# Source paper archive

- Source title: Evaluating novel sources of resistance to Fusarium oxysporum f. sp. cubense tropical race 4 (TR4) in a global banana diversity panel
- Author(s): Olave-Achury, C., Ponndorf, D., Chaparro Garcia, A., Hall, A., McGrann, G., De Vega, J. J.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.18.767592v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.18.767592v1.full.pdf
- PDF SHA-256: 2cb039bfd0a01a24acc05365b81135a7d4db96f0fbe29f9af6597ed411a187a3
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

**&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)**

**(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made**
**&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)**


1 **Evaluating novel sources of resistance to** _**Fusarium**_ _**oxysporum**_ **f. sp.** _**cubense**_
2 **tropical race 4 (TR4) in a global banana diversity panel**


3


4 Carolina Olave-Achury &#91;1&#93;, Daniel Ponndorf &#91;2&#93;, Angela Chaparro Garcia &#91;2&#93;, Anthony Hall &#91;1&#93;,
5 Graham McGrann &#91;2&#93;, Jose J De Vega &#91;1 &#93;


6


7 1 Earlham Institute, Norwich Bioscience Institutes (NBI), Norwich Research Park
8 (NRP), Norwich NR4 7UZ, UK.


9 2 Tropic Biosciences UK Ltd., Innovation Centre, Norwich Research Park (NRP),
10 Norwich NR4 7GJ, UK.


11 @ Correspondence to jose.devega@earlham.ac.uk


12


13 **Abstract**


14 _Fusarium oxysporum_ f. sp. _cubense_ tropical race 4 (Foc TR4) poses a major threat
15 to global banana production, yet quantitative variation in resistance across _Musa_
16 diversity remains poorly characterised. In this study, 21 accessions representing
17 many of the major cultivated _Musa_ groups and wild relatives were evaluated for their
18 response to Foc TR4 under controlled-environment conditions. Significant variation
19 in Foc TR4-associated corm discolouration was observed across the panel. Eight
20 accessions had significantly less corm discolouration than the susceptible control
21 Dwarf Cavendish: five were classified as resistant (all plants with more than 80%
22 healthy corm area): _Musa_ _itinerans_, _M. schizocarpa_, FHIA-02, FHIA-21 and the
23 plantain Curare, and three as tolerant, which response was significantly different to
24 the Dwarf Cavendish control but with some plants affected: _M. ornata_, Pelipita and
25 Mai’a Maoli Koa’e. All accessions with only A genomes were susceptible or very
26 susceptible, as were _M. textilis_ and the B-containing triploids Pome (AAB) and Silver
27 Bluggoe (ABB). Five of the seven accessions carrying at least one B genome were
28 resistant or tolerant, suggesting that B-genome backgrounds merit wider screening
29 as potential sources of resistance. These results identify valuable sources of
30 resistance to Foc TR4 and provide a basis for banana breeding and further
31 investigation of the genetic basis of resistance at the cultivar level.


32


33 Keywords: banana; Fusarium wilt; Foc TR4; germplasm screening; image analysis;
34 _Musa_ ; quantitative resistance


35 Working title: TR4 response across banana cultivars


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


36 **Introduction**


37


38 Bananas, members of the genus _Musa_, are large herbaceous plants cultivated
39 across tropical and subtropical regions of Southeast Asia, Africa, and the Americas.
40 They are among the most important crops worldwide, with more than 183 million
41 tonnes of bananas and plantains produced in 2023 (FAO 2025). The domestication
42 and selection of seedless fruit led to the fixation of parthenocarpy and sterility in
43 cultivated bananas. As a result, these crops are propagated mainly vegetatively. At
44 the same time, thousands of years of diversification among wild genotypes, together
45 with human selection of hybrids, have given rise to the wide range of edible bananas
46 and plantains available today. More than 7,000 _Musa_ accessions are preserved in
47 global germplasm collections curating genetic diversity derived from geographical
48 adaptation and local selection (Ruas _et al._ 2017). The largest _ex situ_ collection is
49 held at the International _Musa_ Germplasm Transit Centre (ITC) in Leuven, Belgium,
50 with 1,704 accessions. Tracing the origins of triploid cultivated hybrids remains
51 challenging because their ancestry is likely to be complex, involving multiple
52 hybridisation and backcrossing events (Martin _et al._ 2023; Higgins _et al._ 2024).


53


54 Fusarium wilt of banana (FWB) has long affected banana production and is caused
55 by different races of the soil-borne fungal pathogen _Fusarium oxysporum_ f. sp.
56 _cubense_ (Foc) (Pegg _et al._ 2019). The earliest symptoms of FWB are root and
57 rhizome vascular discolouration, but these cannot be detected in the field without
58 destructive sampling. External symptoms appear only after the pathogen has
59 colonised the rhizome. Typical symptoms include chlorosis, wilting, and skirting of
60 older leaves due to vascular blockage (Ploetz 2006). Leaf collapse at the petiole
61 base is then followed by plant death. Other symptoms include malformation of
62 younger leaves, splitting at the base of the pseudostem, and collapse of older leaves
63 while they are still green. In resistant lines of _Musa_ _acuminata_ subsp. _malaccensis_, a
64 GFP-tagged Foc race 4 strain showed restricted movement and no sporulation in the
65 rhizome (Chen _et al._ 2019), providing novel insights into how resistance against this
66 pathogen may operate in _Musa_ spp.


67


68 An epidemic of Foc race 1 in the Americas devastated production of Gros Michel
69 (AAA), the main export banana during the first half of the 20th century (Ploetz 2015).
70 The soil-borne nature of the disease left infested land unsuitable for Gros Michel

71 cultivation (Ploetz 2006; Pegg _et al._ 2019). Cavendish (AAA) cultivars became the
72 dominant export bananas in the 1960s and 1970s because of their resistance to Foc
73 race 1. As banana plantations expanded in Asia, Cavendish genotypes spread
74 worldwide. From the late 1960s to the 1990s, more aggressive forms of FWB
75 emerged in East Asia and Australia, associated with tropical and subtropical race 4


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


76 (TR4 and STR4). The distribution of Foc vegetative compatibility groups (VCGs) in
77 Southeast Asia, the centre of origin of _Musa_ spp., supports a co-evolution hypothesis
78 in which TR4 was present in native soils before Cavendish (AAA) plantations were
79 established; these plantations were later abandoned following the rapid spread of the
80 disease between 1990 and 1993 (Ploetz and Pegg 1997; Buddenhagen 2009;
81 Mostert _et al._ 2017). Foc TR4 has since spread beyond Asia and Oceania to the
82 Middle East and Africa (Maymon _et al._ 2020; van Westerhoven _et al._ 2023) and to
83 South America, where it has been reported in Colombia, Peru and Venezuela
84 (García-Bastidas _et al._ 2020; Acuña _et al._ 2022; Mejías Herrera _et al._ 2023) and,
85 most recently, in Ecuador in December 2025 (Pachacama _et al._ 2026). Under
86 tropical growing conditions, TR4 appears to be pathogenic to most AAA cultivars, in
87 addition to cultivars susceptible to races 1 and 2 (Ploetz 2015; Mostert _et al._ 2017;
88 Mintoff _et al._ 2021).


89


90 Genetic variation in response to FWB has been reported in wild banana germplasm,
91 some hybrids, and mutant lines in several phenotypic studies (Li _et al._ 2015; Zuo _et_
92 _al._ 2018; Chen _et al._ 2019; Mintoff _et al._ 2021; Thangavelu _et al._ 2025), reflecting the
93 increased attention prompted by the spread of the pathogen and the projected
94 losses associated with the disease over the last decade (Cook _et al._ 2015; Scheerer
95 _et al._ 2018). To date, several resistance or anti-apoptosis genes have been
96 introduced into elite clonal commercial backgrounds through transgenesis or gene
97 editing (Dale _et al._ 2017; Rocha _et al._ 2021; Harding _et al._ 2025). Resistance to Foc
98 appears to be genotype-specific; several wild and cultivated diploid accessions,
99 including Calcutta 4 (AA), Zebrina (AA), Pisang Lilin (AA), Pisang Jari Buaya (AA)
100 and Tuu Gia (AA), are frequently reported as sources of resistance to race 1 and
101 TR4 (Rocha _et al._ 2021). These TR4-resistant diploids have been used as parents in
102 breeding programmes to generate diploid or tetraploid hybrids. However, these
103 hybrids have not met all export market requirements and therefore have not replaced
104 Cavendish (Ploetz 2015; Thangavelu _et al._ 2020). The _Musa_ B genome has been

105 associated with abiotic stress tolerance and contains exclusive miRNA families

106 putatively related to stress tolerance (Davey _et al._ 2013); B subgenomes have also
107 been proposed as a rich source of disease-resistance genes (Xie _et al._ 2024). BB
108 germplasm has rarely been screened for response to TR4, and when tested the
109 responses of BB accessions have been inconsistent across studies (Zuo _et al._ 2018;
110 Thangavelu _et al._ 2025). Still, seven plantains (AAB) showed high resistance to TR4
111 in greenhouse and field experiments (Zuo _et al._ 2018). Additional cultivars carrying a
112 B genome and reported as resistant or tolerant to Foc TR4 include FHIA-21 (AAAB),
113 Plantain number 2 (AAB), Popoulu (AAB), Pisang Nangka (AAA-AAB) and Akpakpak
114 (AAB) (Zuo _et al._ 2018; Thangavelu _et al._ 2025).


115


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


116 Here, we aimed to quantify variation in response to Foc TR4 across a panel of
117 banana accessions selected to represent broad global _Musa_ diversity, including
118 major cultivated groups and wild relatives, under controlled-environment conditions.
119 By combining external symptom assessment with image-based quantification of
120 internal corm discolouration, we sought to identify resistant and tolerant germplasm
121 with greater resolution than categorical disease scoring allows. This study's novelty
122 lies in using a diversity panel structured from previous population structure analysis
123 (Higgins _et al._ 2024) to represent under-evaluated _Musa_ groups and in applying
124 quantitative corm phenotyping and mixed-model analysis to compare accessions
125 across experiments. Together, these approaches provide a clearer basis for
126 prioritising resistance sources for breeding and for investigating resistance at the
127 cultivar level rather than inferring it from genome composition alone.


128


129


130 **Materials and methods**


131


132 **Plant material**


133 Twenty-one banana accessions were used in this study (Table 1): 19 accessions
134 obtained as individual rooted plantlets from the International _Musa_ Germplasm
135 Transit Centre (ITC; Bioversity International, Leuven, Belgium) under a Standard
136 Material Transfer Agreement in 2022, and two Cavendish-subgroup lines used as
137 additional TR4 susceptible reference lines, Grande Naine ITC0180 and Tropic
138 Grande Naine, provided by Tropic Biosciences UK Ltd (Norwich, UK). We selected
139 the 19 ITC accessions, as detailed below, based on the population structure of the
140 banana collection at the Colombian genebank (AGROSAVIA) described by Higgins
141 et al. (2024).


142


143 Higgins et al. (2024) identified thirteen clonal clusters in a panel of 188 banana and
144 plantain accessions: nine cultivar groups named according to the passport data
145 stored in the _Musa_ Germplasm Information System (MGIS; Ruas _et al._ 2017), one
146 group of unknown AAA accessions lacking passport information, and three groups of
147 synthetic tetraploids from breeding programmes. That study also described wild
148 _Musa_ species and thirty-four accessions not assigned to any cluster. From this
149 population structure, we selected one representative from each of the thirteen
150 clusters, plus five wild species, and sourced them from ITC. An accession of the
151 Pome subgroup (AAB) was also included to clarify the differences between the
152 FHIA-02 accession classified as AAAB in Higgins et al. (2024) and the FHIA-02
153 accessions classified as AAAA in MGIS. When selecting a representative accession
154 for each group, we prioritised accessions duplicated in both the Colombian


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


155 genebank (AGROSAVIA; Higgins et al. 2024) and the ITC collections; if missing, we
156 selected another member of the same genetic group; or, failing that, an accession of
157 the same subgroup and ploidy. ITC codes, genome groups and subgroups follow the
158 ITC passport data supplied with the material (Table 1).


159


160 **Tissue culture and micropropagation**


161 Upon arrival from the ITC, plants were left in the vessels they were shipped in for a
162 recovery period. After acclimatisation, the plants were transferred to rooting medium
163 (RM). Individual rooted plants were later placed in tissue culture vessels containing
164 RM and kept in a growth cabinet at 25°C, 80% relative humidity, and a 16 h light/8 h
165 dark photoperiod until they were ready for subsequent refreshment,
166 micropropagation, or transfer to soil. RM included Murashige and Skoog salts (MS
167 salts), vitamins and a gelling agent. To increase the number of plants,
168 micropropagation was performed using propagation medium (PM; Fig. 1A–E). Plants
169 were cleaned of dead tissue and trimmed, and explants were divided and transferred
170 to culture jars containing PM. New propagation cycles can be started as soon as
171 two weeks or as late as three months after micropropagation, depending on the
172 plant’s growth rate and experimental needs.


173


174 **Soil establishment**


175 Plants were refreshed in vitro into new RM at least one week before transfer from

176 RM to soil to allow for root recovery. Only plants that had at least two roots longer
177 than 1 cm and three leaves were transferred. Plants were transferred to a sterile soil

178 mix (JIC barley mix: 60% peat, 20% grit/sand, and 20% perlite) in cell trays, then
179 kept in a controlled-environment cabinet under controlled temperature, photoperiod
180 and high relative humidity (Fig. 1F–K). Plants were irrigated by automatic flooding
181 and were covered during the initial acclimatisation period. After acclimatisation, the
182 plants were transplanted into pots with the same soil mix and grown for three weeks
183 before TR4 inoculation assays.


184 The composition of the tissue-culture media (RM and PM), together with specific
185 details of the associated micropropagation, culture and acclimatisation procedures,
186 is proprietary to Tropic Biosciences UK Ltd (Norwich, UK) and is therefore not

187 disclosed. All other materials and methods are described in full.


188


189 **Foc TR4 inoculum**


190 Inoculum was produced from Foc TR4 isolate UK0001, recovered from a
191 symptomatic plant at the Eden Project (Cornwall, UK) and sequenced by
192 Warmington et al. (2019). Genomic analyses place this isolate close to the


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


193 Indonesian TR4 isolate II-5 at the sequence level (Warmington _et al._ 2019). For
194 inoculum production, 20 µL of a −80°C stock of isolate UK0001 was plated onto
195 potato dextrose agar (PDA) and incubated for 7 days at 25°C in the dark. A 5 mm
196 plug was transferred onto a new PDA plate and incubated for a further 7 days. Three
197 plugs from this second plate were added to 120 mL of potato dextrose broth (PDB)
198 and incubated in a water-bath shaker at 25°C and 180 rpm for 5 days. The liquid
199 culture was filtered through sterile cheesecloth to remove mycelial fragments. The
200 filtrate was divided evenly between three 50 mL Falcon tubes and centrifuged for 10
201 minutes at 3,000 rpm at 20°C. The pellet in each tube was rinsed twice in 40 mL of
202 distilled water and centrifuged for 5 minutes at 3,000 rpm. After the second wash, the
203 pellet in the first tube was resuspended in 15 mL of distilled water and transferred to
204 the next 50 mL tube, repeating the resuspension until all inoculum was combined. A
205 0.5 mL aliquot from each tube was taken for quality control, along with the mock PDB
206 control. Finally, all the TR4 inoculum was pooled, and a 0.5 mL aliquot was taken for
207 spore counting, using a Neubauer chamber. Inoculum was prepared at 2 × 10 �
208 spores/mL, with 200 mL applied per plant.


209


210 The control inoculum for the mock-inoculations was prepared in the same way, using
211 TR4-free PDB incubated alongside the inoculum preparations.


212


213 **Experimental design and inoculation**


214 Inoculation assays were conducted in a controlled-environment cabinet as six
215 independent experiments between 2023 and 2025, numbered 1 and 3–7; a further
216 experiment (experiment 2) was lost to a cabinet failure and no data from it were
217 used. Each experiment was treated as a block, and the four internal quadrants of the
218 cabinet (A–D) were treated as replicates. Because the number of plants available
219 per accession was limited by micropropagation, each experiment included a subset
220 of the accessions, giving an incomplete block design in which each accession other
221 than Dwarf Cavendish was tested in two to four experiments. Within each
222 experiment, TR4-inoculated plants of each accession were distributed across the
223 quadrants alongside mock-inoculated plants of the same accessions. Cultivar Dwarf
224 Cavendish was included as the susceptible control in every experiment and
225 quadrant, to allow responses to be compared among experiments and quadrants. In
226 total, 180 plants were inoculated with Foc TR4 and 65 plants were mock-inoculated;
227 the 47 mock-inoculated plants with a recorded accession are listed in Table 1. All
228 accessions were tested in a minimum of two independent inoculation experiments.


229


230 Five days before inoculation, plants were moved onto individual trays and placed in
231 the cabinet, and water was withheld before inoculation. On the day of inoculation, the
232 number of open leaves and, in some experiments, plant height were recorded (Table


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


233 S1). Control plants were mock-inoculated first, followed by TR4-inoculated plants, to
234 minimise the risk of cross-contamination. Inoculation involved gently pouring 200 mL
235 of mock solution or spore suspension (2 × 10 &#91;5&#93; spores/mL) around the base of the
236 pseudostem to maximise contact with the roots (García-Bastidas _et al._ 2019). From
237 one week after inoculation, plants were watered from below with 200 mL of water per
238 tray twice weekly, until the end of the assay or until plant death.


239


240 **External and internal scoring**


241 We scored external symptoms weekly from inoculation (week 0) to six weeks after
242 inoculation (wpi) by counting the total number of open leaves and the number of
243 yellow (wilting) leaves.  Leaves were recorded as yellow when more than 50% of
244 their area was discoloured. Plant height was also recorded at inoculation (week 0)
245 and at 6 wpi, and pseudostem splitting was noted, in a subset of experiments (Table
246 S1). At 6 wpi, each plant was photographed before internal symptoms were
247 assessed. For this, the pseudostem was cut 10 cm above the corm, the roots were
248 removed and the soil was washed away with water; the remaining stem was then cut
249 longitudinally to expose the inner corm tissue, and both halves were photographed
250 with a 5 cm label for scale and colour reference. Camera, lighting, and focal distance
251 were standardised across experiments. Using ImageJ (Schneider _et al._ 2012), the
252 total corm area was manually outlined and measured, and the healthy (non253 discoloured) area was then delimited using the colour contrast between healthy and
254 discoloured tissue and measured (Table S2). The percentage of healthy corm area
255 (100 × healthy area/total corm area) was calculated for each of the two halves of
256 each corm, and completely discoloured corms were recorded as 0% healthy area.
257 Corms of mock-inoculated plants were sectioned and photographed in the same
258 way.


259


260 **Data analysis**


261 All analyses were performed in R (R Core Team 2024), and figures were drawn with
262 ggplot2 (Wickham 2016). For external symptoms, TR4-inoculated plants were
263 grouped by the response class of their accession (R, T, S or VS; see below), and the
264 total number of leaves and the percentage of yellow leaves were each analysed with
265 a single linear mixed model fitted to all weekly observations (weeks 0–6), with group
266 (the four TR4 groups and mock-inoculated plants), week and their interaction as
267 fixed effects and block (experiment × quadrant), accession and plant as random
268 effects. Plants that died during experiments were assigned their last recorded
269 number of leaves and 100% yellow leaves for the remaining weeks. Within each
270 week, each TR4 group was compared with the mock-inoculated plants using
271 Dunnett’s test (Table S3). Height gain from inoculation to 6 wpi was analysed with a
272 linear mixed model including group as a fixed effect and block and accession as


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


273 random effects. The proportion of plants with pseudostem splitting is shown with
274 Wilson 95% confidence intervals (Fig. S1).


275


276 For the percentage of healthy corm area, data from TR4-inoculated plants were
277 analysed as an incomplete block design using best linear unbiased estimates
278 (BLUEs) from a linear mixed model fitted with lme4 (Bates _et al._ 2015) and lmerTest
279 (Kuznetsova _et al._ 2017), in which accessions (treatments) were fitted as fixed
280 effects and replicates (quadrants) and blocks (experiments) within replicates as
281 random effects: � ��� �� �� � �� � �� �� �� ���, where, � ��� is the percentage of

282 healthy corm area measured on a corm half of the � &#91;��&#93; accession in the � &#91;��&#93; replicate
283 (quadrant) within the � &#91;��&#93; block (experiment), � is the overall mean, � � is the fixed
284 effect of the accession, � � is the random effect of the replicate (cabinet quadrant), � ��
285 is the random effect of the block (experiment) within the replicate, and � ��� is the
286 residual error. The null hypothesis was that all accession effects were equal �� � : � � �
287 � � ���� � �, while the alternative hypothesis was that at least one accession effect
288 differed. The model was fitted in R as pct_healthy ~ accession + (1 | quadrant) + (1 |
289 quadrant:experiment). Model assumptions were checked using residual plots for
290 normality and homogeneity of variance. BLUEs and contrasts were obtained with
291 emmeans (Lenth 2024), using the Kenward–Roger approximation for the degrees of
292 freedom (Halekoh and Højsgaard 2014) (Table S4). Each accession was compared
293 with Dwarf Cavendish, the susceptible control included in every experiment, using
294 Dunnett’s test (Table S5), and all pairs of accessions were compared using Tukey’s
295 test (Table S6).


296 Accessions were assigned to response classes from the corm data. Accessions
297 whose BLUE was significantly higher than that of Dwarf Cavendish (Dunnett’s test, _P_
298 &lt; 0.05) were classified as resistant (R) when all their corm halves had more than
299 80% healthy area, and as tolerant (T) otherwise. Accessions whose BLUE was
300 significantly lower than that of Dwarf Cavendish were classified as very susceptible
301 (VS), and all remaining accessions as susceptible (S).


302


303 **Results**


304 Twenty-one accessions (Table 1) were compared across six TR4 inoculation
305 experiments conducted from 2023 to 2025. Poor propagation resulted in fewer plants
306 being tested for four accessions: _M. acuminata_ subsp. _malaccensis_ (two plants in
307 two experiments), _M. itinerans_ (three plants in two experiments), Curare (Plantain
308 subgroup, AAB; three plants in two experiments) and Pelipita (ABB; six plants in
309 three experiments). Missing replicates in blocks were accounted for in the mixed
310 model used for data analysis. Of the 65 mock-inoculated plants, 47 were included in
311 the analysis of external symptoms; the remaining 18 were excluded because the


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


312 plant died for reasons unrelated to the inoculation or their accession was not

313 recorded.


314


315 **Internal scoring of TR4 symptoms**


316 Longitudinal sections of the corm at 6 wpi revealed contrasting internal symptoms
317 among accessions, ranging from no discolouration in accessions such as _M._
318 _schizocarpa_ to extensive discolouration in accessions such as Figue Sucrée (Fig.
319 2A). The percentage of healthy corm area was quantified for both halves of each
320 corm. Mock-inoculated corms showed no discolouration (Fig. S2). All corm halves of
321 FHIA-02, FHIA-21, _M. itinerans_, _M. schizocarpa_ and Curare had more than 80%
322 healthy area, whereas more than half of the corm halves of Pome, Figue Sucrée,
323 Grande Naine ITC0180 and FHIA-17 had 5% or less healthy area (Fig. S3).


324


325 We estimated BLUEs of the percentage of healthy corm area with experiments
326 (blocks) and cabinet quadrants (replicates) treated as random effects, to account for
327 variation due to the experimental design (Table S4). Healthy corm area in the
328 susceptible control varied among experiments, with a mean percentage of healthy
329 corm area per experiment ranging from 15% to 59% for Dwarf Cavendish (Fig. 3A;
330 Fig. S4), supporting the use of a data analysis model that accounts for experiment
331 effects. The residuals of the mixed model were approximately normally distributed
332 (Fig. S5A) and evenly distributed above and below zero across fitted values (Fig.
333 S5B); the deviations in the tails and the diamond-shaped pattern near 0% and 100%
334 reflect the bounded nature of the response variable.


335


336 The differences between the BLUE of each accession and that of Dwarf Cavendish,
337 the control, with 95% confidence intervals, are shown in Fig. 3B. Accessions with
338 positive values were more resistant to TR4 than the control, and those with negative
339 values were more susceptible. The BLUE of Dwarf Cavendish was 34.8% healthy
340 corm area. Eight accessions had significantly higher percentages of healthy corm
341 area than Dwarf Cavendish (Dunnett’s test, _P_ &lt; 0.001; Fig. 3B; Table S5), with
342 differences of 35 to 69 percentage points, and Tukey pairwise comparisons did not
343 show significant differences among these eight accessions (Table S6). Five of them,
344 FHIA-02, FHIA-21, _M. itinerans_, _M. schizocarpa_ and Curare, were classified as
345 resistant (R), because all their corm halves had more than 80% healthy area (Fig.
346 3C). The other three, _M. ornata_, Pelipita and Mai’a Maoli Koa’e, were classified as
347 tolerant (T), because some of their plants had less than 80% healthy corm area
348 (minimum plant means of 51–67%). Two accessions, Grande Naine ITC0180 (AAA)
349 and Pome (AAB), had significantly lower percentages of healthy corm area than
350 Dwarf Cavendish ( _P_ &lt; 0.05 and _P_ &lt; 0.01, respectively; Fig. 3B, C) and were
351 classified as very susceptible (VS); 60% and 75% of their corm halves, respectively,


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


352 had 5% or less healthy area (Fig. S3). The remaining 11 accessions did not differ
353 significantly from Dwarf Cavendish and were classified as susceptible (S).


354 The response class of each accession is summarised in Table 1. The resistant and
355 tolerant accessions comprised three wild species with neither A nor B genomes ( _M._
356 _itinerans_, _M. schizocarpa_ and _M. ornata_ ) and five accessions with at least one B
357 genome (FHIA-02 and FHIA-21, AAAB; Curare and Mai’a Maoli Koa’e, AAB; Pelipita,
358 ABB). All accessions with only A genomes (AA, AAA and AAAA), and the IITA hybrid
359 2145 assigned to an AAA cluster, were susceptible or very susceptible.


360


361 **External scoring of TR4 symptoms**


362 External symptoms were scored weekly for six weeks as the number of open leaves
363 and the percentage of yellow leaves (Fig. 4). For this analysis, TR4-inoculated plants
364 were grouped by the response class of their accession, determined from the internal
365 symptoms (see above): resistant (R; 28 plants of five accessions), tolerant (T; 23
366 plants of three accessions), susceptible (S; 111 plants of 11 accessions) and very
367 susceptible (VS; 18 plants of two accessions). Resistant and tolerant accessions had
368 significantly less corm discolouration than the susceptible control Dwarf Cavendish,
369 and very susceptible accessions significantly more (see Internal scoring of TR4
370 symptoms). These were compared with leaf development and health of mock371 inoculated plants. The total number of leaves increased steadily over the six weeks
372 and was similar in all TR4-response groups and in mock plants, except plants of VS
373 accessions which had significantly more leaves following TR4-inoculation than mock374 inoculated plants at week 2 ( _P_ = 0.03; Fig. 4A; Table S3). In contrast, plants of S and
375 VS accessions had a significantly higher percentage of yellow leaves than mock
376 plants from week 4 onwards ( _P_ &lt; 0.001; Fig. 4B), reaching on average about 55%
377 and 80% yellow leaves, respectively, at week 6, compared with about 30% in mock
378 plants. Plants of R and T accessions did not differ significantly from mock plants at
379 any week (Fig. 4B). Leaf yellowing and wilting are typically observed in banana
380 plants following TR4 infection because the fungus colonises the xylem and disrupts
381 the normal growth of the plant. Dwarf Cavendish (AAA) showed an atypical
382 progression of symptoms following TR4-inoculation, with the number of new leaves
383 comparable to or higher than in mock-inoculated plants. However, these new leaves
384 emerged more quickly and were stunted and deformed, with a small surface area.
385 New leaves emerged in rapid succession from the same position on the
386 pseudostem, which did not elongate between successive leaves, so the leaves
387 remained clustered at the top of a short pseudostem. In plants of resistant
388 accessions, the rate of leaf emergence was similar between TR4- and mock389 inoculated plants, and yellow leaves, when present, were attributable to senescence,
390 as they were observed at a similar rate  in both treatments.


391 Height gain from inoculation to 6 wpi did not differ significantly between any TR4392 response group and mock-inoculated plants ( _P_ ≥ 0.24), although plants of S and VS


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


393 accessions gained on average about 13 and 11 cm, compared with about 16 cm in
394 mock plants (Fig. S1A). Pseudostem splitting was rare (4.3% of evaluated TR4395 inoculated plants) and was not observed in mock-inoculated plants or in plants of R
396 accessions (Fig. S1B).


397


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


398 **Discussion**


399 This study set out to examine the response to Foc TR4 across a genetically diverse
400 panel of Musa spp., and identified new sources of resistance within it. The
401 accessions assessed represented, to a large extent, the primary and tertiary
402 diversification of _Musa_ spp., including some of the most common and widely grown
403 cultivars worldwide: nine cultivar subgroups recorded in MGIS plus Pelipita, three
404 synthetic tetraploids and an IITA hybrid, together with five wild species. We obtained
405 the accessions representing each cluster from ITC, whereas the full diversity panel
406 had previously been sequenced from the AGROSAVIA collection in Colombia
407 (Higgins _et al._ 2024).


408


409 Four of the five major genomic groups were represented: AA, AAA, AAB and ABB,
410 with only the AB hybrids absent. No wild _M. balbisiana_ (BB) accession was included.
411 AB interspecific diploids are under-represented in germplasm collections because
412 they arose through secondary diversification in regions where the natural
413 distributions of the wild progenitor species overlap, mainly in South Asia (Fuller and
414 Madella 2009). Nevertheless, the absence of interspecific diploids does not imply
415 that their genetic background is absent from the panel. Analysis of AB homologous
416 exchanges supported contributions from AB diploids to the Bluggoe (ABB) and
417 Pelipita (ABB) clusters, both of which were included (Cenci _et al._ 2021). Including _M._
418 _schizocarpa_ (S genome donor) was also important, as this species contributed to
419 many cultivar groups, though to varying extents, as supported by chromosome
420 painting and Relative Averaged Alignment studies (Martin _et al._ 2023; Higgins _et al._
421 2024). The presence of S ancestry across a range of genome compositions provides
422 evidence of early recombination between _M. acuminata_ and _M. schizocarpa_ before
423 the establishment of cultivated AAA, AAB, and ABB cultivars (Heslop-Harrison and
424 Schwarzacher 2007; Li _et al._ 2013).


425


426


427


428 All accessions with only A genomes (AA, AAA and AAAA) were susceptible or very
429 susceptible to TR4, including the well-known susceptible cultivars Dwarf Cavendish
430 (AAA) and Gros Michel (AAA). In this study, the _M. acuminata_ subsp. _malaccensis_
431 (AA) accession tested appeared susceptible (two plants), although other genotypes
432 of this wild sexual species have been reported as resistant to Foc TR4 (Dale _et al._
433 2017; Chen _et al._ 2019). Natural variation in TR4 responses is expected in sexually
434 reproducing taxa where recombination occurs. Resistance to TR4 has already been
435 experimentally validated within the A genome: the _RGA2_ gene, isolated from a
436 resistant _M. acuminata_ subsp. _malaccensis_ genotype, conferred resistance when
437 introduced into a Cavendish background, and the resulting transgenic line has


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


438 recently been approved for commercial production and consumption in Australia
439 (Dale _et al._ 2017; Harding _et al._ 2025). Similarly, the _M. textilis_ accession tested here
440 was susceptible, although its response was bimodal, with five of 13 plants having
441 more than 80% healthy corm area and three having none. Natural variation in
442 response to TR4 might therefore be expected in this species, which originates from
443 the Philippines, where TR4 is present (Mostert _et al._ 2017); to our knowledge, no
444 resistant genotype of this species has yet been reported. Other susceptible
445 accessions included Silver Bluggoe (ABB), IITA hybrid 2145 and Pome (AAB). FHIA446 02 (AAAB) represented a cluster labelled ‘Pome AAAB’ by Higgins et al. (2024),
447 because one of its members was recorded as Pome. Notably, FHIA-02 was
448 resistant, whereas the Pome (AAB) accession tested here was among the most
449 susceptible accessions in our study (Fig. 3).


450


451 The presented results suggest that the B genome may represent an additional
452 source of resistance to TR4. Five of the eight resistant or tolerant accessions
453 contained at least one B genome, and five of the seven B-containing accessions
454 tested were resistant or tolerant, compared with none of the accessions with only A
455 genomes. The association is therefore one of frequency rather than sufficiency:
456 Pome (AAB) and Silver Bluggoe (ABB) carried a B genome yet were among the
457 most susceptible accessions in the panel, and three of the eight resistant or tolerant
458 accessions carried neither an A nor a B genome. This interpretation has limitations.
459 No _M. balbisiana_ (BB) or AB accession was tested. Three of the eight resistant or
460 tolerant accessions were wild species with neither A nor B genomes, and the two
461 resistant FHIA tetraploids derive from breeding programmes that used improved
462 diploid parents selected for disease resistance, so part of their resistance may derive
463 from A-genome donors. Moreover, _M. balbisiana_ (BB) and other wild diploids have
464 not been studied as extensively as A-genome accessions. One wild _M. balbisiana_
465 accession was highly susceptible to TR4, in contrast to five resistant wild _Musa_ spp.,
466 after inoculation with a locally sourced TR4 strain in China (Li _et al._ 2015). Two
467 independent studies tested the accession Tani BB (ITC1120) against TR4 under
468 glasshouse and field conditions, using the same inoculation and scoring
469 methodology. Zuo et al. (2018) initially reported Tani BB as resistant in the
470 glasshouse and immune in the field. However, the same accession was later
471 reported by Thangavelu et al. (2025) as moderately resistant in both glasshouse and
472 field conditions. Given these differing outcomes, we noted that the Foc TR4 isolate
473 and inoculum concentration used in the glasshouse inoculations differed between
474 the two studies. Moreover, the field experiments did not report the Foc TR4 isolate or
475 concentration, as the study used soils previously contaminated with TR4. It is also
476 unclear how disease scores were translated into resistance categories, and the raw
477 scoring data were not made available (Thangavelu _et al._ 2025). Differences in
478 response to TR4 reported even for the same clone across studies highlight the
479 challenge of reproducibility in inoculation assays. Zuo et al. (2018) reported Pisang


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


480 Ceylan AAB (ITC1441) as resistant, with no infection detected in the rhizome of field481 grown plants after two years and Foc TR4 confirmed by PCR in infected tissues.
482 However, the same accession (ITC1441) was later reported by Zhan et al. (2022) as
483 showing intermediate resistance. Both studies evaluated the same clone and the
484 same Foc TR4 strain in the glasshouse experiments, and the plants were grown in
485 the same field in China. Furthermore, the other three accessions shared between the
486 two studies showed similar responses to Foc TR4 (Zuo _et al._ 2018; Zhan _et al._
487 2022).


488


489 We observed variation in the TR4 response of accessions across inoculation
490 experiments under our bioassay conditions. The effects of the experimental design
491 were accounted for by estimating BLUEs from a model that treated accession effects
492 as fixed while accounting for the random effects of the incomplete experimental
493 blocks and of the quadrants within the cabinet. This was important because the
494 response of the susceptible control varied considerably among experiments (Fig.
495 3A). Including Dwarf Cavendish in every experiment (22 plants analysed in total)
496 provided a common reference against which all accessions could be contrasted,
497 even though each of the other accessions was tested in only two to four
498 experiments. The assays used young plants under controlled conditions, with an
499 endpoint of 6 wpi, so slower-developing infections may have been under-scored, and
500 responses in the field may differ. In addition, we classified _M. itinerans_ and Curare as
501 resistant based on three plants each, although the same _M. itinerans_ accession
502 (ITC1526) has also been rated highly resistant elsewhere (Zuo _et al._ 2018; Rouard _et_
503 _al._ 2026).


504


505 Several of our results agree with previous screenings of the same or related
506 accessions. The _M. itinerans_ accession tested here (ITC1526) has been rated highly
507 resistant to TR4 (Zuo _et al._ 2018), and FHIA-02 has been reported as resistant to
508 TR4 in glasshouse and field studies (Chen _et al._ 2019; Mintoff _et al._ 2021); both ITC
509 accessions are included in a recently curated subset of TR4-resistant germplasm
510 (Rouard _et al._ 2026). Most plantains show high resistance to TR4 (Zuo _et al._ 2018;
511 Zhan _et al._ 2022), consistent with Curare's resistance. In contrast, accessions of the
512 same subgroup have sometimes given contrasting results: two accessions of the
513 Maia Maoli/Popoulu subgroup differed in their response to TR4 (Zhan _et al._ 2022),
514 Pelipita accessions have been rated from highly resistant to susceptible (Rouard _et_
515 _al._ 2026), and Pome-subgroup accessions were resistant in field and glasshouse
516 trials in China (Zhan _et al._ 2022), whereas the Pome accession tested here
517 (ITC0076) was very susceptible. Such variation within subgroups, together with the
518 variable responses of Mai’a Maoli Koa’e and Pelipita in our study, supports
519 investigating resistance at the cultivar level.


520


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


521 Further studies of AAB/ABB cultivars and AB hybrids will be needed to clarify the role
522 of the B genome in TR4 resistance. The source of resistance should be investigated
523 at the cultivar level rather than at the genomic-group level, as variation within groups
524 has already been observed. Variation in response to TR4 at the cultivar level is likely
525 associated with specific combinations of genes and regulatory elements involved in
526 resistance, such as NLR immune receptors, of which RGA2 is one example (Dale et
527 al. 2017). Characterisation of NLRs across the _Musa_ genus should therefore be a
528 priority for future research. There is also a need for a robust and reproducible
529 scoring system for TR4 bioassays that is less prone to subjective assessment and to
530 the resulting loss of statistical power. In this regard, a continuous measure of corm
531 discolouration, such as the image-based percentage of healthy corm area used here,
532 is preferable to categorical scales. Screening that scores infection on a discrete
533 scale rather than as a quantitative trait reduces a continuous response to broad
534 categories, which discards information and reduces statistical power. Categorical
535 scales may also saturate: the corms of Dwarf Cavendish plants in our study ranged
536 from 0% to 66% healthy area, a range that would fall within the two most severe
537 classes of a typical six-class rhizome discolouration scale. Categorical scoring also
538 makes comparison across studies more difficult, particularly when the raw scores are
539 not made available.


540


541 **Conclusions**


542 The quantitative phenotyping used here revealed substantial variation in response to
543 Foc TR4 across a panel designed to capture understudied _Musa_ diversity. Five
544 accessions were classified as resistant and three as tolerant, whereas all accessions
545 carrying only A genomes were susceptible or very susceptible in this panel, including
546 well-established susceptible controls. Among them is the M. acuminata subsp.
547 malaccensis accession tested, which was represented by only two plants and for
548 which resistant genotypes have been reported elsewhere; the result therefore
549 reflects the specific accessions assessed rather than the A genome as a whole.
550 Although resistance cannot be assigned to a single genomic group, the frequency of
551 resistant or tolerant responses among accessions carrying at least one B genome
552 suggests that this genomic background may represent an underexplored source of
553 resistance against TR4; two B-containing accessions were nonetheless highly
554 susceptible, so a B genome is not by itself sufficient for resistance. In addition,
555 continuous corm discolouration measurements allowed accessions to be compared
556 across experiments on a common quantitative scale and avoided the limitations
557 associated with categorical disease classes. These findings provide a basis for
558 prioritising germplasm in resistance breeding and for future studies aimed at
559 identifying the genes and regulatory mechanisms underlying TR4 resistance.


560


561 **Data Availability**


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


562 The data supporting the findings of this study, including the weekly external symptom
563 scores per plant, corm measurements and model estimates, are provided as
564 Supporting Information (Tables S1–S6). The R scripts used for the analyses and
565 figures will be deposited in a GitHub public repository
566 (https://github.com/DeVegaGroup/TR4_response_screening).


567


568 **Acknowledgements**


569 We thank Leo Wilkinson for assistance with the first inoculation experiment, and
570 Finlay Taylor, Sophie Thomas, and the rest of the pathology and media teams at
571 Tropic Biosciences UK Ltd. We also acknowledge the International Musa Germplasm
572 Transit Centre (Bioversity International, Leuven, Belgium) for providing the
573 germplasm. The authors used Generative AI (Claude, Anthropic) to help check the
574 curated data, reproduce the statistical analyses, and edit the manuscript text. The
575 authors reviewed and verified all outputs and take full responsibility for the content.


576


577 **Sources of funding**


578 COA is supported by the Biotechnology and Biological Sciences Research Council
579 (UKRI-BBSRC) through the Norwich Research Park Doctoral Training Programme
580 (BB/T008369/1), with Tropic Biosciences UK Ltd as an industrial CASE partner. This
581 study was also partially funded by the Biotechnology and Biological Sciences
582 Research Council (BBSRC), part of UKRI, via Earlham Institute’s Strategic
583 Programme Grant “Decoding Biodiversity” (BBX011089/1) and its constituent work
584 package BBS/E/ER/230002B (Decode WP2 Genome Enabled Analysis of Diversity
585 to Identify Gene Function, Biosynthetic Pathways, and Variation in Agri/Aquacultural
586 Traits).


587


588 **Conflict of interest**


589 DP, AC and GM are employed by Tropic Biosciences UK Ltd, the industrial partner of
590 the studentship. The other authors declare no conflict of interest.


591


592 **Supporting Information**


593 **Figure S1.** **Foc TR4 did not significantly reduce height gain, and pseudostem**
594 **splitting was rare.** Plant height gain and pseudostem splitting after inoculation with
595 Foc TR4. (A) Adjusted mean height gain from inoculation (week 0) to 6 weeks after
596 inoculation, with 95% confidence intervals, for mock-inoculated plants and TR4597 inoculated plants grouped by the response class of their accession (R, resistant; T,
598 tolerant; S, susceptible; VS, very susceptible), from a linear mixed model with block


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


599 and accession as random effects; n, number of plants. (B) Percentage of plants with
600 pseudostem splitting noted, with Wilson 95% confidence intervals; numbers are
601 plants with splitting/plants scored.


602 **Figure S2.** **Corm discolouration occurs only in TR4-inoculated plants.**
603 Photographic summary of external and internal symptoms at 6 weeks after
604 inoculation. (A) Representative TR4- and mock-inoculated plants and longitudinal
605 corm sections of a resistant and a susceptible accession; scale bars, 5 cm. (B–D)
606 Corm halves of one TR4-inoculated plant (upper two rows) and one mock-inoculated
607 plant (lower two rows) per accession (one accession per column) for (B) resistant
608 accessions (green frame), (C) tolerant accessions (purple frame) and (D) susceptible
609 and very susceptible accessions (red frame). Yellow outlines show the healthy corm
610 areas delimited in ImageJ.


611 **Figure S3. All corm halves of resistant accessions stay above 80% healthy**
612 **area.** Distribution of classes of healthy corm area scored for each accession tested.
613 Bars show, for each accession, the proportion of corm halves of TR4-inoculated
614 plants in each class of healthy corm area at 6 weeks after inoculation; n, number of
615 plants. Accessions are ordered as in Fig. 3C, and the strip above the bars shows the
616 response class (R, resistant; T, tolerant; S, susceptible; VS, very susceptible).


617 **Figure S4. Healthy corm area of the Dwarf Cavendish control varies between**
618 **experiments.** Variability of the Dwarf Cavendish control. (A) Percentage of healthy
619 corm area of the susceptible control Dwarf Cavendish in each experiment after
620 removal of one outlier; symbols indicate the cabinet quadrant (A–D). (B) Difference
621 between the BLUE of each accession and that of Dwarf Cavendish, with 95%
622 confidence intervals, coloured by response class; filled symbols, _P_ &lt; 0.05 (Dunnett’s
623 test).


624 **Figure S5. Residuals of the corm mixed model are approximately normal.**
625 Mixed-model diagnostics of TR4 inoculation scoring data. (A) Normal quantile–
626 quantile plot of the residuals of the mixed model for the percentage of healthy corm
627 area. (B) Residuals against fitted values; the diamond-shaped pattern near 0% and
628 100% reflects the bounded response variable.


629 **Table S1.** Weekly external symptom scores per plant (total and yellow leaves, plant
630 height and pseudostem splitting).


631 **Table S2.** Corm area measurements for each corm half at 6 weeks after inoculation.


632 **Table S3.** External symptoms: adjusted means per week for each TR4-response
633 group and mock-inoculated plants, and Dunnett’s comparisons with mock-inoculated
634 plants.


635 **Table S4.** Best linear unbiased estimates (BLUEs) of the percentage of healthy corm

636 area for each accession.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


637 **Table S5.** Comparison of each accession with the susceptible control Dwarf
638 Cavendish (Dunnett’s test).


639 **Table S6.** Pairwise comparisons between accessions (Tukey’s test).


640


641


642 **References**


643 **Acuña R, Rouard M, Leiva AM,** _**et al.**_ **2022** . First report of _Fusarium oxysporum_ f.
644 sp. _cubense_ Tropical Race 4, causing Fusarium wilt in Cavendish bananas in Peru.
645 _Plant Disease_ **106** : 2268.


646 **Bates D, Mächler M, Bolker B, Walker S** . **2015** . Fitting linear mixed-effects models
647 using lme4. _Journal of Statistical Software_ **67** : 1–48.


648 **Buddenhagen I** . **2009** . UNDERSTANDING STRAIN DIVERSITY IN FUSARIUM
649 OXYSPORUM F. SP. CUBENSE AND HISTORY OF INTRODUCTION OF

650 “TROPICAL RACE 4” TO BETTER MANAGE BANANA PRODUCTION. _Acta_
651 _Horticulturae_ **828** : 193–204.


652 **Cenci A, Sardos J, Hueber Y,** _**et al.**_ **2021** . Unravelling the complex story of
653 intergenomic recombination in ABB allotriploid bananas. _Annals of Botany_ **127** : 7–20.


654 **Chen A, Sun J, Matthews A,** _**et al.**_ **2019** . Assessing Variations in Host Resistance to
655 Fusarium oxysporum f sp. cubense Race 4 in Musa Species, With a Focus on the
656 Subtropical Race 4. _Frontiers in Microbiology_ **10** : 1062.


657 **Cook DC, Taylor AS, Meldrum RA, Drenth A** . **2015** . Potential Economic Impact of
658 Panama Disease (Tropical Race 4) on the Australian Banana Industry. _Journal of_
659 _Plant Diseases and Protection_ **122** : 229–237.


660 **Dale J, James A, Paul J-Y,** _**et al.**_ **2017** . Transgenic Cavendish bananas with
661 resistance to Fusarium wilt tropical race 4. _Nature Communications_ **8** : 1496.


662 **Davey MW, Gudimella R, Harikrishna JA, Sin LW, Khalid N, Keulemans J** . **2013** .
663 “A draft Musa balbisiana genome sequence for molecular genetics in polyploid, inter664 and intra-specific Musa hybrids.” _BMC Genomics_ **14** : 683.


665 **FAO** . **2025** . FAOSTAT: Crops and livestock products. Rome.
666 https://www.fao.org/faostat/en/#data/QCL (Accessed 01/06/2026).


667 **Fuller DQ, Madella M** . **2009** . Banana Cultivation in South Asia and East Asia: A
668 review of the evidence from archaeology and linguistics. _Ethnobotany Research and_
669 _Applications_ **7** : 333–351.


670 **García-Bastidas FA, Quintero-Vargas JC, Ayala-Vasquez M,** _**et al.**_ **2020** . First
671 Report of Fusarium Wilt Tropical Race 4 in Cavendish Bananas Caused by Fusarium
672 odoratissimum in Colombia. _Plant Disease_ **104** : 994.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


673 **García-Bastidas FA, Van Der Veen AJT, Nakasato-Tagami G, Meijer HJG,**
674 **Arango-Isaza RE, Kema GHJ** . **2019** . An Improved Phenotyping Protocol for
675 Panama Disease in Banana. _Frontiers in Plant Science_ **10** : 1006.


676 **Halekoh U, Højsgaard S** . **2014** . A Kenward-Roger approximation and parametric
677 bootstrap methods for tests in linear mixed models – the R package pbkrtest. _Journal_
678 _of Statistical Software_ **59** : 1–32.


679 **Harding R, Paul J, James A,** _**et al.**_ **2025** . QCAV ‐ 4, the first genetically modified
680 Cavendish (cv. Grand Nain) banana resistant to Fusarium wilt tropical race 4
681 approved for commercial production and consumption. _Plant Biotechnology Journal_
682 **23** : 3628–3637.


683 **Heslop-Harrison JS, Schwarzacher T** . **2007** . Domestication, Genomics and the
684 Future for Banana. _Annals of Botany_ **100** : 1073–1084.


685 **Higgins JA, Osorio-Guarín JA, Olave-Achury AC,** _**et al.**_ **2024** . Characterizing
686 subgenome recombination and chromosomal imbalances in banana varietal
687 lineages. _Annals of Botany_ **133** : 349–364.


688 **Kuznetsova A, Brockhoff PB, Christensen RHB** . **2017** . lmerTest package: tests in
689 linear mixed effects models. _Journal of Statistical Software_ **82** : 1–26.


690 **Lenth RV** . **2024** . emmeans: Estimated marginal means, aka least-squares means. R
691 package version.


692 **Li WM, Dita M, Wu W, Hu GB, Xie JH, Ge XJ** . **2015** . Resistance sources to
693 _Fusarium oxysporum_ f. sp. _cubense_ tropical race 4 in banana wild relatives. _Plant_
694 _Pathology_ **64** : 1061–1067.


695 **Li L-F, Wang H-Y, Zhang C,** _**et al.**_ **2013** . Origins and Domestication of Cultivated
696 Banana Inferred from Chloroplast and Nuclear Genes. _PLoS ONE_ **8** : e80502.


697 **Martin G, Cottin A, Baurens F,** _**et al.**_ **2023** . Interspecific introgression patterns
698 reveal the origins of worldwide cultivated bananas in New Guinea. _The Plant Journal_
699 **113** : 802–818.


700 **Maymon M, Sela N, Shpatz U, Galpaz N, Freeman S** . **2020** . The origin and current
701 situation of Fusarium oxysporum f. sp. cubense tropical race 4 in Israel and the
702 Middle East. _Scientific Reports_ **10** : 1590.


703 **Mejías Herrera R, Hernández Y, Magdama F,** _**et al.**_ **2023** . First Report of Fusarium
704 Wilt of Cavendish Bananas Caused by _Fusarium oxysporum_ f. sp. _cubense_ Tropical
705 Race 4 in Venezuela. _Plant Disease_ **107** : 3297.


706 **Mintoff SJL, Nguyen TV, Kelly C,** _**et al.**_ **2021** . Banana Cultivar Field Screening for
707 Resistance to Fusarium oxysporum f.sp. cubense Tropical Race 4 in the Northern
708 Territory. _Journal of Fungi_ **7** : 627.


709 **Mostert D, Molina AB, Daniells J,** _**et al.**_ **2017** . The distribution and host range of
710 the banana Fusarium wilt fungus, Fusarium oxysporum f. sp. cubense, in Asia. _PLOS_
711 _ONE_ **12** : e0181630.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


712 **Pachacama S, Mostert D, Guevara J,** _**et al.**_ **2026** . First Report of Fusarium
713 oxysporum f. sp. cubense Tropical Race 4 Causing Fusarium Wilt to Bananas in
714 Ecuador. _New Disease Reports_ **54** : e70150.


715 **Pegg KG, Coates LM, O’Neill WT, Turner DW** . **2019** . The epidemiology of
716 Fusarium wilt of banana. _Frontiers in plant science_ **10** : 1395.


717 **Ploetz RC** . **2006** . Fusarium Wilt of Banana Is Caused by Several Pathogens
718 Referred to as Fusarium oxysporum f. sp. cubense. _Phytopathology_ **96** : 653–656.


719 **Ploetz RC** . **2015** . Management of Fusarium wilt of banana: A review with special
720 reference to tropical race 4. _Crop Protection_ **73** : 7–15.


721 **Ploetz R, Pegg K** . **1997** . Fusarium wilt of banana and Wallace’s line: Was the
722 disease originally restricted to his Indo-Malayan region? _Australasian Plant_
723 _Pathology_ **26** : 239.


724 **R Core Team** . **2024** . R: A language and environment for statistical computing.
725 &#91;Online&#93;.


726 **Rocha A de J, Soares JM da S, Nascimento F dos S,** _**et al.**_ **2021** . Improvements in
727 the Resistance of the Banana Species to Fusarium Wilt: A Systematic Review of
728 Methods and Perspectives. _Journal of Fungi_ **7** : 249.


729 **Rouard M, Blomme G, Ocimati W,** _**et al.**_ **2026** . Navigating Fusarium wilt of
730 bananas: a ready-to-use subset of resistant Musa genotypes. _Frontiers in Plant_
731 _Science_ **17** : 1753570.


732 **Ruas M, Guignon V, Sempéré G,** _**et al.**_ **2017** . MGIS: managing banana (Musa spp.)
733 genetic resources information and high-throughput genotyping data. _Database_ **2017** :
734 bax046.


735 **Scheerer L, Pemsl D, Dita M, Vicente LP, Staver C** . **2018** . A quantified approach to
736 projecting losses caused by Fusarium wilt Tropical race 4. _Acta Horticulturae_ **1196** :
737 211–218.


738 **Schneider CA, Rasband WS, Eliceiri KW** . **2012** . NIH Image to ImageJ: 25 years of
739 image analysis. _Nature Methods_ **9** : 671–675.


740 **Thangavelu R, Loganathan M, Arthee R, Prabakaran M, Uma S** . **2020** . Fusarium
741 wilt: a threat to banana cultivation and its management. _CAB Reviews: Perspectives_
742 _in Agriculture, Veterinary Science, Nutrition and Natural Resources_ **15** .


743 **Thangavelu R, Loganathan M, Backiyarani S,** _**et al.**_ **2025** . Screening of exotic
744 banana accessions for their resistance to fusarium wilt race 1 and tropical race 4 in
745 India. _Scientific Reports_ **15** : 25060.


746 **van Westerhoven AC, Meijer HJG, Houdijk J,** _**et al.**_ **2023** . Dissemination of
747 Fusarium Wilt of Banana in Mozambique Caused by _Fusarium odoratissimum_
748 Tropical Race 4. _Plant Disease_ **107** : 628–632.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


749 **Warmington RJ, Kay W, Jeffries A,** _**et al.**_ **2019** . High-Quality Draft Genome
750 Sequence of the Causal Agent of the Current Panama Disease Epidemic.
751 _Microbiology Resource Announcements_ **8** : e00904-19.


752 **Wickham H** . **2016** . _ggplot2: Elegant Graphics for Data Analysis_ . Springer-Verlag
753 New York.


754 **Xie W-Z, Zheng Y-Y, He W,** _**et al.**_ **2024** . Two haplotype-resolved genome assemblies
755 for AAB allotriploid bananas provide insights into banana subgenome asymmetric
756 evolution and Fusarium wilt control. _Plant Communications_ **5** : 100766.


757 **Zhan N, Kuang M, He W,** _**et al.**_ **2022** . Evaluation of Resistance of Banana
758 Genotypes with AAB Genome to Fusarium Wilt Tropical Race 4 in China. _Journal of_
759 _Fungi_ **8** : 1274.


760 **Zuo C, Deng G, Li B,** _**et al.**_ **2018** . Germplasm screening of Musa spp. for resistance
761 to Fusarium oxysporum f. sp. cubense tropical race 4 (Foc TR4). _European Journal_
762 _of Plant Pathology_ **151** : 723–734.


763


764


765 **Figure legends**


766 **Figure 1. Banana plants were micropropagated and established in soil before**
767 **inoculation.** (A–E) Tissue culture and micropropagation: (A) plantlets as received
768 from ITC; (B) plantlets transferred to a dish; (C) individual plants separated and
769 cleaned of dead tissue; (D) plants prepared for division; (E) explants placed in a jar
770 of propagation medium (PM) to start a micropropagation cycle. (F–K) Plant
771 refreshment and soil establishment: (F, G) plants removed from PM jars and cleaned
772 of dead tissue and medium; (H) individual plants cultured in rooting medium (RM)
773 until transfer to the growth cabinet; (I) plants transferred to trays with soil mix and
774 covered during the initial acclimatisation period to maintain high humidity; (J) plants
775 in pots, ready for inoculation assays; (K) plants re-potted when larger plants are
776 required.


777 Alt text: Eleven photographs showing the steps of banana tissue culture,
778 micropropagation and transfer of plantlets to soil trays and pots.


779 **Figure 2. Corm discolouration after Foc TR4 inoculation varies widely among**
780 **accessions.** Internal symptoms of Foc TR4 infection at 6 weeks after inoculation.
781 (A) Mock- and TR4-inoculated plants and longitudinal corm sections of a susceptible
782 accession (Figue Sucrée, ITC0107, AA) and a resistant accession ( _M. schizocarpa_,
783 ITC0926, SS). (B) Representative corm halves of a TR4-inoculated plant of each
784 accession, framed by response class: resistant (green), tolerant (purple) and very
785 susceptible (red) in the upper row, and susceptible (orange) in the lower row. Yellow
786 outlines show the healthy area delimited in ImageJ.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


787 Alt text: Photographs of banana plants and corm sections showing healthy corms in
788 a resistant wild species and dark, discoloured corms in a susceptible cultivar,
789 followed by one corm section for each of the 21 accessions.


790 **Figure 3. Eight accessions keep significantly more healthy corm tissue than**
791 **Dwarf Cavendish.** Internal symptoms of Foc TR4 infection quantified as the
792 percentage of healthy corm area at 6 weeks after inoculation (wpi). (A) Percentage
793 of healthy corm area (plant means) of all TR4-inoculated plants in each experiment;
794 diamonds, Dwarf Cavendish; circles, other accessions, coloured by response class;
795 n, number of plants. (B) Difference between the best linear unbiased estimate
796 (BLUE) of each accession and that of the susceptible control Dwarf Cavendish, with
797 95% confidence intervals; filled symbols, significant differences (Dunnett’s test, _P_ &lt;
798 0.05); open symbols, not significant. (C) Percentage of healthy corm area of each
799 corm half by accession; symbols indicate the cabinet quadrant (A–D), boxes are
800 coloured by response class (R, resistant; T, tolerant; S, susceptible; VS, very
801 susceptible) and n is the number of plants. Asterisks indicate significant differences
802 from Dwarf Cavendish (Dunnett’s test; * _P_ &lt; 0.05, ** _P_ &lt; 0.01, *** _P_ &lt; 0.001). The
803 green dotted line marks 80% healthy corm area, the threshold used to classify
804 resistant accessions. Boxes (A, C) show the median and interquartile range, and
805 whiskers extend to 1.5 times the interquartile range.


806 Alt text: Forest plot and box plots of the percentage of healthy corm area by
807 accession and by experiment; eight accessions have much higher healthy corm area
808 than the Dwarf Cavendish control, and two have lower.


809 **Figure 4. Leaf yellowing rises from 4 wpi in susceptible accessions but not in**
810 **resistant or tolerant ones.** External symptoms after inoculation with Foc TR4.
811 Adjusted means of (A) the total number of leaves and (B) the percentage of yellow
812 leaves from inoculation (week 0) to 6 weeks after inoculation, for TR4-inoculated
813 plants grouped by the response class of their accession (left, resistant (R) and
814 tolerant (T) accessions; right, susceptible (S) and very susceptible (VS) accessions)
815 and for mock-inoculated plants (grey open circles and dotted line, shown in both
816 panels). Means were estimated with a single linear mixed model for each trait fitted
817 to all weekly observations, with group, week and their interaction as fixed effects and
818 block (experiment × quadrant), accession and plant as random effects (Table S3);
819 error bars, 95% confidence intervals. Asterisks indicate significant differences from
820 mock plants within a week (Dunnett’s test; * _P_ &lt; 0.05, *** _P_ &lt; 0.001). The numbers of
821 plants and accessions in each group are given in the key. Data from experiments 1

822 and 3–7.


823 Alt text: Line plots of total leaves and percentage of yellow leaves over six weeks
824 after inoculation; susceptible and very susceptible groups have a higher percentage
825 of yellow leaves than mock plants from week 4, whereas resistant and tolerant
826 groups resemble mock plants.


827


828 **Table 1. Banana germplasm accessions used in this study, number of plants inoculated and response to Foc TR4.**































829

|Accession|ITCa&lt;br&gt;code|Biological&lt;br&gt;status|Species/&lt;br&gt;genome&lt;br&gt;group|Subspecies/subgroup|Cluster(Higginsetal&lt;br&gt;.&lt;br&gt;2024)|No of TR4 -inoculated&lt;br&gt;.&lt;br&gt;plantsscored|Mock-inoculated&lt;br&gt;plantsscored|TR4&lt;br&gt;response|
|---|---|---|---|---|---|---|---|---|
||||||||||
|_Musa_ _itinerans_|ITC1526|Wild|_itinerans_|var._xishuangbannaensis_|–|3|0|R|
|_Musa_ _acuminata_ subsp._malaccensis_|ITC0250|Wild|_acuminata_&lt;br&gt;(AA)|subsp._malaccensis_|–|2|1|S|
|_Musa_ _ornata_|ITC0370|Wild|_ornata_|subsp. _ornata_|–|9|1|T|
|_Musa_ _schizocarpa_|ITC0926|Wild|_schizocarpa_&lt;br&gt;(SS)|–|–|8|5|R|
|_Musa_ _textilis_|ITC1072|Wild|_textilis_ (TT)|subsp. _textilis_|–|13|2|S|
||||||||||
|Figue Sucrée|ITC0107|Cultivar|AA|Sucrier|Cl01_Sucrier_AA|9&lt;br&gt;|2|S|
|DwarfCavendish|ITC0002|Cultivar|AAA|Cavendish|Cl02_Cavendish_AAA|24b|4|S (control)|
|Gros Michel|ITC1122|Cultivar|AAA|Gros Michel|Cl03_Gros_Michel_AAA|8|1|S|
|IITA hybrid 2145/1320|ITC1448|Improved&lt;br&gt;cultivar|Unknownc|–|Cl04_unknown_AAA|8|6|S|
|Green Red|ITC0485|Cultivar|AAA|Red|Cl05_Red_AAA|9|0|S|
|FHIA-17|ITC1264|Improved&lt;br&gt;cultivar|AAAA|Synthetic tetraploid|Cl06_AAAA|9|2|S|
|Guineo|ITC0005|Cultivar|AAA|Mutika/Lujugira|Cl07_Mutika_AAA|8|4|S|
|Pome|ITC0076|Cultivar|AAB|Pome|–|8|3|VS|
|FHIA-02|ITC0505|Improved&lt;br&gt;cultivar|AAAB|Synthetic tetraploid|Cl08_Pome_AAABd|8|2|R|
|FHIA-21|ITC1332|Improved&lt;br&gt;cultivar|AAAB|Synthetic tetraploid|Cl09_Africa_AAAB|6|0|R|
|Mai’a Maoli Koa’e|ITC1167|Cultivar|AAB|Maia Maoli/Popoulu|Cl10_Popoulu_AAB|8|4|T|
|Curare|ITC1165|Cultivar|AAB|Plantain|Cl11_Plantain_AAB|3|0|R|
|Silver Bluggoe|ITC0364|Cultivar|ABB|Bluggoe|Cl12_Bluggoe_ABB|10|2|S|
|Pelipita|ITC0396|Cultivar|ABB|–|Cl13_Pelipita_ABB|6|1|T|
|Grande Naine ITC0180|ITC0180|Cultivar|AAA|Cavendish|–|10|4|VS|
|Tropic Grande Naine|–|Cultivar|AAA|Cavendish|–|11|3|S|



830 ITC codes, names, genome groups and subgroups follow the ITC passport data supplied with the material. R, resistant; T, tolerant; S, susceptible; VS, very
831 susceptible (see Materials and methods). &#91;a&#93; The International Musa Germplasm Transit Centre. &#91;b&#93; Corm data were obtained for 23 plants, of which 22 were
832 analysed after removal of one outlier. &#91;c&#93; Genome group not recorded in the ITC passport data; assigned to an AAA cluster by Higgins et al. (2024). &#91;d&#93; Cluster
833 labelled ‘Pome AAAB’ by Higgins et al. (2024). Mock-inoculated plants scored, mock plants with a recorded accession; a further 18 mock-inoculated plants
834 were excluded (see Results).


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.18.767592; this version posted November 20, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.18.767592)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


# **A**

**Susceptible acc.: Figue Sucrée (ITC0107, AA)** **Resistant acc.:** _**M. schizocarpa**_ **(ITC0926, SS)**


Mock (-) TR4 infected Mock (-) TR4 infected


# **B**























**Grande**














|A&lt;br&gt;Otheraccessions DwarfCavendish&lt;br&gt;100&lt;br&gt;80&lt;br&gt;Healthycormareaat6wpi&lt;br&gt;60 an)&lt;br&gt;tme&lt;br&gt;plan&lt;br&gt;40&lt;br&gt;(% ,&lt;br&gt;20&lt;br&gt;0&lt;br&gt;n=18 n=34 n=31 n=44 n=24 n=27&lt;br&gt;1 3 4 5 6 7&lt;br&gt;Experiment|B&lt;br&gt;n .s . p &lt; 0 .05&lt;br&gt;FHIA 21(AAAB)&lt;br&gt;−&lt;br&gt;FHIA 02(AAAB)&lt;br&gt;−&lt;br&gt;M itinerans&lt;br&gt;.&lt;br&gt;M schizocarpa&lt;br&gt;.&lt;br&gt;Curare(AAB)&lt;br&gt;M ornata&lt;br&gt;.&lt;br&gt;Pelipita(ABB)&lt;br&gt;Mai'aMaoliKoa'e(AAB)&lt;br&gt;TropicGrandeNaine(AAA)&lt;br&gt;M textilis&lt;br&gt;.&lt;br&gt;GreenRed(AAA)&lt;br&gt;SilverBluggoe(ABB)&lt;br&gt;Guineo(AAA)&lt;br&gt;GrosMichel(AAA)&lt;br&gt;M a subsp malaccensis&lt;br&gt;. . .&lt;br&gt;FigueSucrée(AA)&lt;br&gt;FHIA 17(AAAA)&lt;br&gt;−&lt;br&gt;IITAhybrid2145&lt;br&gt;GrandeNaineITC0180(AAA)&lt;br&gt;Pome(AAB)&lt;br&gt;50 0 50 100&lt;br&gt;−&lt;br&gt;DifferencefromDwarfCavendish&lt;br&gt;(BLUE % healt hycorm ;95%CI)&lt;br&gt;,|
|---|---|
|**C**&lt;br&gt;control&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;*&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;**&lt;br&gt;n=22&lt;br&gt;n=3&lt;br&gt;n=8&lt;br&gt;n=9&lt;br&gt;n=6&lt;br&gt;n=9&lt;br&gt;n=10&lt;br&gt;n=11&lt;br&gt;n=9&lt;br&gt;n=8&lt;br&gt;n=8&lt;br&gt;n=8&lt;br&gt;n=2&lt;br&gt;n=3&lt;br&gt;n=9&lt;br&gt;n=8&lt;br&gt;n=13&lt;br&gt;n=8&lt;br&gt;n=6&lt;br&gt;n=8&lt;br&gt;n=10&lt;br&gt;80%&lt;br&gt;0&lt;br&gt;20&lt;br&gt;40&lt;br&gt;60&lt;br&gt;80&lt;br&gt;100&lt;br&gt;FHIA−21 (AAAB)&lt;br&gt;FHIA−02 (AAAB)&lt;br&gt;_M. itinerans_&lt;br&gt;_M. schizocarpa_&lt;br&gt;Curare (AAB)&lt;br&gt;_M. ornata_&lt;br&gt;Pelipita (ABB)&lt;br&gt;Mai'a Maoli Koa'e (AAB)&lt;br&gt;Tropic Grande Naine (AAA)&lt;br&gt;_M. textilis_&lt;br&gt;Green Red (AAA)&lt;br&gt;Silver Bluggoe (ABB)&lt;br&gt;Dwarf Cavendish (AAA)&lt;br&gt;Guineo (AAA)&lt;br&gt;Gros Michel (AAA)&lt;br&gt;_M. a._ subsp. _malaccensis_&lt;br&gt;Figue Sucrée (AA)&lt;br&gt;FHIA−17 (AAAA)&lt;br&gt;IITA hybrid 2145&lt;br&gt;Grande Naine ITC0180 (AAA)&lt;br&gt;Pome (AAB)&lt;br&gt;Healthy corm area at 6 wpi (%)&lt;br&gt;Resistant (R)&lt;br&gt;Tolerant (T)&lt;br&gt;Susceptible (S)&lt;br&gt;Very susceptible (VS)&lt;br&gt;Quadrant&lt;br&gt;A&lt;br&gt;B&lt;br&gt;C&lt;br&gt;D|**C**&lt;br&gt;control&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;*&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;***&lt;br&gt;**&lt;br&gt;n=22&lt;br&gt;n=3&lt;br&gt;n=8&lt;br&gt;n=9&lt;br&gt;n=6&lt;br&gt;n=9&lt;br&gt;n=10&lt;br&gt;n=11&lt;br&gt;n=9&lt;br&gt;n=8&lt;br&gt;n=8&lt;br&gt;n=8&lt;br&gt;n=2&lt;br&gt;n=3&lt;br&gt;n=9&lt;br&gt;n=8&lt;br&gt;n=13&lt;br&gt;n=8&lt;br&gt;n=6&lt;br&gt;n=8&lt;br&gt;n=10&lt;br&gt;80%&lt;br&gt;0&lt;br&gt;20&lt;br&gt;40&lt;br&gt;60&lt;br&gt;80&lt;br&gt;100&lt;br&gt;FHIA−21 (AAAB)&lt;br&gt;FHIA−02 (AAAB)&lt;br&gt;_M. itinerans_&lt;br&gt;_M. schizocarpa_&lt;br&gt;Curare (AAB)&lt;br&gt;_M. ornata_&lt;br&gt;Pelipita (ABB)&lt;br&gt;Mai'a Maoli Koa'e (AAB)&lt;br&gt;Tropic Grande Naine (AAA)&lt;br&gt;_M. textilis_&lt;br&gt;Green Red (AAA)&lt;br&gt;Silver Bluggoe (ABB)&lt;br&gt;Dwarf Cavendish (AAA)&lt;br&gt;Guineo (AAA)&lt;br&gt;Gros Michel (AAA)&lt;br&gt;_M. a._ subsp. _malaccensis_&lt;br&gt;Figue Sucrée (AA)&lt;br&gt;FHIA−17 (AAAA)&lt;br&gt;IITA hybrid 2145&lt;br&gt;Grande Naine ITC0180 (AAA)&lt;br&gt;Pome (AAB)&lt;br&gt;Healthy corm area at 6 wpi (%)&lt;br&gt;Resistant (R)&lt;br&gt;Tolerant (T)&lt;br&gt;Susceptible (S)&lt;br&gt;Very susceptible (VS)&lt;br&gt;Quadrant&lt;br&gt;A&lt;br&gt;B&lt;br&gt;C&lt;br&gt;D|


