags and put in a refrigerator or cooling box to avoid quality loss upon the further analysis in the molecular lab. For each sample collected, notes were taken about the altitude, longitude and latitude, site location, and the host specimen. Collected Fusarium samples were stored at the Plant Resources Center (PRC), Ha Noi, Viet Nam (Table 1).

Figure 1. 

Distribution map of localities in northern Viet Nam where Fusarium wilt was observed. Colours indicate different Fusarium strains or species. Squares indicate Fusarium infections of wild bananas; circles indicate infections of cultivated bananas.

In order to observe possible morphological differences between the Fusarium wilt isolates collected from the wild and cultivated northern Vietnamese Musa accessions, we followed the approach of Groenewald et al. (2006) in which different Foc isolates were grown on different media. Infected discoloured pseudostem tissue samples were cut into 2–3 cm pieces and placed on the Komada medium (Komada 1975). After a few days, fungal Fusarium colonies were transferred to plates with different media and then put in a growing chamber at 25 °C until the colonies reached 2–3 cm. The different isolates were grown on four distinct nutritious media to observe the Fusarium wilt in different culture conditions: PDA (Potato Dextrose Agar), SNA (Spezieller Nährstoffarmer Agar), CLA (Carnation Leaf Agar), and OMA (Oatmeal Agar) (Nirenberg 1976). The PDA medium consisted of 200g potato dextrose, 20g D-glucose, and 20g agar dissolved in 1000ml distilled water, whereas the SNA medium consisted of 1g KH₂PO₄, 1g KNO₃, 0.5g MgSO₄•7H₂O, 0.5g KCl, 0.2g D-glucose and 0.2g D-sucrose dissolved in 1000ml distilled water. The CLA medium contained aseptic carnation leaves and 20g agar dissolved in 1000ml distilled water. The OMA medium consisted of 50g oatmeal and 20g agar dissolved in 1000ml distilled water. The growth of the Fusarium isolates on the different media took place under in-vitro conditions with temperatures between 23 and 27 °C (Pérez et al. 2003). After seven days of incubation, the developing colonies were morphologically investigated under a light microscope (400x magnification). The coloration of the colony, the morphology, and the size of the conidia were determined. The colony reverse colour was determined on PDA medium after incubation at room temperature, using the colour charts of Rayner (1970). In addition to colony colour, the aroma of the different cultures was assessed as a strong rank odour generated by mature cultures which is a typical characteristic for TR4 infections. In the first stage of culturing, we characterized the isolates as Fusarium spp. emanated from mycelium morphology and the presence of different types of conidia. The study of Maryani et al. (2019a) was used as a reference to classify the Foc lineages into different sublineages further. All obtained Fusarium isolates were stored in the Plant Resources Center (PRC), Ha Noi, Viet Nam.

Figure 2. 

A overall view of a banana plant infected by Fusarium wilt B detailed view of wilted plant C radial cutting of Fusarium-infected banana pseudostem D tangential cutting of Fusarium-infected banana pseudostem.

In order to extract high-quality DNA from the Fusarium wilt isolates collected and cultured, we used the pure mycelium cultures generated for the morphological characterization of the banana wilt. Total genomic DNA was isolated using a modified CTAB protocol based on the study of Lin et al. (2008) and Dellaporte et al. (1983). After the addition of 5ml TNE buffer (100 mM Tris-HCl, 50 mM EDTA, 50 mM NaCl, 8 μM β-mercaptoethanol, 1% SDS, pH 8.0) to the sampled mycelium, the samples were incubated for 1 h (65 °C). Subsequent to the lysis phase, 1.66ml NaOAc (5M) was added and centrifuged. Chloroform-isoamylalcohol (24/1 v/v) extraction was done twice, followed by isopropanol precipitation at -32 °C for 12h. After centrifugation at 4 °C, the pellet was washed twice (75% ethanol), air-dried, and dissolved in 100µl TE buffer (10mM TrisHCl, 0.1mM EDTA; pH 8)

Amplification reactions of rpb1, rpb2, and tef1a were carried out using standard PCR (20µl). Reactions were initiated with a 3 min heating at 95 °C followed by 30 cycles consisting of 95 °C for 30s, 55–65 °C (rpb1 and rpb2) and 53°-59 °C (tef1a) for 60s, and 72 °C for 60s. Reactions ended with a 3 min incubation at 72 °C. Primers designed by O’Donnell et al. (1998) were used to sequence tef1a, whereas primers for rpb1 and rpb2 were adopted from O’Donnell et al. (2010). PCR products were purified using an ExoSap purification protocol. Purified amplification products were sequenced by the Macrogen sequencing facilities (Macrogen, Seoul, South Korea).

Raw sequences were assembled using Geneious Prime (Biomatters, New Zealand). Automatic alignment was conducted with MAFFT (Katoh et al. 2002) using an E-INS-i algorithm, a 100PAM/k = 2 scoring matrix, a gap open penalty of 1.3, and an offset value of 0.123. Manual fine-tuning of the aligned dataset was performed in Geneious Prime.

Fusarium sequence data of rpb1, rpb2, and tef1a was extracted from GenBank (September 20, 2020) using the ‘NCBI Nucleotide extraction’ tool in Geneious Prime. Together with the newly generated sequences for the 19 Vietnamese Fusarium wilt accessions, the total sequence data matrix consisted of 529 specimens divided over 201 species (Suppl. material 1: Table S1). Of those, 72 species belong to different closely related genera of Fusarium within the Nectriaceae family were chosen as outgroup (Albonectria, Bisifusarium, Cosmospora, Cyanonectria, Cylindrocarpon, Fusicolla, Geejayessia, Luteonectria, Macroconia, Microcera, Neocosmospora, Pycnofusarium, Rectifusarium, Setofusarium; sensu Crous et al. 2021). Newly generated sequences were deposited in the GenBank sequence database (Table 1). Furthermore, in order to compare the newly collected Vietnamese Foc accessions with the known VCGs, the sequence dataset included Foc samples representing all VCGs (see Table S1; Ordonez et al. 2015), except for VCG01212 and VCG0129, from which only one locus was available, thereby causing phylogenetic biases due to the occurrence of too much missing data.

Possible incongruency between the different datasets was inferred by conducting an ILD test (Farris et al. 1995) as implemented in PAUP* v.4.0b10 (Swofford 2003) with the following parameters applied: simple taxon addition, TBR branch swapping, and heuristic searches of 1000 repartitions of the data. Despite the well-known sensitivity of the ILD test (Barker and Lutzoni 2002), the results of this test were compared in light of the resolution and support values for each of the single gene topologies. As a result, the possible conflict between data matrices was visually inspected by searching for conflicting relationships within each topology (obtained per single sequence data matrix) that were supported by a Maximum Likelihood (ML) support value > 70% (hard vs. soft incongruence; Johnson and Soltis 1998; Pirie 2015). A conflict was assumed to be significant if two different relationships for the same set of taxa (one being monophyletic and the other non-monophyletic) were observed in rival trees. ML analyses were conducted under the RAxML search algorithm (Stamatakis 2014) with the GTRGAMMAI approximation of rate heterogeneity for each gene. ML bootstrapping was carried out on five hundred bootstrapped datasets using the RAxML Rapid bootstrap algorithm (ML-BS).

The best-fit nucleotide substitution model for each dataset was selected using jModelTest 2.1.4. (Posada 2008) out of 88 possible models under the Akaike information criterion (AIC). The GTR+I+G, TVM+G, and HKY+I+G were determined as the best-fit substitution models for rpb1, tef1a, and rpb2 markers, respectively. Consequently, we used a mixed-model approach to apply different evolutionary models on each DNA region of the combined dataset (Ronquist and Huelsenbeck 2003). Bayesian inference analyses were conducted with MrBayes v3.2.6 (Ronquist et al. 2012) on three individual data partitions and a combined data matrix. Each analysis was run twice for 20 million generations. Trees were sampled every 5000^th generation. Chain convergence and ESS parameters were inspected with TRACER v.1.4 (Rambaut and Drummond 2007). Only nodes with Bayesian posterior probabilities (BPP) above 0.95 were considered as well supported by the data (Suzuki et al. 2002).

Sequence characteristics of all data matrices analysed are summarized in Table 2. Despite the fact that sometimes not all gene markers could be sequenced, their absence did not influence the overall phylogenetic results, as sufficient nucleotide variation was present. The partition homogeneity test found no significant incongruence between all three sequence datasets (with all P-values being larger than 0.05). Visual examination of the two different partitions of the combined dataset corroborates this congruency analysis.

Phylogenetic analyses of the 19 Fusarium wilt isolates found in various northern Vietnamese bananas showed that although overall morphological characterisation pointed towards F. oxysporum f. sp. cubense, it was clear that they had various evolutionary origins (Fig. 3c, d). Of the 19 accessions analysed, two (FOC61 and FOC56) were placed within the F. odoratissimum clade (as defined by Maryani et al. 2019a, Fig. 3d) and for which pathogenicity tests by Maryani et al. (2019a) showed that the members of this group caused infections in Cavendish and Gros Michel AAA banana varieties. In addition, VCG 01213 and VCG 01216 are positioned close to FOC61 and FOC56. As a result, two of the 19 (10.5%) northern Vietnamese Fusarium wilt isolates are assumed to be Foc-TR4 (also taking the morphological characterization into account). Interestingly, one of the two isolates characterised as Foc-TR4 (FOC61) infected a Cavendish plantation in Vinh Phuc province, whereas the other infection of Foc-TR4 (FOC56) took place on ABB Tay banana cultivars situated on a smallholder farm in Nam Dinh province (Table 1). The largest group (13 accessions; 68.5%) of Fusarium wilt isolates in northern Vietnamese bananas belong to the recently delineated F. tardichlamydosporum clade (Fig. 3d). Pathogenicity tests carried out for this clade by Maryani et al. (2019a) have indicated a large infection rate in Gros Michel cultivars for this lineage, and therefore members of the F. tardichlamydosporum clade are consequently classified as Foc-Race 1. Furthermore, the isolates that fell within the F. tardichlamydosporum were also most closely related to VCG 0125, a known Foc-Race 1 representative. In northern Viet Nam, infections of Foc-Race 1 occurred both on wild and cultivated accessions. For the cultivated accessions, the Foc-Race 1 was only found on the ABB Tay banana cultivar, yet was clearly spread in northern Viet Nam as it was found in eight different provinces (Ha Giang, Yen Bai, Lao Cai, Bac Giang, Nam Dinh, Ha Nam, Tuyen Quang, Ha Noi; Table 1). Most interestingly, Foc-Race 1 was also identified (isolate FOC5) in an individual of the wild banana Musa lutea (section Callimusa). Here the infected accession grew sympatrically with other individuals of M. lutea as well as with M. itinerans. The area where this infection occurred was a steep, abandoned rice terrace in Yen Bai province where hundreds of individuals of both wild species co-occurred and were rather close to one of the smallholder farms where Foc-Race 1 was also detected (isolate FOC6-1). In addition to the Foc-Race 1 infections caused by Fusarium tardichlamydosporum, F. duoseptatum is also classified as a Foc-Race 1 Fusarium wilt (see Maryani et al. 2019a). An infection of this latter Foc isolate (FOC38) was found only once in northern Viet Nam (Nghe An province; c. 5% of the Fusarium wilt infections), where it infected the ABB Tay banana cultivars that were grown on a smallholder farm. The VCGs that occurred in the same clade as FOC38 are VCG 01223 and VCG 01217, with the latter being known as a Foc-Race 1 representative (e.g., Katan 1999; Fraser-Smith et al. 2014).

Figure 3. 

Maximum Likelihood topology obtained via heuristic search algorithm of the combined rpb1, rpb2 and tef1a data matrix. Bootstrap support (ML-BS) values above 50 are indicated with a dot, ML-BS values above 75 are indicated with an asterisk. No indication above the branches indicates a ML-BS value below 50. Newly included accessions are indicated in red. FOSC: Fusarium oxysporum species complex, FFSC: Fusarium fujikuroi species complex.

In addition to the Foc-Race 1 and Foc-TR4 infections, two Foc isolates (FOC4 and FOC10) were found in northern Viet Nam (10.5%) that belong to the recently described F. cugenangense (Maryani et al. 2019a; Fig. 3d). Hitherto, this Fusarium species was considered to be strictly Indonesian (see Maryani et al. 2019a). Pathogenicity tests conducted for representatives of F. cugenangense by Maryani et al. (2019a) have demonstrated that it only causes a mild infection in Gros Michel and Cavendish and is regarded as non-pathogenic for the above-mentioned AAA cultivars. However, our results clearly show that the infection of this isolate also occurred on ABB Tay banana cultivars in northern Viet Nam, where it had a large impact on the fitness of the infected host plants. Although additional confirmation is needed, Maryani et al. (2019a) assume that representatives of the F. cugenangense clade should be considered as Foc-Race 2 (Maryani et al. 2019a; Fig. 3d), yet more thorough analyses need to be carried out in order to further confirm this hypothesis. The VCG that occurred in the same clade as FOC4 and FOC10 is VCG 01221.

A final Fusarium wilt infection (FOC58) that was regarded upon collection in the field and during morphological screening as a Foc infection (c. 5% of the Fusarium wilt infections observed in this study) was not situated in the F. oxysporum species complex (FOSC) but was a distinct lineage sister to F. fujikuroi (Fig. 3c), a well-known pathogen of rice (e.g., Wulff et al. 2010; Choi et al. 2018). This Fusarium isolate was the prime infection source of ABB Tay bananas cultivated on a small plantation for local use in Nam Dinh province (Fig. 1).

To better manage the significant threat of Foc dispersion in northern Viet Nam, the correct identification and abundance of the Foc strains that cause Fusarium wilt in bananas in the region are necessary. This is the basis for eradication-confinement and suppression-contention measures (Perez-Vicente et al. 2014). Since the survey of Vakili et al. (1968), Foc Race 1 has been considered as the main Foc infecting edible bananas in Viet Nam. With the emergence of Foc TR4 it remained unclear how abundant this new pathogenic Foc strain had become in Viet Nam. Although officially present in Viet Nam for only a few years (Hung et al. 2017; Zheng et al. 2018), Foc TR4 was already observed on Cavendish bananas in 1998 in Southern China (Hu et al. 2006). A few years later, in 2002, Foc TR4 was also found in Chinese regions adjacent to northern Viet Nam (Hu et al. 2006; Li et al. 2013). With the current shift of Cavendish cultivation in Asia from China to its neighbouring countries Laos, Myanmar, and Viet Nam, there is also an active spread of Fusarium pathogens through transportation of planting material, farming equipment, and contaminated soil from China (Zheng et al. 2018), so that Foc TR4 can quickly become the most dominant Foc race in Viet Nam affecting banana cultivation.

The present study applies the FOSC species delimitation concept of Maryani et al. (2019) to delineate the Foc lineages sampled in northern Viet Nam more thoroughly. Furthermore, incorporating the different VCGs in the current phylogenetic dataset allowed us to link the Vietnamese Foc isolates with one of the currently known VCGs that have been assessed in the past. Based on the compatibility of the novel material with the VCGs present in the same clade and their specific species allocation following the species delineation concept of Maryani et al. (2019), we linked the northern Vietnamese Foc isolates to one of the known Foc Races. Accordingly, our results show that Foc isolates that are phylogenetically situated within a lineage containing Foc Race 1 are among the most common isolates in northern Viet Nam, causing 74% of all the infections. A more in-depth molecular characterisation shows that of these, 13 out of 14 isolates are representatives of F. tardichlamydosporum, whereas one isolate is a representative of F. duoseptatum. Whereas F. tardichlamydosporum is commonly present throughout the northernly oriented Northeastern region and Red River Delta, F. duoseptatum is not present in these more northerly oriented geographic regions but occurs in the more centrally oriented North Central region in Viet Nam. Also, from a global distributional perspective, F. tardichlamydosporum is much more widespread than F. duoseptatum, with the first species located in Australia, Indonesia, Malaysia, Honduras, and Brazil, and the latter is only known to date from Indonesia and Malaysia (Maryani et al. 2019a).

Interestingly, the Foc isolates that are phylogenetically situated within the F. odoratissimum species lineage (linked to Foc-TR4 infections; Maryani et al. 2019a) account for only 10% of the Fusarium wilt in northern Viet Nam, demonstrating that Foc-TR4 has not yet become a dominant banana pathogen, unlike other countries in Asia where there is a tendency to grow Cavendish cultivars as large monocultures, such as in China, the Philippines, and Taiwan. The F. odoratissimum accessions found in the current study were located in the River Delta region of northern Vietnam, provinces Vinh Phuc and Nam Dinh, which are rather distant from Foc-TR4 infected regions in Southern China. This indicates a gradual spreading in Viet Nam of F. odoratissimum (=F. oxysporum f. sp. cubense TR4) towards the south as the TR4 isolates analysed by Zheng et al. (2018) were collected in the upper North of Viet Nam in the Lao Cai province at only a few kilometres from the border with China (Yunnan). At the moment, it is unclear whether the occurrence of Foc-TR4 in Viet Nam is still in an initial lag phase, with the potential of largely increasing its distribution range in the country if conditions would enable the disease to spread (Pegg et al. 2019). Especially the replacement of citrus plantations and maize fields by Cavendish monocultures provides an ideal basis for Foc-TR4 to rapidly spread as plants available for infection become less limited. A more worrying observation is that Foc-TR4 is not only found in Cavendish bananas in Viet Nam, but it also poses a threat to local banana varieties as is observed in the current study where F. odoratissimum is found in ABB Tay banana cultivars.

The current study demonstrates that the presence of F. cugenangense (linked to Foc-Race 2 infections; Maryani et al. 2019a) also accounts for 10% of the Fusarium wilt in northern Viet Nam. In general, Foc-Race 2 infections occur on triploid ABB Bluggoe varieties and its closely related cooking cultivars (Jones, 2000). Besides Bluggoe cooking bananas, Foc-Race 2 also infects the tetraploid AAAA Bodles Altafort hybrid between Gros Michel (AAA) and Pisang Lilin (AA) (Stover and Simmonds 1987). In addition, experimental infection of Ensete ventricosum demonstrated that this important Ethiopian crop is highly susceptible to Foc-Race 2 (Ploetz, 2005). With the confirmation of F. cugenangense also present in representatives of infected ABB Tay banana cultivars, it is clear that Fusarium wilt caused by Foc-Race 2 is potentially more widespread than has often been assumed.

In addition to the Fusarium isolates collected from northern Vietnamese bananas belonging to FOSC, an infection with symptoms similar to Foc wilt was observed, yet the cultured isolate did not belong to FOSC. The morphological colony characteristics were comparable to those observed for FOSC cultures through having a pale purple colony reverse colour that became whitish towards the periphery with a dry colony surface appearance. However, when considering its phylogenetic position within the Fusarium genus, this isolate did not fall within F. oxysporum representatives, but was a member of the F. fujikuroi species complex (FFSC) where it is the sister lineage of F. fujikuroi. It is not uncommon that several Fusarium species cause the same disease pattern as this phenomenon has also been identified in mango deformity (Lima et al. 2009) and sugar beet wilt (Burlakoti et al. 2012). Within the FFSC, some species are known to be pathogenic for some Musa cultivars (F. proliferatum, F. verticillioides, F. sacchari, F. lumajangense, F. desaboruense and F. musae; Maldonado-Bonilla et al. 2019; Van Hove et al. 2002; Huang et al. 2019; Maryani et al. 2019b), yet to date, no other species of the FFSC - except for the above-mentioned - was identified as a pathogen for Musa. From a phylogenetic point of view, the novel Fusarium isolate that was obtained from the infected tissue of a triploid ABB Tay banana cultivar in Nam Dinh province is sister to F. fujikuroi. Fusarium fujikuroi is a widespread phytopathogen causing the bakanae disease in various Oryza sativa cultivars (rice), but it is also known to have a major impact on many other economically important crops (e.g., maize, wheat). This result increases our knowledge of the diversity of the Fusarium species that cause wilt symptoms on bananas. More importantly, it also demonstrates the urgent need to accurately identify plant pathogens that are morphologically very difficult to distinguish from each other in the field.

Although mainly observed on cultivated bananas, Foc has also been rarely recorded on wild Musa species (Ploetz and Pegg 2000). Waite (1963) noticed that Fusarium wilt also occurred on M. acuminata, M. balbisiana, M. schizocarpa and M. textilis. Since these specific Musa species belong to different sections in the genus - section Musa and Callimusa (Australimusa) - it is therefore likely that Foc can also infect other wild bananas. The current finding of a Foc isolate - phylogenetically situated within a Foc-Race 1 lineage - on a wild representative of M. lutea, could indicate that wild species are perhaps more susceptible to Fusarium wilt than previously assumed. Besides the visual symptoms of Fusarium wilt on this single specimen of M. lutea, none of the hundreds of individuals of M. itinerans and M. lutea surveyed in the same population showed any sign of Foc infection. This lack of visual symptoms either implies that F. oxysporum f. sp. cubense could have been absent from all those other wild accessions or that this Fusarium isolate related to Foc-Race 1 was present but failed to cause the disease in the other wild bananas. If the latter assumption is true, this could indicate that the pathogen has not necessarily co-evolved together with its host in Southeast Asia as postulated by Vakili (1965). Nevertheless, it is evident that F. oxysporum is omnipresent throughout the native distribution range of the Musa genus, and its infections take place when the plant is weakened due to external biotic or abiotic stressors, and the endophytic equilibrium is disturbed.

The current study demonstrates that bananas in northern Viet Nam which are infected by Fusarium wilt are characterised by a various range of Fusarium species (F. cugenangense, F. odoratissimum, F. duoseptatum and F. tardichlamydosporum) that belong to the Fusarium oxysporum species complex. Of these, the latter was most commonly present in cultivated bananas infected by Fusarium wilt, whereas the other species are less prominently present, yet in equal amounts. Fusarium tardichlamydosporum also occurred in a wild accession of Musa lutea, indicating that wild bananas might function as a sink for Foc.