Single torradovirus infections explain the mysterious cassava frogskin disease in the Americas Jenyfer Jimenez1, Sara Caicedo1,2, Juan M. Pardo1, Alejandra Gil-Ordóñez1,3, Robert Alvarez-Quinto4, Dimitre Mollov5 & Wilmer J. Cuellar1 Cassava frogskin disease (CFSD) emerged in the Americas in the 1970s, but its causal agent has to date remained a mystery. The clonal propagation of cassava, high incidence of mixed infections, unknown alternative hosts, and root symptoms taking two or more crop cycles to develop, have made it difficult to identify the causal agent. Consequently, most studies on CFSD have produced a catalogue of pathogens occurring in affected plants. Using a sentinel approach, we captured single-pathogen infections in fields with high incidence of root symptoms. Eight months after being exposed to CFSD, we detected < 6.9% incidence of root symptoms in sentinel plants. Plants were then propagated and transferred to a screenhouse for a second infection cycle and storage root development. Interestingly, molecular diagnostics did not identify an association with phytoplasma or reovirids—pathogens historically reported in CFSD-infected plants—but indicated that single-infections by torradoviruses were sufficient to cause the disease. Further analysis by high-throughput sequencing confirmed the presence of torradoviruses in symptomatic roots and allowed unveiling the occurrence of a second torradovirus species in farmers’ fields in Colombia. These new findings should support early interception of infected planting material, development of cassava seed certification standards, breeding and screening for resistance programs, and ultimately significantly reduce the impact of CFSD in cassava. Keywords Cassava, Emergent disease, High-throughput sequencing, Disease diagnostics Cassava, a tropical starchy root crop, is a crucial staple food for millions of people worldwide, with a significant potential for income generation through the sale of fresh storage roots and processed products1. As a food crop, it ranks third in global importance, and fifth as the most-widely used source of starch2. In 2022, Africa’s cassava accounted for 63% of the 330  million tons produced worldwide, Asian countries’ for about 29% and in the Americas for about 8% of global cassava production3. The leading producers of cassava are Nigeria, the Democratic Republic of Congo, and Thailand3. Asia—notably Thailand, Cambodia, Indonesia and Vietnam—also contributes significantly to global cassava production to meet the demand for dried cassava chips and cassava starch, focusing on its use as a commercial livestock feed and an industrial cash crop for products such as biofuel4–6. In the Americas, Brazil is the main producer, leading in 2022 with approximately 17 million tons3. Traditionally in this region, cassava has been used primarily for human consumption and as on-farm animal feed, rather than for starch processing7,8. In this region, processing cassava into starch has been predominantly concentrated in countries such as Brazil, Colombia, and Paraguay6. Each of these regions (Africa, the Americas, and Asia) must deal with distinct disease pressures that limit local cassava production9. This includes transboundary pathogens that constantly challenge regional phytosanitary measures, as exemplified by the recent spread of cassava mosaic disease (CMD) from south India to continental Southeast Asia10. Since its first reported incidence11, cassava frogskin disease (CFSD) has been associated with mixed infections12. This was expected, as cassava is a clonally propagated crop, where pathogens build up during each crop cycle, compromising planting material (stakes) quality and eventually reducing yield13,14. More than seven different pathogen genera have been found in CFSD-affected plants12, suggesting that CFSD could be a 1Virology and Crop Protection Laboratory, Cassava Program, International Center for Tropical Agriculture (CIAT), Recta Cali-Palmira Km 17, Palmira, Colombia. 2Facultad de Ingeniería, Diseño y Ciencias Aplicadas, Universidad ICESI, Cali, Colombia. 3Departamento de Biología, Facultad de Ciencias Naturales y Exactas, Universidad del Valle, 760032 Cali, Colombia. 4Department of Plant Pathology, University of Minnesota, Saint Paul, MN, USA. 5Pest Exclusion and Import Programs, USDA APHIS Plant Protection and Quarantine, Riverdale, MD, USA. email: w.cuellar@cgiar.org OPEN Scientific Reports | (2024) 14:29648 1| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports http://www.nature.com/scientificreports http://crossmark.crossref.org/dialog/?doi=10.1038/s41598-024-81142-2&domain=pdf&date_stamp=2024-11-28 consequence of such microorganism accumulation. These complex mixed infections or pathobiomes15 have not been reported outside the Americas, where cassava is more widely grown. In Africa and Asia, geminiviruses dominate the cassava pathobiome9, while in Southeast Asia and the Pacific Islands one geminivirus (Sri Lankan cassava mosaic virus) and one fungus (Ceratobasidium theobromae) predominate10,16. The absence of these pathogens in the Americas indicates that they are native to Africa and Asia (i.e., they have not arrived with the crop). The common occurrence of more complex mixed infections in the Americas by a distinct pathobiome than that occurring in Africa and Asia, could be related to cassava being native to this region17. Interestingly, of eight different virus species commonly detected in mixed infection in the Americas, only two have been shown to cause disease in single infection: cassava vein mosaic virus18 and cassava common mosaic virus19,20. Mixed pathogen infections have confused CFSD diagnostics for more than 50 years12. Initially, clostero-like virus particles and mycoplasmas (later renamed as phytoplasmas) were detected in CFSD-affected roots, followed by the identification of several species of double-stranded RNA (dsRNA), which suggested mixed infections21. Experiments in the early 2000s revealed the presence in infected tissues of phytoplasmas and different strains of one reovirid, originally named as cassava frogskin-associated virus (CsFSaV; Fam. Sedoreoviridae, Genus: Oryzavirus)22. Reovirids are a group of viruses composed of up to 12 segments of dsRNA23. These findings explained previous observations and the hypothesis that either of them (phytoplasma or CsFSaV) causes the disease has been accepted since 200922,24–27. Despite this, molecular diagnostics targeting each of these pathogens have given unclear results, and a high proportion of plants with root symptoms of the disease have tested negative for phytoplasma presence by PCR. Likewise, CsFSaV is commonly detected in plants that do not show root symptoms of the disease12. High-throughput sequencing (HTS) analysis of CFSD symptomatic samples apparently single-infected by either phytoplasma or CsFSaV, would unravel the occurrence of more complex mixed infections in these plants28,29, suggesting that either hypothesis (phytoplasma or CsFSaV) could be revised. As clear CFSD symptoms form in the mature cassava roots, experimental reproduction of root symptoms with any candidate pathogen, is a lengthy task, starting by separating each candidate from the mix. One approach is to screen germplasm collections, after samples have gone through several rounds of cleaning (e.g. via thermotherapy, meristem culture, etc.), to find single pathogen infections and then evaluate them in different combinations. Such an approach, however, would require at least three crop cycles (~ 3 years), after establishing the in-vitro plantlets in a screenhouse, to normalize treatments and reproduce root symptoms. This could explain why most studies on CFSD diagnostics have resulted in listing a catalogue of different pathogens found in affected plants22,24–30. Although no major advance has been reported on CFSD etiology, a recent review of four decades of CFSD research12 has shed some light on the nature of the pathogen: (1) CFSD is an infectious disease and its causal agent can be transmitted by grafting; (2) transmission of CFSD root symptoms requires an aerial vector; (3) frogskin leaf mosaic is associated with mixed infections in only a couple of genotypes, and is not a generic symptom of the disease; and (4) all cassava stakes from a plant showing root symptoms of CFSD can develop the disease, which suggests that the pathogen infects cassava systemically. To identify the causal agent of CFSD, we considered these observations and used the CFSD-susceptible cassava genotype “Reina” (CM6740-7), which develops clear root symptoms of CFSD, as a ‘sentinel’ exposed to a high disease pressure in the field. Sentinel plants should display visible symptoms at an early stage of infection and are generally used as early warning guides of infection31. Using this approach, we captured single-pathogen infections and uncovered that, single infections by the cassava torrado-like virus (CsTLV; Family Secoviridae, Genus Torradovirus) were sufficient to explain root symptoms of CFSD. This was confirmed by routine molecular diagnostics and HTS. Interestingly, two of the pathogens widely thought to be the causal agent of CFSD, phytoplasmas24 and CsFSaV22, were found not to be associated with the disease. Only torradoviruses could be found in symptomatic roots that could explain CFSD. A deeper analysis of torradovirus diversity in Colombia revealed the common occurrence of a distinct CsTLV species, explaining the high percentage of false negative results obtained in previous analyses targeting another species28,32. Our results highlight the importance of diagnostics validation under experimental conditions and understanding pathogen diversity, to decipher the etiology of diseases in complex scenarios. We recommend applying this new information for the early detection and management of CFSD in the Americas and to prevent its spread to Africa and Asia. Results Low incidences of root symptoms and single infections detected in sentinel plants after eight months When analyzing plants from two independent sentinel trials (named REP1 and REP2), only a small percentage developed clear root symptoms of CFSD (degree 3 or above), two sentinels out of 36 in REP1 (5.5%), and three sentinels out of 43 in REP2 (6.9%). In contrast, sentinel plants infected with the CFSD inoculum CM5460-10 showed root symptoms in 95.6% of the plants with an average severity scale degree of 4.5, with 5 being the most affected (Table 1; Supplementary Table 1). On average, agronomic traits such as dry matter and number of affected roots per plant, were significantly different between sentinel (Healthy) and mixed infected plants (Diseased). However, after one year, the effect of CFSD on yield (fresh weight of roots per plant) was not significant (Fig.  1). These results were as expected according to previous field trial records of the disease, as reviewed by Pardo et al. 12. Most unexpectedly, molecular diagnostics of sentinel plants for all pathogens previously reported in CFSD-affected plants12,22,24,25,27–29, detected only one plant infected with phytoplasma and another one with CsFSaV in REP1 (planting date August 2022), none of which showed root symptoms of CFSD. On the other hand, in REP2 (planting date August 2023), no phytoplasma was detected while CsFSaV was present in almost all plants (with or without root symptoms of CFSD). Cassava polero-like virus (CsPLV; Fam. Solemoviridae, Gen. Polerovirus), another virus commonly occurring in mixed infections, was not detected in these trials. In contrast, all plants showing clear root symptoms of CFSD were positive to CsTLV, Scientific Reports | (2024) 14:29648 2| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports and among these, CsTLV was detected in single infections in two plants, one in each REP experiment and these plants were further analyzed (Supplementary Table 1). Therefore, sentinel trials allowed us to identify single pathogen infections, and statistical analysis of molecular diagnostics results versus occurrence of root symptoms of CFSD suggested that neither phytoplasma nor CsFSaV were associated with the disease, and indicated that single infections by CsTLV in sentinel plants, were sufficient to explain CFSD. Metagenomics confirms torradoviruses in single-infected symptomatic roots while CsFSaV and phytoplasma were not detected To validate the single occurrence of CsTLV in sentinel plants with root symptoms of CFSD, these plants were used as a planting material source that was then transferred to a soil-bed inside an insect-proof screenhouse, for a second infection cycle and to allow for storage root symptoms to replicate (Fig. 2). A second round of molecular diagnostics from young leaves confirmed the absence of CsFSaV and phytoplasma in these plants (not shown). Further analysis included molecular diagnostics from total nucleic acids extracted directly from root tissue. The protocol described here produced on average 300–700 µg/µl of total nucleic acids per 300 mg of fresh root tissue. Again, standard nested PCR and LAMP25,33 failed to detect phytoplasma in these samples and RT-PCR did not amplify any CsPLV PCR product (Supplementary Table 1). To avoid false negative results by (RT-)PCR or LAMP diagnostics, total nucleic acids extracted from affected root tissue were submitted for HTS producing an average of 245,556 (± 229,114) and 12,331,331 (± 2,950,105) non-host-associated PE150 Q > 30 sequences Fig. 1. Characterization of healthy and diseased roots after eight months in the field. The left panel shows an example of a healthy (degree 1) (A) and a diseased root (degree 4) (B) with the characteristic root symptoms of CFSD. On the right panel we show the comparative results of a one-year field evaluation. As expected, the effect of CFSD on yield in kg (C) was not as significant after one crop cycle, even when using infected stakes as starting material. Nevertheless, significant differences can already be observed on dry matter content (D), the frequency of affected roots versus the severity scale recorded (E) and the percentage of affected roots per plant (F), indicating that the environment where the plants were grown was optimal for disease development. D, E, F box plots showing significant differences (p < 0.05), according to the t-Student test. ns = no significant. n Root symptoms Phytoplasma 16Sr III CsFSaV CsPLV CsTLV-1 CsTLV-2 REP1 (2022) 36 5.5% 2.8% 2.8% 0% 0% 2.8% REP2 (2023) 43 6.9% 0% 97.7% 0% 2.3% 39.5%  Reina 38 0% 0% 0% 0% 0% 0% ReinaCM5460−10* 23 95.6% 4.4% 95.6% 91.3% 8.7% 100% Table 1. Using sentinel plants to capture single pathogen infections in the field. Healthy susceptible plants of genotype CM6740-7 (Reina) were used as sentinels and exposed to an infected field in two different years (REP1 and REP2) and analyzed one month before harvest. CsTLV-1 and CsTLV-2 refer to two distinct species of torradoviruses found in cassava (see below). Virus buildup in REP2 was detected in almost all plants, mainly the double infection of CsFSaV with CsTLV-2. To verify that the field conditions were adequate for the development of CFSD root symptoms we used the same sentinel plants previously graft-infected with the CM5460-10 inoculum12(*). Scientific Reports | (2024) 14:29648 3| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports from DNA-seq (Supplementary Table 2, Sheet 1, Samples 1–4) and RNA-seq (Supplementary Table 2, Sheet 1, Samples 5–8), respectively. While HTS of small interfering RNA (siRNA-seq) produced 68,179,102 SE75 reads of 18–28 nt (Supplementary Table 2, Sheet 2). According to the rarefaction curve of DNA-seq and RNA-seq data, sampling sufficiency was achieved in all samples (Supplementary Fig. 1). DNA-seq and RNA-seq analyses did not detect a single read belonging to phytoplasma or CsFSaV groups in CFSD-affected plants, confirming molecular diagnostic results. Instead, DNA-seq and RNA-seq results showed that the only suspected plant pathogen found exclusively in CFSD roots was torradovirus CsTLV (Fig. 2, Supplementary Table 2, Sheet 4–5) and in fact a full genome could be assembled, with an average coverage of 3,150X for RNA1 and 8,861X for RNA2 (see below). In a previous work, we found out that a disease thought to be caused by a phytoplasma, was in fact associated with a fungus16. We therefore checked for fungal pathogens, but no additional taxonomic trends could be detected in the diseased samples’ microbial profile. Finally, HTS of siRNA, a gold-standard method for virus discovery34, failed to detect any other virus contig that could explain the root symptoms of CFSD, except for CsTLV. Interestingly, healthy root samples tested positive for cassava common mosaic virus (Fam. Alphaflexiviridae, Gen. Potexvirus) (Fig. 2). This is one of several species of potexviruses infecting cassava in Colombia, and the only one among them that causes single-infection disease in cassava20,35. Although this virus could be commonly found in cassava, it has never been associated with CFSD, as is again confirmed here by its occurrence in the roots of plants that do not show root symptoms even after two crop cycles. Interestingly, the CsTLV isolate present in symptomatic root samples shared less than 65% aa identity with our reference sequence Yop12 (GenBank UAW09555), indicating that a wider CsTLV diversity was present in Colombia. Fig. 2. Metagenomics analysis of CsTLV single-infected symptomatic roots. (A) Healthy root of a sentinel plant next to (B) a root of a plant infected only with CsTLV showing clear symptoms (degree 4) of CFSD. Notice the darker skin color and characteristic “lips” of the diseased root covering the whole root. Taxonomic classification of microbial contigs obtained from healthy roots (1−2) and plants with root symptoms of CFSD (3−4) using Kraken2 with the nt database (2023-05-02). Taxonomic classification of the eight most-abundant taxa of microbiome and plant viruses based on shotgun metagenomics of DNA (C) and RNA (D). Rarefaction curves of DNA-seq and RNA-seq data indicate that sampling sufficiency was achieved in all samples (Supplementary Fig. 1). Scientific Reports | (2024) 14:29648 4| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports Two species of torradovirus infecting cassava plants with CFSD in Colombia Geographically distinct CsTLV isolates were characterized in parallel with sentinel and disease surveys work, prompted by previous results where plants showing CFSD root symptoms proved negative to CsTLV using PCR primers designed from sequences of isolates Sec13 and Yop1232,36. Several near complete CsTLV genomes were assembled using the protocol described here. RNA-seq results produced an average of 580,303 (± 598,462) of non-host-associated PE150 Q > 30 sequences (Supplementary Table 2, Sheet 1, Samples 9–13). All isolates unveiled a conserved CsTLV genome organization generally consisting of a single polyprotein encoded in RNA1 and two overlapping proteins encoded in RNA232,36,37 (Supplementary Table 3). We identified all ORFs and conserved domains described in the reference genome Yop12, which is also observed in other members of the genus Torradovirus38 and confirmed the occurrence of a predicted Maf/Ham1 domain32 in RNA1 of all isolates (Supplementary Table 3). After comparative analysis including all other CsTLV genomes available in GenBank, we consistently observed the formation of two distinct phylogenetic groups that correspond to two distinct species of torradovirus (according to the latest ICTV taxonomic demarcation criteria using the protease- polymerase (Pro-Pol) region) (Fig. 3). The first group of CsTLV isolates, hereby named CsTLV-1 is formed by our reference genome, isolate Yop1232, isolate CM5460-10 and isolate Sec13 from Colombia, and isolate PV- 1279 isolated from Brazil (GenBank accession no. UZN89639). All these isolates were originally field collected from plants with root symptoms of CFSD, except for isolate Sec13, which was originally detected in an in vitro plant28. No other subdivision could be recognized in this group. The second group, hereby called CsTLV-2 shares only 64% aa identity in the Pro-Pol region, with the CsTLV-1 group, and is organized into two subgroups, whose nucleotide identities (~ 94% aa identity) and geographical locations would allow us to consider them as two strains (Fig. 3). Using RNA-seq and siRNA information we assembled a near full genome of a representative of CsTLV-2. This isolate, named Pal52, had an overall sequence identity of 64% in aa with the type species of CsTLV-1, isolate Yop1232. Although all conserved domains are present in both CsTLV species, we noticed a change in the putative active site of the Maf/Ham1 domain, from GLR in CsTLV-1 to DLR in CsTLV-2, which align with the conserved SHR site found in the Maf/Ham1 domain of other cassava infecting viruses39 (Fig. 4). Improved diagnostics shows the association of torradoviruses with CFSD in farmers’ fields New PCR primers designed using novel sequence information of CsTLV-1 and CsTLV-2 isolates were used to explore the association of these viruses with root symptoms in farmers’ fields. These primers targeted the RdRP and Maf/Ham1 region from RNA1 for CsTLV-1 and CsTLV-2, respectively. As mixed infections with CsTLV-1 and CsTLV-2 could be detected in some plants (e.g., CM5460-10 or Yop12 in Fig. 3), we analyzed all samples with both sets of primers. Our results showed that one species of torradovirus (CsTLV-2) is commonly detected, in fact only samples from a specific geographic region (Yopal) were infected with CsTLV-1 (Supplementary Table 4). Coincidentally it was the CsTLV-2 found in most plants that displayed root symptoms of CFSD, which explains why previous diagnostics designed to target CsTLV-1 (the only species known at the time28,32), would produce false negatives. Although we also detected a high number of PCR-positive plants that did not show root symptoms, this proportion was lower in fields with zero incidence of root symptoms, or in fields where “clean seed” was reportedly used by farmers, who indicated the use of planting material from regions where CFSD is not reported, as a disease management strategy. In all cases, > 93% of plants showing clear root symptoms of CFSD (degree 3 or above) were positive to CsTLV (Table 2). It was interesting that, unlike our results with the sentinels, CsFSaV and CsPLV were consistently detected in almost all field collected samples, including fields that used disease-free planting material (Supplementary Table 4). This result indicates that these viruses are not associated with CFSD, but that they can be commonly detected in affected plants. In these surveys, to reduce the “noise” that tolerant genotypes could introduce to our diagnostics, we worked with only one genotype in each region (‘Brasilera’ in Meta and ‘Veronica’ in Córdoba), both known to be susceptible to CFSD. On the other hand, samples collected in Yopal in 2020 and Granada in 2023 were all from different genotypes, but all showing root symptoms of CFSD (Supplementary Table 4). In conclusion, all field-collected plants displaying root symptoms of CFSD analyzed in this study were infected by CsTLV, resulting in a 93.10% sensitivity of our diagnostic tests (Table 2). On the other hand, CsFSaV was detected in equal proportions in healthy (~94%) and diseased (~97%) plants, while phytoplasma incidence was lower in both cases, having a relatively higher incidence in symptomatic plants (~21%) than in healthy plants (~6%). Discussion CFSD was a difficult puzzle to solve due to the biology of the disease (symptoms are only observed in mature storage roots) and the common occurrence of mixed pathogen infections in the vegetatively propagated cassava host. However, results obtained over decades of research12 altogether pointed to several characteristics of the elusive causal agent: (1) it is graft-transmissible; (2) infects the plant systemically; (3) it requires an aerial vector for transmission; and (4) leaf symptoms are an exception observed in only a few cassava genotypes. Several groups attempting to identify the causal agent of CFSD already evidenced the occurrence of mixed infections, identifying different pathogens in their studied plants22,24,28,29. In practice, reproducing CFSD root symptoms under screenhouse conditions requires 2‒3 crop cycles when starting with clean planting material. Development of symptoms after only one year in the field is likely due to the inadvertent use of already mixed-infected planting material12. Our results, obtained using mixed infected planting material (inoculum CM5460-10), show this as a reproducible phenomenon (95.6%  incidence of root symptoms in one year), compared to the use of clean planting material (< 6.9% incidence of root symptoms) (Table 1). It is noteworthy that torradoviruses occurred in mixed infections in all locations where high CFSD-root symptoms incidences were observed (Supplementary Table 4). The use of the CFSD-susceptible genotype Reina (CM6740-7) as a sentinel in a high disease-incidence scenario aimed to capture single pathogen infections, as we reasoned that not all pathogens identified in mixed Scientific Reports | (2024) 14:29648 5| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports Fig. 3. Phylogenetic relationships of torradoviruses infecting cassava with root symptoms of CFSD. (A) Genome analysis of cassava-infecting torradoviruses shows a high diversity of geographically distinct isolates, which organize into two different species (sharing less than 65% aa identity between them). The names in red indicate the type species for each group. Genomes obtained in this work are indicated with an asterisk. The two subgroups observed in the CsTLV-2 group share 90–91% aa identity. (B) Two sets of PCR primers were designed based on the new genomic information available and tested for CsTLV in a set of infected samples. Notice that samples 1 and 3 are infected by both CsTLV species. (I) PCR for internal control cDNA. (II) PCR using specific primers for CsTLV-1. (III) PCR using specific primers for CsTLV-2. Isolates: Yop12 (lane 1), Sec13 (lane 2), CM5460-10 (lane 3). Samples from Meta (lanes 4–6), Arauca (lanes 7–9), Bolivar (lanes 10–12) and Córdoba (lanes 13–15) C+: Positive control, C-: Healthy cassava, Crx: Reaction control for cDNA synthesis, W: Water control, M: 1 Kb Plus DNA Ladder (Invitrogen) black arrowheads indicate 633 and 606 bp bands (II−III), the white arrowhead indicates the 181 bp band of the internal control. Scientific Reports | (2024) 14:29648 6| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports infections will be transmitted at the same rate by aerial vectors. Furthermore, the experimental fields with high incidence of CFSD serving as an inoculum source did not receive pest management during the duration of the sentinel experiments, increasing the probability of transmission of the different pathogens. The different infection percentages in REP1 versus REP2 could indicate that different insect groups visited each REP in the different years the sentinel plants were exposed; for example, the high percentage of CsFSaV only in REP2 (Table 1). Further work should address the identification of specific insect vectors for CsFSaV and CsTLV. Our results discarded a role for CsFSaV and phytoplasma in forming the characteristic root symptoms of CFSD. These hypotheses, that either CsFSaV or a phytoplasma were the causal agents of CFSD, have remained unchallenged for more than 15 years, since these pathogens were identified in Colombia22,24, and later in Paraguay, Costa Rica, and Brazil25,27,40. Nevertheless, an association between the specific pathogen and the root symptoms could not be found in any of these reports. A similar situation was observed for cassava witches’ broom disease (CWBD) in Southeast Asia, where the phytoplasma hypothesis did not stand diagnostic tests carried out by independent laboratories and instead a fungus was uncovered to be causing the disease16,41. The low titers and uneven distribution of phytoplasma within the diseased plant were considered as an explanation for the high percentage of putative false negatives obtained during molecular diagnostics of field samples. Phenotype Number of samples CsTLV RT-PCR (-) CsTLV RT-PCR (+) Healthy 90 66 24 Diseased 29 2 27** Unclear symptoms* 1 1 0 Table 2. Root symptoms and occurrence of CsTLV in farmers’ fields. RT-PCR analysis of field samples (top youngest leaves from plants with and without root disease symptoms) showed the association of CsTLV with CFSD. Fields were surveyed following a diagonal transect over a 1 ha field and samples were collected every 4th plant along the transects. All PCR bands obtained were confirmed to correspond to CsTLV-1 or CsTLV-2 by ONT sequencing. The sensitivity (93.10%) was calculated as TP/(TP + FN)*100, and specificity (73.33%) calculated as TN/(TN + FP)*100. TP, FN, TN, FP denotes True Negative, True Positive, False Negative, and False Positive, respectively. Additional data is available in Supplementary Table 4. *Unclear symptoms indicate plants with indistinguishable root symptoms. **Four of these plants were infected with CsTLV-1 only, all other samples were infected with CstLV-2 (See Supplementary Table 4). Fig. 4. The genome organization of CsTLV-2 isolate Pal52. A near complete genome of CsTLV-2 was assembled from RNA-seq data and is shown here with the viral siRNA mapped to its consensus sequence (average siRNA coverage of ~ 587X (RNA1) and ~ 1123X (RNA2). The bipartite genome of Pal52 shows the characteristic organization of torradoviruses: RNA1 (GenBank: PQ493399) composed of a single polyprotein encoding replication domains and a Maf/Ham1 domain at its 3’ end while RNA2 (GenBank: PQ493410) encodes two overlapping ORFs encoding virus movement and coat protein domains37,38. Sequences in red indicate some of the insertions present only in CsTLV-1 (See Supplementary Table 2 for more details). Arrowheads in black and red indicate the approximate location of diagnostic PCR-primers for CsTLV-1 and CsTLV-2, respectively. The sequence in green indicates the conserved active site of the Maf/Ham1 domain in CsTLV-2. The three conserved aa ‘SHR’ in the Maf/Ham1 domain of cassava-infecting ipomoviruses39 is changed to ‘GLR’ in CsTLV-1 and to ‘DLR’ in CsTLV-2. Scientific Reports | (2024) 14:29648 7| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports However, even the highly sensitive DNA-seq and RNA-seq results from samples taken directly from the affected tissue (affected roots) failed to detect any contigs belonging to the phytoplasma group (Fig. 2). On the other hand, CsFSaV was ubiquitous in farmers’ fields, and our sentinel results indicate that they may be efficiently transmitted (see REP2 in Table 1; Supplementary Table 1). Besides, its dsRNA nature also makes these latter more stable RNA molecules during RNA isolation processes for molecular diagnostics, explaining why they are consistently detected in cassava21,22,26,28. Nevertheless, the common occurrence of CsFSaV in healthy plants and its absence in CFSD symptomatic roots, as revealed by RNA-seq analysis, eliminate any role for this group of viruses in causing CFSD. Further studies should investigate any synergistic effect this virus may have in co- infection with torradoviruses. Here we show that torradoviruses are the only group of pathogens that can sufficiently explain the observed CFSD root symptoms. The low incidence of root symptoms along with > 39% torradovirus incidence observed in REP2 (Table  1) is likely related to late virus transmissions and agrees with previous observations, where root symptoms require two or three crop cycles to develop, if one starts with clean planting material12. On the other hand, using mixed-infected ‘Reina’ plants showed that the field conditions during these experiments were appropriate for root symptom development during only one crop cycle. The latter allowed us to agronomically reproduce several disease characteristics (Fig. 1). The more severe effects that mixed infections have in the field fit a seed degeneration model. Seed degeneration can be defined as an increase in pest and/or pathogen incidence or severity, associated in this case with reduction in yield or quality of storage roots over successive cycles of vegetative propagation42. We have also observed more severe symptoms in roots after a second year even in plants single-infected with CsTLV under screen house conditions. Under farmers’ conditions, the cumulative effect from additional pathogens in planting material can mirror what we have observed here and should be more thoroughly studied. The high incidence of a second species of torradovirus (CsTLV-2) as unveiled by HTS, explains in part the conflicting results obtained when using torradovirus generic primers43 or those originally designed based on CsTLV-132. Improving the diagnostic protocol for CsTLV using newly obtained virus genome information (Fig. 4) made it possible to confirm the association of torradoviruses with CFSD root symptoms in recently collected samples from farmers’ fields (Supplementary Table 4). In our experience, the relatively low specificity (73.33%) of our diagnostic test (Table 2) is likely due to CsTLV infections occurring when storage roots have already developed, as with our results in REP2 (Table 1). It is noteworthy that > 93% (37/40) of field-collected plants in the eastern plains and > 96% (57/59) in the north coast of Colombia, were infected with the reovirid CsFSaV. This was the case whether the plants displayed root symptoms or not ‒ as with our results observed in REP2 (Supplementary Table 4), and likely due to the presence of higher populations of the aerial vector. This high occurrence of the reovirid CsFSaV infecting cassava explains its common detection if one analyzes a small number of CFSD symptomatic plants only18,21. In conclusion, our results identify CsTLV-2 as a sufficient agent causing CFSD, and field data shows that plants infected with CsTLV-1 (see Table 2 and Supplementary Table 4), can also develop root symptoms. The virus can be detected in top leaves and roots, and stakes from different parts of an infected plant develop the characteristic root symptoms (Fig.  2). This indicates that CsTLV infects cassava systemically. This is a clear difference from CWBD, where the localized accumulation of the fungal pathogen allows parts of the stem to be disease-free44. It is interesting that we detected the virus even before leaf symptoms develop (a characteristic response of ‘Reina’), confirming that molecular diagnostics should replace virus indexing by grafting and leaf symptom observation. Although virus surveillance using leaf samples efficiently detected CsTLV-1 and CsTLV-2, in this case we recommend developing protocols for molecular diagnostics directly from root samples to minimize the number of false negatives. Furthermore, as experimental data suggest that CFSD is transmitted by an aerial vector, we should indicate that most torradoviruses described so far are transmitted by whiteflies and aphids38,45–49. While the specific insect vector of cassava-infecting torradoviruses is still unknown, the low incidence of root symptoms (less than 6.9%) in fields with up to 39.5% of infected plants with CsTLV-2 (Table 1) confirms that although insect transmission is effective, CFSD root symptoms require more than one crop cycle to emerge in large incidences. Therefore, CFSD could be initially managed by using clean planting material. At the same time, our results confirm that the sudden occurrence of high incidences of CFSD must be driven by the inadvertent use and distribution of already infected planting material. These new findings should support cassava seed certification standards, early interception of infected planting material, and breeding and screening for resistance programs, aiming to reduce the impact of CFSD in affected regions. Materials and methods Plant material and nucleic acid extraction Two months-old virus-free cassava plants of the CFSD susceptible variety, CM6740-7 known as “Reina”, were used as sentinels, grown in a CFSD ‘contaminated’ field for eight months in two experiments named REP1 (36 plants) and REP2 (43 plants). These were carried out in 2022 and in 2023, respectively at the International Center for Tropical Agriculture (CIAT), Palmira, Colombia. Plants were sampled (top youngest leaves and roots symptom photographic record) one month before harvest, and planting material (stakes) was propagated for a second infection cycle, and then transferred to an insect-proof screenhouse at CIAT. Transferring to a screenhouse would stop further infections. Propagated plants were then transferred to a soil bed to allow for storage root development. Reina stakes inoculated with CM5460-10 were used as positive controls for root symptom development, and healthy Reina stakes were used as a source of experimental planting material, and as negative controls for root symptoms development. Plants were grown and maintained at 28 ± 5 °C and 70–80% of relative humidity. Field sample collections were carried out non-destructively at harvest time, and in accordance with relevant institutional guidelines and legislation. Samples Met3-153 and Ara4-141 were collected in 2022 in the eastern Scientific Reports | (2024) 14:29648 8| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports plains of Colombia (Departments of Meta and Arauca, respectively). Samples Bol18-254, Bol18-258 and Bol22- 227 were collected in 2017 from fields located on the north coast of Colombia (Department of Bolivar) and Sample Pal52 was collected in Palmira, Department of Valle del Cauca, from one of the sentinels’ experiments. The severity scale (1–5) for CFSD root symptoms used in this work was designed using the susceptible variety “Reina”, and is described as follows: 1 = Healthy (none of the roots showed symptoms); 2 = Lips in at least one section of one root; 3 = Lips in more than one section of two or more roots; most sections of each root remain without symptoms; 4 = More than one root totally covered with lips, forming a honeycomb-like appearance; the root size and thickness is not significantly changed. 5 = High density of lip rows resembling a honeycomb-like appearance covering the whole root, reduced storage root thickness, and formation of numerous fibrous roots. Total DNA and RNA from leaf samples was extracted following our standard protocol, without DNAse treatment needed before cDNA synthesis50. For total nucleic acid extraction from roots, we proceeded by harvesting them and washing them in water and in ethanol 70%. Using a kitchen peeler tool, we collected the tissue just below the root skin, which contains the peripheral phloem tissue. The slices were cut into smaller pieces and collected in 2 mL tubes for liquid nitrogen treatment. Approximately 300 mg of fresh root tissue were collected per root sample. After grinding with a plastic mini-pestle, we added CTAB and continued the protocol as described in Jimenez et al.50. High-throughput sequencing and bioinformatic analysis For complete virus genome assembly, total RNA was extracted from silica-dried leaves of cassava collected in Colombia (Supplementary Table 3), using Guanidine Thiocyanate 5 M for initial extraction and the RNeasy Plant Mini Kit (Qiagen) for purification. Libraries were prepared using TruSeq Stranded Total RNA with Ribo- Zero Plant kit, and sequenced on an Illumina NovaSeq 6000 platform (Psomagen, USA) to generate 6G (~ 40 M reads) of PE150 reads per sample. High-throughput sequencing (HTS) raw data was trimmed with BBDuk Adapter/Quality Trimming v38.84 (Brian Bushnell - DOE Joint Genome Institute) and de novo assembled using SPAdes v3.13.051 assembler. Contigs were filtered out using BLASTn (e-value < 1e − 25) to remove those matching cassava DNA (GenBank: GCF_001659605.2). Non-host contigs (40–50%) were searched against the complete NCBI non-redundant (nr) protein database (2023-07-28) using BLASTx (e-value < 1e − 25). Viral contigs were further analyzed using Geneious® Prime software v2023.1.2 (Biomatters, New Zealand). For microbial profile analysis total nucleic acids extracted from roots were processed with Illumina NovaSeq 6000 platform (Novogene, USA). Libraries were prepared using shotgun metagenomics (DNA-seq) and whole transcriptome sequencing (RNA-seq) protocols to generate 6G and 12G of PE150 reads per sample, respectively. Additionally, total RNA from root samples were sent for small interfering RNA sequencing (siRNA-seq)34 using an Illumina NextSeq 500 platform (Fasteris, Switzerland) with the small RNA-Seq gel free protocol, to produce SE75 sequences. DNA and RNA-seq raw data was quality controlled with FastQC v0.11.952 and filtered with Trimmomatic v0.3953 to remove adapter and Q > 30 sequences. Post-QC DNA and RNA-seq reads were aligned against the cassava genome (GenBank: GCF_001659605.2) to filter out host reads using BWA v0.7.1754 and STAR v2.7.11a55, respectively. SAMtools v1.3.156 performed sam-bam conversion from which non-host reads were retained. Non-host reads were subsequently assembled with MegaHIT v1.2.957, and taxonomically identified using Kraken2 v2.1.258,59 with the NCBI nucleotide (nt) database (2023-05-02). Subsequent taxonomic analyses were performed with phyloseq package60 in R project v4.2.161. Furthermore, siRNA-seq data was filtered by size, keeping reads of 18–28 bp. De novo assembly was then performed with Velvet v1.2.1062 using kmer values from 15 to 25 at steps of 2 (hash_length: 15,25,2). Contigs produced by each assembly were taxonomically inspected with DIAMOND v0.9.2463 by performing a DIAMOND-BLASTx (e-value < 1e − 25) against the nr database (2023-07-28). Additionally, a targeted search for cassava viruses (AH015299.2, KC505249.1, KC505250.1, KC505251.1, KC505252.1, NC_001658.1) and the newly assembled CsTLV-2 sequences was performed using BLASTn (e-value < 1e − 25, length ≥ 30 bp, identity ≥ 40%). From the de novo assembly of non-host RNA-seq data, contigs identified as CsTLV were inspected in order to recover the complete genome. Subsequently, an alignment of the post-QC RNA-seq reads against each genome segment (RNA1 and RNA2) of CsTLV was performed using BWA. The sam-bam conversion was performed with SAMtools and the mean sequencing depth was calculated with Qualimap v2.2.2a64. Finally, phylogenetic trees were constructed with MEGA v.665 based on the amino acid (aa) sequences of the protease-polymerase (Pro-Pol) region (RNA1) of torradovirus sequences available in GenBank (last accessed in August 2024) and the sequences obtained in this work. Trees were re-constructed using the Neighbor-Joining parameters with 1000 bootstrap replications and distances were calculated using the Poisson model for aa. Molecular diagnostics All plants were monitored for the presence of viruses and phytoplasma one month before harvest and then every three months after being transferred to an insect-proof screenhouse. For all RT-PCR assays, as the internal cDNA synthesis control, we used the nad5 (NADH dehydrogenase subunit 5) gene with primers nad5-s (5’- G A T G C T T C T T G G G G C T T C T T G T T-3’) and nad5-as (5’- C T C C A G T C A C C A A C A T T G G C A T A A-3’), as indicated in Menzel et al.66. For RNA viruses, we used PCR primers as described in Pardo et al.12 except for primers targeting torradoviruses. Newly designed diagnostic primers for isolates of CsTLV species 1 and 2, were developed from a multiple alignment that included all CsTLV genomes available in GenBank plus the ones assembled in this work. Primers for species 1 CsTLV1_RdRp-4042_F (5’- C T G C C T C T C A A G A A G A T T T A- 3’) and CsTLV1_RdRp_4657_R (5’- C A C T T C A G C T A T G C G A G G-3’) and for species 2 CsTLV2_Ham_6563_F (5’-AATRCATACGTGCACTCCCA-3’) and CsTLV2_Ham_7149_R (5’- A T T T C T T T A C A G A C A A G G T G-3’) were designed to amplify a fragment from RNA 1 of 633 (RdRp region) and 606 bp (Ham region), respectively. PCR programs were carried out in a Mastercycler Nexus® as follows: initial denaturation at 94  °C for 5 min followed by 30 cycles of denaturation at 94 °C x 30 s, annealing at 57 °C x 40 s and extension at 72 °C x 40 s. After Scientific Reports | (2024) 14:29648 9| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ http://www.nature.com/scientificreports a final extension at 72 °C x 10 min. PCR products were resolved by agarose gel-electrophoresis. All PCR products sequences were confirmed using Oxford Nanopore Technology (ONT) (Oxford Nanopore Technologies, UK). The occurrence of phytoplasma 16SrIII-L was checked using nested PCR with primers P1/Tint followed by R16F2n/R16R225,67, and LAMP using a rotor gene protocol as described33. Field surveys and disease evaluation To validate the results obtained from the sentinel plants (i.e., whether CsTLV can be associated with CFSD root symptoms), we inspected cassava fields of approximately eight-month-old plants located in the eastern plains (Department of Meta, province of Granada) and the north coast (Department of Córdoba, province of Monteria) of Colombia. In these locations, only the genotypes ‘Brasilera’ (Meta) and ‘Veronica’ (Córdoba) were surveyed for CFSD root symptoms. First, we calculated the incidence of the root symptoms by using a diagonal transect and 15 observations per ha. Leaf samples were collected from each plant and, in some cases, stems were also collected and transferred to an insect-proof screenhouse for further analyses. Leaf samples were dried in silica gel for at least three days, as previously described50, before they were examined for viruses and phytoplasma. In addition, different sets of historically collected leaf samples from plants with clear CFSD root symptoms were re- checked for CsTLV and other suspected pathogens. These samples included those collected from affected fields in the eastern plains of Colombia (Granada and Yopal) during the last recorded outbreak of CFSD with 100% incidence of root symptoms in this region (Supplementary Table 4). The identity of PCR products obtained in these tests was confirmed by ONT (Oxford Nanopore Technologies, UK) sequencing. Data availability The sequences reported in this study have been deposited in the National Center for Biotechnology Informa- tion (NCBI) database with BioProject PRJNA1151691 and BioSamples SAMN43322541 to SAMN43322549 and SAMN44378827 to SAMN44378831. The data presented in this study are available in the following Supplemen- tary Materials section. Received: 13 September 2024; Accepted: 25 November 2024 References 1. Manganyi, B., Lubinga, M. H., Zondo, B. & Tempia, N. Factors influencing cassava sales and income generation among cassava producers in South Africa. Sustainability 15(19), 14366 (2023). 2. Cruz, I. A. et al. Valorization of cassava residues for biogas production in Brazil based on the circular economy: An updated and comprehensive review. Clean. Eng. Technol. 4, 100196 (2021). 3. FAOSTAT. Rome: FAO. https://www.fao.org/faostat/en/#data/QCL/visualize Accessed 01/06/2024 to 02/08/2024 (2022). 4. Byju, G. & Suja, G. Mineral nutrition of cassava. Adv. Agron. 159, 169–235 (2020). 5. Marx, S. Cassava as feedstock for ethanol production: a global perspective. Bioethanol production from food crops. Academic press, 101–113 (2019). 6. Scott, G. J. A review of root, tuber and banana crops in developing countries: Past, present and future. Int. J. Food Sci. Technol. 56(3), 1093–1114 (2021). 7. Rivera, T., Labarta, R., Andrade, R., Calle, F., & Becerra López-Lavalle, L. A. Farm management and varietal choice in cassava- based production systems in Colombia. Publication CIAT No. 524. International Center for Tropical Agriculture (CIAT). Cali, Colombia (2021). 8. Fathima, A. A., Sanitha, M., Tripathi, L. & Muiruri, S. Cassava (Manihot esculenta) dual use for food and bioenergy: A review. Food Energy Secur. 12(1), e380 (2023). 9. Legg, J. P. et al. Cassava virus diseases: Biology, epidemiology, and management. Adv. Virus Res. 91, 85–142 (2015). 10. Siriwan, W. et al. Surveillance and diagnostics of the emergent Sri Lankan cassava mosaic virus (Fam. Geminiviridae) in Southeast Asia. Virus Res. 285, 197959 (2020). 11. Pineda López, B., Jayasinghe, W. U. & Lozano, J. C. La enfermedad cuero de sapo en yuca (Manihot esculenta Crantz). ASIAVA 4, 10–12 (1983). 12. Pardo, J. M. et al. Cassava frogskin disease: Current knowledge on a re-emerging disease in the Americas. Plants 11, 1841 (2022). 13. Thomas-Sharma, S. et al. Seed degeneration in potato: the need for an integrated seed health strategy to mitigate the problem in developing countries. Plant Pathol. 65(1), 3–16 (2016). 14. Shirima, R. R. et al. Assessing the degeneration of cassava under high-virus inoculum conditions in coastal Tanzania. Plant Dis. 103(10), 2652–2664 (2019). 15. Bass, D., Stentiford, G. D., Wang, H. C., Koskella, B. & Tyler, C. R. The pathobiome in animal and plant diseases. Trends Ecol. Evol. 34(11), 996–1008 (2019). 16. Leiva, A. M. et al. Ceratobasidium sp. is associated with cassava witches’ broom disease, a re-emerging threat to cassava cultivation in Southeast Asia. Sci. Rep. 13(1), 22500 (2023). 17. Cock, J. H., & Connor, D. J. Cassava. Crop physiology case histories for major crops Academic Press, Belin. 588–633 (2021). 18. Calvert, L. A., Ospina, M. D. & Shepherd, R. J. Characterization of cassava vein mosaic virus: A distinct plant pararetrovirus. J. Gen. Virol. 76(5), 1271–1278 (1995). 19. Costa, A.S., & Kitajima, E.W. Cassava common mosaic virus. C.M.I/A.A.B. Descriptions of Plant Viruses 90 (1972). 20. Collavino, A., Zanini, A. A., Medina, R., Schaller, S. & Di Feo, L. Cassava common mosaic virus infection affects growth and yield components of cassava plants (Manihot esculenta) in Argentina. Plant Pathol. 71(4), 980–989 (2022). 21. Cuervo, M. Caracterización de los ácidos nucleicos de doble cadena asociados a enfermedades similares a las ocasionadas por virus en yuca (Manihot esculenta Crantz). Fitopatol. Colomb. 14, 10–17 (1990). 22. Calvert, L. A., Cuervo, M., Lozano, I., Villareal, N. & Arroyave, J. Identification of three strains of a virus associated with cassava plants affected by frogskin disease. J. Phytopathol. 156, 647–653 (2008). 23. Matthijnssens, J. et al. ICTV virus taxonomy profile: Sedoreoviridae 2022. J. Gen. Virol. 103(10), 001782 (2022). 24. Alvarez, E. et al. Characterization of a phytoplasma associated with frogskin disease in cassava. Plant Dis. 93, 1139–1145 (2009). 25. de Oliveira, S. A. S. D. et al. First report of a 16SrIII-L phytoplasma associated with frogskin disease in cassava (Manihot esculenta Crantz) in Brazil. Plant Dis. 98(1), 153–153 (2014). 26. de Souza, A. N., da Silva, F. N., Bedendo, I. P. & Carvalho, C. M. A phytoplasma belonging to a 16SrIII-A subgroup and dsRNA virus associated with cassava frogskin disease in Brazil. Plant Dis. 98(6), 771–779 (2014). Scientific Reports | (2024) 14:29648 10| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ https://www.fao.org/faostat/en/#data/QCL/visualize http://www.nature.com/scientificreports 27. Cardozo, L., Pardo, J. M., Zacher, M., Torres, A. & Alvarez, E. First report of a 16SrIII phytoplasma associated with frogskin disease in cassava (Manihot esculenta) in Paraguay. Plant Dis. 100, 1492 (2016). 28. Carvajal-Yepes, M. et al. Unraveling complex viral infections in cassava (Manihot esculenta Crantz) from Colombia. Virus Res. 186, 76–86 (2014). 29. de Oliveira, S. A. S. et al. First report of cassava torrado-like virus, cassava polero-like virus and cassava new alphaflexivirus associated with cassava frogskin disease in Brazil. J. Plant Pathol. 102, 247–247 (2020). 30. Chaparro-Martinez, E. I. & Trujillo-Pinto, G. First report of frog skin disease in cassava (Manihot esculenta) in Venezuela. Plant Dis. 85(12), 1285–1285 (2001). 31. Lovell-Read, F. A., Parnell, S., Cunniffe, N. J. & Thompson, R. N. Using ‘sentinel’ plants to improve early detection of invasive plant pathogens. PLoS Comput. Biol. 19(2), e1010884 (2023). 32. Leiva, A. M., Jimenez, J., Sandoval, H., Perez, S. & Cuellar, W. J. Complete genome sequence of a novel secovirid infecting cassava in the Americas. Arch. Virol. 167, 1–4 (2022). 33. Cuervo, M. et al. Manual de procedimientos del laboratorio de sanidad de germoplasma. Certificación sanitaria del germoplasma de yuca. Palmira, Colombia: Centro Internacional de Agricultura Tropical (CIAT), Programa Recursos Genéticos. 2da. Edición. 63 p. https://hdl.handle.net/10568/49635 (2017). 34. Kreuze, J. F. et al. Complete viral genome sequence and discovery of novel viruses by deep sequencing of small RNAs: A generic method for diagnosis, discovery and sequencing of viruses. Virology 388(1), 1–7 (2009). 35. Lozano, I. et al. Resolution of cassava-infecting alphaflexiviruses: Molecular and biological characterization of a novel group of potexviruses lacking the TGB3 gene. Virus Res. 241, 53–61 (2017). 36. Jimenez, J. Caracterización del Virus Cassava Torrado-like Virus en Yuca (Manihot esculenta Crantz) e Identificación de Proteínas Virales Involucradas en la Supresión del Silenciamiento de ARN. Master’s Thesis, Universidad Nacional de Colombia, Palmira, Colombia. (2017). 37. Jimenez, J., Leiva, A. M. & Cuellar, W. J. Cassava torrado-like virus encodes a gene that facilitates the mechanical transmission to Nicotiana benthamiana of Cassava virus X. Sci. Agropecu. 14(2), 213–221 (2023). 38. van der Vlugt, R. A. et al. Torradoviruses. Ann. Rev. Phytopathol. 53(1), 485–512 (2015). 39. Tomlinson, K. R. et al. Cassava brown streak virus Ham1 protein hydrolyses mutagenic nucleotides and is a necrosis determinant. Mol. Plant Pathol. 20(8), 1080–1092 (2019). 40. Pardo, J. M., Truke, M. J., Alvarez, E., Cardozo, L., & Varela, I. A real-time PCR assay to detect and quantify 16SRIII-L and phytoplasmas associated with cassava frogskin disease in Costa Rica and Paraguay. In Proceedings of the 50th Annual meeting Caribbean Food Crops Society—Enhancing Family Farms through Sustainable Energy, Research and Technology, St. Thomas, VI, USA, 7–11 July 2014. 41. Landicho, D. M. et al. Status of cassava witches’ broom disease in the Philippines and identification of potential pathogens by metagenomic Analysis. Biology 13(7), 522 (2024). 42. Gibson, R. W. & Kreuze, J. F. Degeneration in sweetpotato due to viruses, virus-cleaned planting material and reversion: A review. Plant Pathol. 64(1), 1–15 (2015). 43. Verbeek, M., Tang, J. & Ward, L. I. Two generic PCR primer sets for the detection of members of the genus Torradovirus. J. Virol. Meth. 185(2), 184–188 (2012). 44. Gil-Ordoñez, A. et al. Isolation, genome analysis and tissue localization of Ceratobasidium theobromae, a new encounter pathogen of cassava in Southeast Asia. Sci. Rep. 14, 18139 (2024). 45. Verbeek, M., van Bekkum, P. J., Dullemans, A. M. & van der Vlugt, R. A. Torradoviruses are transmitted in a semi-persistent and stylet-borne manner by three whitefly vectors. Virus Res. 186, 55–60 (2014). 46. Lecoq, H. et al. Characterization and occurrence of squash chlorotic leaf spot virus, a tentative new torradovirus infecting cucurbits in Sudan. Arch. Virol. 161, 1651–1655 (2016). 47. Rozado-Aguirre, Z. et al. Detection and transmission of Carrot torrado virus, a novel putative member of the Torradovirus genus. J. Virol. Methods 235, 119–124 (2016). 48. Verbeek, M., Dullemans, A. M. & van der Vlugt, R. A. Aphid transmission of Lettuce necrotic leaf curl virus, a member of a tentative new subgroup within the genus Torradovirus. Virus Res. 241, 125–130 (2017). 49. Wintermantel, W. M., Hladky, L. L. & Cortez, A. A. Genome sequence, host range, and whitefly transmission of the torradovirus Tomato necrotic dwarf virus. Acta Hortic. 1207, 295–302 (2018). 50. Jimenez, J., Leiva, A. M., Olaya, C., Acosta-Trujillo, D. & Cuellar, W. J. An optimized nucleic acid isolation protocol for virus diagnostics in cassava (Manihot esculenta Crantz.). MethodsX 8, 101496 (2021). 51. Bankevich, A. et al. SPAdes: A new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19(5), 455–477 (2012). 52. Brown, J., Pirrung, M. & McCue, L. A. FQC dashboard: Integrates FastQC results into a web-based, interactive, and extensible FASTQ quality control tool. Bioinformatics 33(19), 3137–3139 (2017). 53. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics 30(15), 2114– 2120 (2014). 54. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25(14), 1754–1760 (2009). 55. Dobin, A. et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 29(1), 15–21 (2013). 56. Li, H. et al. The sequence alignment/map format and SAMtools. Bioinformatics 25(16), 2078–2079 (2009). 57. Li, D. et al. MEGAHIT v1.0: A fast and scalable metagenome assembler driven by advanced methodologies and community practices. Methods 102, 3–11 (2016). 58. Wood, D. E. & Salzberg, S. L. Kraken: Ultrafast metagenomic sequence classification using exact alignments. Genome Biol. 15, 1–12 (2014). 59. Wood, D. E., Lu, J. & Langmead, B. Improved metagenomic analysis with Kraken 2. Genome Biol. 20, 1–13 (2019). 60. McMurdie, P. J. & Holmes, S. phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PloS One 8(4), e61217 (2013). 61. RStudio Team. RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL http://www.rstudio.com/ (2020). 62. Zerbino, D. R. & Birney, E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 18(5), 821–829 (2008). 63. Buchfink, B., Xie, C. & Huson, D. H. Fast and sensitive protein alignment using DIAMOND. Nat. Methods 12(1), 59–60 (2015). 64. García-Alcalde, F. et al. Qualimap: evaluating next-generation sequencing alignment data. Bioinformatics 28(20), 2678–2679 (2012). 65. Tamura, K., Stecher, G., Peterson, D., Filipski, A. & Kumar, S. MEGA6: Molecular evolutionary genetics analysis version 6.0. Mol. Biol. Evol. 30(12), 2725–2729 (2013). 66. Menzel, W., Jelkmann, W. & Maiss, E. Detection of four apple viruses by multiplex RT-PCR assays with coamplification of plant mRNA as internal control. J. Virol. Methods 99(1–2), 81–92 (2002). 67. Pardo, J. M. et al. Cassava witches’ broom disease in Southeast Asia: A review of its distribution and associated symptoms. Plants 12(11), 2217 (2023). Scientific Reports | (2024) 14:29648 11| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ https://hdl.handle.net/10568/49635 http://www.rstudio.com/ http://www.nature.com/scientificreports Acknowledgements This work was funded by the U.S. Agency for International Development (USAID) with support from CGIAR’s Plant Health (PHI) and Accelerated Breeding (ABI) initiatives. We thank Viviana Dominguez and Ronald Mon- tes for laboratory and greenhouse technical support. Thanks to Sandra Salazar, Camilo Vargas, Sean Fenstemak- er and Xiaofei Zhang from CIAT’s Cassava Breeding team, for their help with field activities, to Hector Sandoval from Agrosavia for providing samples from Meta and Arauca and to Cristian Olaya for technical advice with RNA extraction. Special thanks to Maritza Cuervo and Diana Niño for their help with detection of phytoplas- ma. We thank Luis Augusto Becerra, Jonathan Newby and Joe Tohme for supporting this study. The authors acknowledge Vincent Johnson (CIAT Science Writing Service) for copy editing this paper. Author contributions J.J performed field work, molecular diagnostics, virus genome characterization and phylogenetic analysis. S.C. and J.M.P. performed field and greenhouse work, molecular diagnostics and agronomic data analysis. J.J. and S.C performed nanopore sequencing of PCR products. A.G.-O. performed bioinformatic analysis from DNA- seq and RNA-seq datasets from root samples. J.J.; S.C.; A.G-O.; J.M.P. and W.J.C. prepared figures and tables. J.M.P. carried out field data curation and evaluation of disease. R.A.-Q and D.M. carried out and supervised sequencing analysis of CsTLV genomes from leaf samples. W.J.C. designed research, supervised laboratory and field work and wrote the main manuscript text. All authors reviewed, contributed, and agreed with the final version of the manuscript. Declarations Competing interests The authors declare no competing interests. Ethical approval The authors declare no competing interests. Additional information Supplementary Information The online version contains supplementary material available at h t t p s : / / d o i . o r g / 1 0 . 1 0 3 8 / s 4 1 5 9 8 - 0 2 4 - 8 1 1 4 2 - 2 . Correspondence and requests for materials should be addressed to W.J.C. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit h t t p : / / c r e a t i v e c o m m o n s . o r g / l i c e n s e s / b y - n c - n d / 4 . 0 / . © The Author(s) 2024 Scientific Reports | (2024) 14:29648 12| https://doi.org/10.1038/s41598-024-81142-2 www.nature.com/scientificreports/ https://doi.org/10.1038/s41598-024-81142-2 https://doi.org/10.1038/s41598-024-81142-2 http://creativecommons.org/licenses/by-nc-nd/4.0/ http://creativecommons.org/licenses/by-nc-nd/4.0/ http://www.nature.com/scientificreports Single torradovirus infections explain the mysterious cassava frogskin disease in the Americas Results Low incidences of root symptoms and single infections detected in sentinel plants after eight months Metagenomics confirms torradoviruses in single-infected symptomatic roots while CsFSaV and phytoplasma were not detected Two species of torradovirus infecting cassava plants with CFSD in Colombia Improved diagnostics shows the association of torradoviruses with CFSD in farmers’ fields Discussion Materials and methods Plant material and nucleic acid extraction High-throughput sequencing and bioinformatic analysis Molecular diagnostics Field surveys and disease evaluation References