http://orcid.org/0000-0001-8561-4172
Figures
Abstract
In Sub-Saharan Africa cassava (Manihot esculenta Crantz) is one of the most important food crops where more than 40% of the population relies on it as their staple carbohydrate source. Biotic constraints such as viral diseases, mainly Cassava Mosaic Disease (CMD) and Cassava Brown Streak Disease (CBSD), and arthropod pests, particularly Cassava Green Mite (CGM), are major constraints to the realization of cassava’s full production potential in Africa. To address these problems, we aimed to map the quantitative trait loci (QTL) associated with resistance to CBSD foliar and root necrosis symptoms, foliar CMD and CGM symptoms in a full-sib mapping population derived from the genotypes AR40-6 and Albert. A high-density linkage map was constructed with 2,125 SNP markers using a genotyping-by-sequencing approach. For phenotyping, clonal evaluation trials were conducted with 120 F1 individuals for two consecutive field seasons using an alpha-lattice design at Chambezi and Naliendele, Tanzania. Previously identified QTL for resistance to CBSD foliar symptoms were corroborated, and a new putative QTL for CBSD root necrosis identified (qCBSDRNc14AR) from AR40-6. Two QTL were identified within the region of the previously recognized CMD2 locus from this population in which both parents are thought to possess the CMD2 locus. Interestingly, a minor but consistent QTL, qCGM18AR, for CGM resistance at 3 months after planting stage was also detected and co-localized with a previously identified SSR marker, NS346, linked with CGM resistance. Markers underlying these QTL may be used to increase efficiencies in cassava breeding programs.
Citation: Garcia-Oliveira AL, Kimata B, Kasele S, Kapinga F, Masumba E, Mkamilo G, et al. (2020) Genetic analysis and QTL mapping for multiple biotic stress resistance in cassava. PLoS ONE 15(8): e0236674. https://doi.org/10.1371/journal.pone.0236674
Editor: Peng Zhang, Shanghai Institutes for Biological Sciences, CHINA
Received: April 9, 2020; Accepted: July 9, 2020; Published: August 5, 2020
Copyright: © 2020 Garcia-Oliveira et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This work was funded by the Bill and Melinda Gates Foundation (BMGF) grant no. OPPGD1016 to MF, IITA. Work by JB and JL was supported by BMGF grant OPPGD1493 to Dr. Steve Rounsley, who at the time was at the University of Arizona. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Cassava (Manihot esculenta Crantz) is widely grown within the tropics of Africa, Asia, and Latin America where it is consumed as a food and utilized as an industrial crop primarily for starch [1]. The broad adaptation, resilience, and excellent source of calories makes cassava a popular choice for African farmers, but its productivity in Africa is far below that of its potential and that achieved in countries outside of the continent. Biotic stresses inflict a large yield penalty and impact the availability of clean planting material [2–4]. These stresses include viral diseases and insect pests, particularly Cassava Mosaic Disease (CMD), Cassava Brown Streak Disease (CBSD), and Cassava Green Mite (CGM) (Mononychellus tanajoa) [4,5]. Although yield losses due to individual biotic constraints vary from year to year, the combined effect of diseases and pests implies that there is potential for near total crop failure under field conditions [6,7]. Viruses that cause CMD and CBSD are spread by the whitefly vector, Bemisia tabaci, and via vegetative stem cuttings. From the 1930s, after the first detailed report on viruses that cause disease in cassava [8], about 15 specific virus species and several strains have been identified in Africa that cause substantial yield losses in different cassava producing regions [4].
Compared to CMD, CBSD received comparatively little attention until the end of 20th century. This is partly because it was endemic to a relatively restricted area in the coastal lowlands of Kenya, Tanzania, and in areas along Lake Malawi in Mozambique and Malawi [9–11]. However, in the past decade it has also been reported at higher altitudes and inland of countries surrounding the Great Lakes regions, including Burundi [12], Rwanda [13], Uganda [7], and eastern DR Congo [14], as well as further afield such as in southern Sudan [15] and the islands off East Africa, including Mayotte Island [16]. In heavily affected regions, farmers have been forced to abandon cassava production [12,14,17], with significant reductions in storage root quality observed in susceptible cultivars and losses of 70% or more [18,19]. The spread of CBSD from East to West Africa is considered a major threat to food security [20].
Breeding for disease resistance continues to be one of the most reliable, sustainable, and cost-effective measures to overcome multiple disease and pest stresses [21]. The first attempts to select for cultivars resistant to CBSD were during the 1930’s at the former Amani Research Station (Tanzania). Studies on CBSD inheritance and gene action indicate that CBSD resistance is multigenic and recessive [22]. Using Mozambican parental lines, Zacarias [23] suggested that non-additive gene effects contribute to the control of CBSD resistance, results that diverged from ones obtained previously by Jennings [24], and later by Kulembeka et al. [25], that showed CBSD resistance in Tanzanian varieties seemed to be due to additive gene effects. Studies to identify QTL associated with resistance to CBSD foliar symptoms and root necrosis have been conducted using biparental populations derived from Tanzanian landraces [26,27] and a Ugandan breeding population [28]. Here we utilized a source of germplasm from South America where additional sources of resistance, and even immunity, to CBSD have recently been identified [29].
CMD is widely distributed and one of the most notorious viral diseases of cassava in Africa, resulting in an estimated annual loss of 34 million tonnes [30]. The disease is best kept under control by the deployment of resistant cultivars [31]. Following the establishment of the Root and Tuber Improvement Program in 1971 at International Institute of Tropical Agriculture (IITA), Ibadan, Nigeria, tremendous progress has been made in producing improved cassava cultivars with resistance to CMD [32]. Among the well-known sources of resistance to CMD (CMD1, CMD2 and CMD3), CMD2 has been extensively utilized in cassava breeding programmes in Africa and recently introduced in Asia particularly India and Thailand [31–33]. Despite numerous studies, causative gene(s) of the widely recognized CMD2 locus [34] have not been identified. The exact nature of this resistance, whether monogenic [35, 36], polygenic or multi-allelic [26,37] or epigenetic [38] is still under debate, although significant new knowledge on the genomic organization of this locus has been provided by Kuon et al. [39]. Additional loci of smaller effect including CMD1 [40] and CMD3 explaining 11% of the phenotypic variance [41] have been recognized in different germplasm sources.
CGM is a notorious arthropod pest that seriously affects the cassava canopy (ability to stay green), particularly in dry seasons. CGM was originally described on cassava in Brazil [42], but occurs in almost all cassava growing regions, particularly in Brazil and Africa, where it causes losses of up to 51% [43,44]. Biological control and deployment of host-plant resistance are important control measures against CGM [45, 46]. Typhlodromalus aripo, a predatory mite introduced from Brazil, is the most effective agent for biological control of CGM in Africa [47], although its effectiveness is sometimes compromised by a failure to establish properly, either because of a lack of a suitable host cassava genotype and/or climatic conditions [48]. In the past, breeding efforts have mainly focused on indirect selection against CGM through selection for associated morphological traits such as pubescent leaves (PL), large compact shoot apices, enhanced leaf retention (LR) and stay green (SG) characteristics [49,50]. Little progress in understanding the inheritance of CGM resistance has been achieved through field studies, although additive gene action is thought to be important [48,51]. Recently, QTL for PL and LR were found within a 4- to 8-MB region on chromosome 8, and for PL and SG on chromosome 12. These co-localized with QTL for CGM severity at 6 MAP [50,52]. Using bulk segregant analysis (BSA), Macea-Choperena et al. [53] identified SSR markers NS1099 and NS346 as having the highest association with CGM resistance in individuals of different families.
In the present study, we aimed to construct a high-density linkage map through SNP genotyping and map QTL for field resistance to CBSD foliar and root necrosis, and CMD and CGM foliar symptoms using a full-sib population derived from a cross between an elite genotype, AR40-6 (source of CBSD, CMD and CGM resistance), and a Tanzanian local variety, Albert (CBSD and CGM susceptible and CMD resistant).
Materials and methods
Generation of mapping population
Among the cassava genotypes previously screened for viral diseases in Tanzania, AR40-6 was identified as having high levels of resistance to CMD, CBSD, and CGM and was thus selected for QTL mapping. AR40-6 was bred at the International Centre of Tropical Agriculture (CIAT), Colombia and has approximately 12.5% hybrid genome from wild species M. esculenta ssp. flabellifolia, 0.3% M. glaziovii–M. esculenta introgression segments [54], 50% CMD resistant variety C39 from IITA [55], 6.25% of Thai origin from Rayong 1 (via Rayong 60) [56] and the remainder approximately 30.25% of South American origin. AR40-6 was positively selected for the CMD2 locus at CIAT using SSR markers. In contrast, Albert is a locally adapted variety popular in South-eastern Tanzania and is susceptible to CBSD and CGM but has good agronomic characters and resistance to CMD [57, 58]. It is a full-sib of TME117 [54] and was found to possess two putative QTL within the CMD2 locus by Masumba et al. [26] in a bi-parental mapping population with the landrace Namikonga. To generate the full-sib F1 mapping population, crosses between AR40-6 (female) and Albert (male) were performed at Naliendele, Tanzania during 2009 according to a standard protocol [59]. A total of 1,015 F1 seeds were obtained and, after a flotation test to eliminate poor quality seed, they were germinated in seed trays on a bench under screen house conditions at the Tanzanian Agricultural Research Institute (TARI), Kibaha, Tanzania. The integrity of the F1 seedlings was tested using SSR markers. For clonal multiplication, only true F1 seedlings were planted in a low disease pressure area, isolated from other cassava plants at Makutupora, Tanzania in March 2010.
Field evaluation for disease and pest resistance/tolerance
A total of 120 F1 individuals were phenotyped in clonal evaluation trials (CET) over two years (2013 and 2014) at two locations namely; Chambezi (5°54’S and 35°57’E at an altitude of 39m above sea level (asl)) and Naliendele (10°23’S and 40°09’E at an altitude of 137m asl) in Tanzania. Further details of agro-climatic and soil conditions are given in S1 Fig. An alpha lattice design was used with two replications. Each genotype was grown in single-row plots of five plants with 1 × 1 m spacing (row-to-row and plant-to-plant). To augment disease pressure, the trials were planted in December which coincides with the short rains and extremely high CBSD pressure in Chambezi [60]. In addition, stakes taken from plants close to the experimental sites that clearly exhibited CMD and CBSD symptoms were planted around the evaluation trials to increase disease pressure. Sufficient replication within trial, across sites and years was also used to reduce the effect of any escape from disease challenge. Foliar symptoms of CBSD, CMD, and CGM were evaluated on a subjective scale of 1 (no symptoms) to 5 (severe leaf symptoms with die back in the case of CBSD) at 3 and 6 months after planting (MAP) [26,59]. For CBSD root necrosis symptoms, roots were harvested 12 MAP and necrosis was scored both qualitatively on a 1 (no visible root necrosis) to 5 (very severe necrosis of affected roots) scale [26,61] and quantified through analysis of photographs using ImageJ software (http://rsbweb.nih.gov/ij/).
Phenotypic data was analysed by performing analysis of variance (ANOVA) using the General Linear Model (Proc GLM) procedure of SAS software (SAS Institute, Cary, NC, USA) including environment (e), genotype (g), genotype by environment interactions (gxe), and replication effects (r) in the model, and from which the components of variance were estimated. Broad sense heritability (h2bs) across environments was computed using the formula; h2bs = σ2g + σ2ge/e + σ2 e/er [62] and classified as suggested by Johnson and Robinson [63].
Genotyping, data analysis and construction of a genetic linkage map
Total genomic DNA was extracted from young leaf tissue for all the F1 full sib family individuals and both parents, using a modified Dellaporta et al. [64] method. Genotyping-by-sequencing (GBS) and single nucleotide polymorphism (SNP) calling was carried out at the University of California, Berkeley and used to develop the consensus map reported by ICGMC [65]. In that study SNPs were called against v4.1 of the draft cassava reference genome sequence but hereere SNPs were re-called against version 5.1. A series of quality checks were performed on the data before calculating the linkages. Markers with InDels, tri- and tetra-allelic loci, loci with identical segregation patterns and any markers with more than 20% missing values, were removed before generating linkage groups (LGs). For JoinMap®4.1, SNP marker data were coded according to the CP option in JoinMap® 4.1 manual for outcrossing species [66]. The threshold of fit was set to be >5.0 with groups defined according to the independence LOD and a recombination frequency <0.4. Segregation distortion was tested using the chi-square goodness-of-fit test in JoinMap. The marker order and distances were calculated using the Maximum Likelihood Mapping function with default settings and a genetic map was constructed using the Kosambi mapping function [67].
QTL analysis
QTL analysis was performed on the mean scores of both replications for CBSD foliar (3 and 6 MAP), CBSD root necrosis, CMD foliar (3 and 6 MAP) and CGM at each environment using the QTL mapping software package Genetic Analysis of Clonal F1 and Double cross (GACD) version 1.1 [68]. Based on the actual number of identified alleles in the two parents and the actual number of identifiable genotypes in the clonal F1 progenies, each marker locus was classified into four categories as described for GACD [68]. Four marker categories i.e., ABCD, A = B, C = D and AB = CD were assigned with the assumption that A and B are two alleles from the female parent at the marker locus, and C and D are two alleles from the male parent at the same locus. Category ABCD represents polymorphism in both parents whereas categories A = B and C = D represent only female and male polymorphism, respectively. On the other hand, category AB = CD represents the same polymorphism pattern in both female and male parents, similar to an F2 population derived from two inbred parents in self-pollinated and cross-pollinated species. Any missing values of marker type were coded as XX. In order to declare QTL significance, LOD thresholds were determined using the empirical formula LOD = χ2αp(df)/2ln as described by Yu et al. [69]. Results are from the .QIC file which reports an adjusted phenotypic variance explained (PVE). A QTL was considered a major QTL if it accounts for more than 10% phenotypic variance [70].
Gene annotation
To identify putative genes underlying the major consistent QTL, the position of markers of interest were first converted from v5.1 to v6.1 on the M. esculenta reference genome (http://phytozome.jgi.doe.gov/pz/portal.html). All the predicted genes in the specific QTL region were extracted from the physical map. GO enrichment was performed by Singular Enrichment Analysis (SEA) using the Comparative Arabidopsis Genome Resource (https://www.arabidopsis.org/). Over- and under-represented annotation-enrichment GO terms were extracted with default parameters and only pathways with p<0.5 values were considered.
Results
One thousand and fifteen F1 seeds were sown, however a germination rate of 21.5% left 218 seedlings for establishment in the multiplication block. Of these, 18 and 41 seedlings were identified by SSR markers as self-fertilized and off-types, respectively. After evaluation, 159 true F1 genotypes including both parents were chosen for genotyping, but a high level of missing data in five samples reduced the number to 154.
Disease and pest traits assessment, and their frequency distribution
Typical symptoms of all the investigated traits were observed adequately among the 120 F1 clones planted in both years and locations in Tanzania. Analysis of variance (ANOVA) for phenotypic traits showed a significant difference for the environmental effect indicating the presence of substantial environmental variance (S1 Table). Higher mean scores of CBSD foliar symptoms at 3 and 6 MAP were observed in both years in Chambezi compared to Naliendele (Fig 1). Similarly, higher root necrosis was observed throughout 2013 at Chambezi, but no significant difference was observed for both mean score and per cent root necrosis between the locations in 2014. Similar levels of CMD and CGM were observed across locations throughout the growing period, except for CGM in Naliendele, where more severe damage was consistently observed at 6 MAP in both years. F1 clone reactions to each of CBSD, CMD, and CGM were strongly correlated between years at each location, whereas correlations between Chambezi and Naliendele locations were poor (S2 Table). Some clones that scored 5 for CBSD root necrosis had die back that led to the death of the plant. High skewness and kurtosis values revealed that traits were not normally distributed (S3 Table). In general, the frequency distribution of reaction of the family against CBSD foliar and root necrosis, and foliar CMD and CGM showed significant skewing toward the resistant parental mean (Fig 1). The higher magnitudes of phenotypic coefficients of variation (PCV) compared to their genotypic coefficients of variation (GCV) for all traits clearly underscore the role of environmental factors on the expression of the studied traits (S3 Table). Moderate to high PCV and GCV were observed for all the studied traits. The broad sense heritability estimates for foliar CBSD, CMD, CGM at 3 and 6 MAP varied from 24.70%– 42.91%, 75.16%– 76.79% and 21.29%– 36.84% across the environments, respectively. Similarly, heritability estimates of 25.12% and 33.26% were observed for CBSD root necrosis if the traits were quantified on the basis of scale (1–5) or per cent root necrosis (S3 Table).
The foliar CBSD, CMD and CGM, and necrosis of root (RN), scores on a 1–5 scale from resistant to susceptible and root necrosis per cent (% area) are shown on the x-axis, and the number of genotypes (F1 families) with each score is plotted on the y-axis.
Construction of linkage map
A total of 3,491 high-quality SNPs were generated by GBS from 147 genotypes. Following the stringent filtering of polymorphic SNPs (≥20% missing values; ≤5% MAF), 41 loci were deleted because they segregated in exactly the same way as another marker; furthermore, one marker was discarded for being tetra-allelic and 18 were discarded for being tri-allelic which left a total of 2,125 SNP markers (S4 Table). Prior to building the final genetic linkage map, 17 genotypes that > 20% missing data were also deleted. The resulting genetic map consisted of 2,125 markers on 18 linkage groups, representing the 18 chromosomes of cassava (2n = 36) with a length of 1730 cM (Fig 2; S4 Table; S2 Fig). Linkage group (LG) length varied from 49.8 cM (Chr. XII) to 121.9 cM (Chr. VI), with an average marker interval length between 0.45 cM to 1.71 cM per chromosome. The highest number of markers was mapped on chromosome VI (161), while the lowest number of markers was mapped to chromosome XVI (45). Overall, the map contained an average of 118 markers per linkage group with an average marker interval length of 0.81 cM.
Horizontal bars on each chromosome represent mapped SNP markers and the scale bar to the right shows the lengths of linkage groups (Chromosomes) as measured in Kosambi cM. A more detailed map with marker names (Manihot esculenta v5.1 reference genome) and their position is presented in S2 Fig.
QTL mapping
To define a genome-wide LOD significance threshold, LOD scores were computed by performing permutation analysis with 1000 iterations. This corresponded to a threshold of LOD 3.25 to 8.49 for the studied traits. The high LOD threshold values are the result of the non-normally distributed phenotypic data, which are highly skewed towards resistance. Therefore, a QTL with a LOD score above 3.0 was considered significant [71]. QTL detected in at least two of the four environments were considered to be consistent [72], because QTL × environment interaction effects were highly significant and no QTL was consistently identified in all the four environments except a QTL for CGM at 3 MAP stage. Common names were assigned to consistent QTL associated with a specific disease or pest if it was detected at different stages (3 MAP or 6 MAP) and mapped to the same, or at least overlapping, genomic regions/marker intervals. A total of 17 consistent QTL were detected on chromosomes V, VIII, IX, X, XI, XII, XIII, XIV, XVI, and XVIII that each explained from 1.49% to 29.03% of phenotypic variation for the studied traits. A summary of these QTL, including their positions, LOD scores, genetic effects [additive effects of female (aF) and male (aM), and dominance effect d], and the adjusted PVE are shown in Table 1. Favourable alleles at each of the flanking markers are given in S5 Table.
QTL for CBSD foliar symptoms
QTL analysis for CBSD foliar resistance at 3 MAP identified one minor but consistent QTL (qCBSDFc11AR) on chromosome XI accounting for 2.5% phenotypic variation (Table 1). The peak position of qCBSDFc11AR was mapped at 10 cM in Naliendele 2013 and 2014 environments. Similarly, one consistent but major QTL (qCBSDFc18AR), explaining 12.87% of the phenotypic variation for CBSD foliar resistance at 6 MAP stage, was identified on chromosome XVIII in both years in Chambezi (Table 1). None of the QTL consistent across years for CBSD foliar resistance were found to be consistent across both 3 and 6 MAP stages.
QTL for CBSD root necrosis
QTL analysis of CBSD root necrosis data obtained by two different methods revealed four consistent QTL on three different chromosomes (VIII, XIV, and XVIII) (Table 1). A single consistent QTL (qCBSDRNSc14AR), with LOD values ranging from 3.86 to 7.12 and contributions to phenotypic variance of 8.18%– 8.72% was detected for root necrosis when the trait was quantified on the basis of visual score (1–5 scale). Contrarily, when root necrosis was quantified as per cent root necrosis area using ImageJ software, three consistent QTL (qCBSDRNAc8AR, qCBSDRNAc14AR, and qCBSDRNAc18AR), accounting for 2.13% to 9.70% phenotypic variation, were detected at chromosome(s) VIII, XIV, and XVIII with mean LOD values ranging from 3.6 to 11.49. Among these QTL, qCBSDRNSc14AR/qCBSDRNAc14AR were detected consistently with different quantification methods for root necrosis. However, all the QTL for root necrosis exhibited PVE less than 10% and were considered as minor QTL.
QTL for CMD
QTL analysis for CMD data at both 3 and 6 MAP stages revealed a total of six putative QTL on chromosomes IX, X, and XII accounting for 3.47%– 45.97% PVE (Table 1). At 3 MAP, QTL analysis identified three consistent QTL (qCMDc9AA, qCMDc10.1AA, and qCMDc12.1AA), among them a major QTL, namely qCMDc12.1AA located at position 22 cM between markers cXII:5237169 and cXII:5330694 on chromosome XII, showed the highest contribution (19.09% PVE) to the variance (Fig 3). Similarly, three consistent QTL (qCMDc10.2AA, qCMDc12.1AA, and qCMDc12.2AA) were also detected for CMD resistance at 6 MAP stage. Of these QTL, qCMDc12.1AA and qCMDc12.2AA were located in adjacent regions on chromosome XII with peak positions at 22 cM and 18 cM, respectively. QTL qCMDc12.2AA for CMD resistance detected at position 18 cM between markers cXII:5224029 and cXII:6745505 on chromosome XII exhibited the highest LOD score up to 32.79 with PVE ranging from 41.12%– 45.97% (numbers after the chromosome designation cXII refer to the SNP base pair position on v5.1 of the cassava reference genome sequence. The sequence of markers on the genetic linkage map is not consistent with that on v5.1 of the physical map). Only major QTL qCMDc12.1AA was co-localized at both 3 and 6 MAP stage. It is worth noting that although qCMDc12.1AA and qCMDc12.2AA are sequentially located to one another at 22 cM and 18 cM on the genetic linkage map, qCMDc12.2AA encompasses qCMDc12.1AA on the physical map as flanking markers are cXII:5224029 and cXII:6745505, and cXII:5237169 and cXII:5330694, respectively. On the latest version of the physical map (v6.1), qCMDc12.1AA is positioned between c12:5530061 bp and c12:5770027 bp and qCMDc12.2AA between c12:5543126 bp and c12:7460818, so there is minor overlap in these QTL.
Trait nomenclature was named in combination of the disease, stage, location and year such as Cassava Mosaic Disease (CMD), 3 or 6 months after planting (3 MAP or 6 MAP) stage, crop season 2013 or 2014 at location Naliendele (NAL13 or NAL14) and Chambezi (CHA13 or CHA14), whereas pooled represents average of the trait from four environments either at 3 or 6 MAP stage.
Comparison of CMD chromosome 12 QTL with previously mapped markers
To more accurately compare the relative positions of SNP markers associated with QTL found in this study with those previously identified, all relevant markers were located on the M. esculenta v6.1 reference genome (http://phytozome.jgi.doe.gov/pz/portal.html) (S6 Table). In the present investigation, qCMDc12.1AA is located just outside the range of previously mapped markers with QTL qCMDc12.2AA [5,543,126–7,460,818 bp (v6.1)] encompassing the secondary peak from Wolfe et al. [37] at 7,032,174 bp (v6.1), Rabbi et al. [35] (S5214.30911) at 6,305,022 bp, Rabbi et al. [36] (S5214.780931) at 6,911,164 bp and overlapping with a peak of Masumba et al. [26] qCMDc12.2A [6,648,605–8,645,322 bp (v6.1)]. Previously described SSR markers associated with CMD resistance (SSRNS169, SSRNS158 and SSRY28) were identified in the interval 7,541,052–7,731,970 bp (v6.1), just outside the range of qCMDc12.2AA (Fig 3). This confirmed the co-localization of CMD2 locus with qCMDc12.2AA. SSR marker NS198, associated with the CMD3 locus, mapped to interval cXII:1106091 –cXII:1275959 on chromosome XII which was found approximately 30.3 cM from the CMD2 locus (SSRY28). In the M. esculenta v6.1 reference genome, the positions of the markers defining CMD2 and CMD3 loci confirmed that CMD3 locus is located further towards the beginning of the chromosome than the CMD2 locus on chromosome XII.
QTL for CGM
A total of five consistent QTL were detected for CGM resistance at 3 and 6 MAP (Table 1). Of these, four QTL (qCGMc5AR, qCGMc9AR, qCGMc13AR, and qCGMc18AR) for CGM resistance were identified at 3 MAP on chromosomes V, IX, XIII, and XVIII accounting for 1.35% to 11.06% phenotypic variance. Noticeably, QTL qCGMc9AR, explaining phenotypic variance between 1.76% to 7.44% was the only one consistently detected in all four environments. At 6 MAP only one QTL, qCGMc16AR, was detected for CGM resistance and accounted for phenotypic variance of 7.27% and had a LOD value of 2.59 across the years at Chambezi. All the QTL for CGM resistance were either specific to 3 or 6 MAP.
Comparison with previously linked molecular markers for CGM and putative candidate genes in the QTL region
Of the three SSR markers previously reported for CGM resistance (NS346, NS1009, and NS1099) [53,73], markers NS346 and NS1099 mapped to chromosome XVIII, whereas the position of marker NS1009 was not confirmed on the M. esculenta v6.1 reference genome because only the forward primer could be located on chromosome XIV (S6 Table). Marker NS346 was found to co-localize with the minor QTL qCGMc18AR detected here. A total of 68 putative genes were present in the QTL qCGMc18AR region located between markers cXVIII:3701705 and cXVIII:3265545 (0.524 Mbp region) (Table 1; S7 Table) and included gene families of specific interest, such as receptor-like kinases (RLK) and transcription factors, particularly three defense-related leucine-rich repeat (LRR) RLK genes.
Discussion
The viral diseases CBSD and CMD together cause over US$1 billion losses in annual production of cassava, and most of these losses affect smallholder farmers growing cassava as a food staple and selling surplus for income generation [74]. Although CMD and CBSD often occur in the same field, and co-infections do occur, there is no evidence for synergistic interaction between the viruses that cause CMD and CBSD [7,31,75]. The impact of CGM is more variable, being much more severe in dry seasons, but has not been widely quantified. One of the most effective means of combating these biotic constraints is the development of multiple disease and pest resistant varieties, in addition to biological control for CGM [21,76]. The identification of QTL associated with resistance would facilitate marker-assisted selection (MAS), which would enable selection for these traits at the seedling stage, dramatically reducing population sizes and enhancing efficiencies in breeding. In addition, this would enable pre-emptive breeding for CBSD resistance in West Africa, by applying artificial selection in the absence of disease pressure.
During the development of the mapping population, seed set was good, but germination was poor. Although seeds were stored for several months to break dormancy and flotation was used to eliminate non-viable seed, germination was only 21.5%. Poor germination was attributed to diurnal fluctuations in soil temperature as seeds were sown in pots on a bench. When pots were placed at ground level, germination increased. It is recommended that seed is germinated in seed beds, rather than in pots [27].
In the present investigation, substantial segregation was observed for reaction to CBSD foliar and root necrosis, and foliar CMD and CGM, indicating wide genetic variability among the full-sib family population (Fig 1, S3 Table). None of the progeny, however, exhibited immune responses to the diseases and pest under field conditions when considering performance across all the test environments. The skewed frequency distribution of all the studied traits observed toward the resistant phenotype could be due to the fact that extreme susceptibles succumbed to the disease(s) in the first few months after planting, and/or indicate that inheritance is oligogenic, as has been found in other host-pathogen systems including pest and viral disease resistance in cassava [27], and/or be due to environmental variation (moderately low or extremely higher disease/pest pressure) under field conditions [77]. This discrepancy may influence the ability to identify consistent QTL across multiple environments and becomes much more complex if the casual organism has multiple strains such as CBSD and CMD [26].
GBS has enabled a large number of SNP markers to be identified in a cost-effective manner and has facilitated the construction of high-density linkage maps in numerous crops, including cassava [35,36]. Here, stringent filters were used to generate high-quality SNPs, meaning that no imputation was necessary, although the number of markers was less than in other studies [35,36].
QTL associated with tolerance to CBSD foliar symptoms
A minor but consistent QTL (qCBSDFc11AR) accounting for between 1.43%–1.65% phenotypic variation for foliar CBSD resistance at 3 MAP stage was detected on chromosome XI (18.4Mbp– 18.7Mbp). Nzuki et al. [27] detected qCBSDRNFc11KR associated with both root and foliar symptoms on the same arm of this chromosome but at a slightly different location (15.6–15.7Mbp). Kawuki et al. [28] reported five significant SNPs associated with CBSD root necrosis resistance on chromosome XI covering 3.88 Mbp between 19,872,319 to 23,751,929 bp that explained 14.6% of the phenotypic variation, but these SNPs failed to reach the genome-wide Bonferroni significance threshold. The right arm of chromosome XI thus appears to be implicated in tolerance to both CBSD root necrosis and foliar symptoms, although with relatively small effect. In accordance with this study, Nzuki et al. [27] also observed QTL for CBSD foliar symptoms on chromosome XVIII (qCBSDFc18Ka and -b), albeit at slightly different positions, at 5.3–5.4 Mbp and 5.7 to 6.1 Mbp, as opposed to 10.1–10.9 Mbp in this study. It thus appears that chromosome XVIII has QTL influencing both foliar and root necrosis.
QTL associated with tolerance to CBSD root necrosis
Root necrosis was scored using two different methods; a subjective scale of 1–5 (RNS), and a quantitative method through the measurement of the percentage area of browning (RNA). Both methods identified a QTL on the right arm of chromosome XIV in three of the four environments (qCBSDRNSc14/qCBSDRNAc14). In addition, RNA identified two additional QTL on chromosomes VIII and XVIII, indicating greater resolution of this method of phenotyping. The QTL on chromosome XIV has so far not been detected by any other study [26–28], although Masumba et al. [26] reported a QTL associated with foliar resistance on the left arm of the same chromosome. Locus qCBSDRNSc14/qCBSDRNAc14 could represent a new source of tolerance to CBSD root necrosis in AR40-6.
CBSD root and foliar symptoms mainly influenced by different loci
In the present investigation, we assessed foliar CBSD symptoms at 3 and 6 MAP stages, and CBSD root necrosis on the basis of visual scale (1–5) as well as per cent root necrosis area by ImageJ program [78]. Two loci (qCBSDFc11AR and qCBSDFc18AR) for foliar CBSD and three genomic regions (qCBSDRNAc8AR, qCBSDRNSc14AR/qCBSDRNAc14AR and qCBSDRNAc18AR) for root necrosis were detected. This suggests that resistance to CBSD root necrosis is largely under different genetic control to CBSD foliar symptoms. It has also been observed in field conditions that foliar symptoms and root necrosis are often not highly correlated, as some varieties with clear leaf symptoms may fail to show root symptoms, while others that do not express leaf symptoms may produce root symptoms [6]. This is also consistent with the findings of Masumba et al. [26]. Similarly, Nzuki et al. [27] also reported that most of the QTL were associated with either root necrosis or foliar symptoms, except QTL qCBSDRNFc11KR and qCBSDRNFc15K, which were consistent across root and foliar symptoms, supporting the notion that resistance to foliar and root symptoms of CBSD are largely under different genetic control.
QTL associated with tolerance to CMD
The first analysis of CMD resistance in an F1 population derived from a cross between the Nigerian landrace TME 3 (CMD resistant) and an improved line TMS 30555 (CMD susceptible) suggested Mendelian inheritance was governed by a single dominant locus designated as CMD2 and tagged with SSR marker SSRY28 [34]. Subsequently, CMD2 was confirmed in another Nigerian landrace TME 7 [79]. Now several molecular markers tightly linked with CMD2 have been reported [26,34–37,41,79,80], although the designation of linkage group was different; group R [34], 8 [79], 16 [35,36] and chromosome XII [26].
In recent years, as larger population sizes have been used, coupled with higher density markers, it appears that two closely linked loci may be influencing CMD resistance, or this phenomenon may result from a multi-allelic effect. Two closely linked loci have been reported by Wolfe et al. [37] and Masumba et al. [26], and are also detected here (qCMD12.1AA and qCMD12.2AA). Large sections of repetitive DNA found within the CMD2 locus by Kuon et al. [39] could explain this.
In the present investigation, qCMDc12.1AA is located just outside the range of previously mapped markers, while the range of QTL qCMDc12.2AA encompasses the secondary peak found by Wolfe et al. [37] and both markers of Akano et al. and Rabbi et al. [34,35] and overlapping with a peak of Masumba et al. [26]. Previously described SSR markers associated with CMD resistance (SSRNS169, SSRNS158, and SSRY28) were identified just outside the range of qCMDc12.2AA but confirms the co-localization of qCMDc12.2AA with the CMD2 locus (Fig 3). qCMDc12.1AA was the most stable and effective QTL, accounting for 17.77%–17.85% and 3.47%–7.64% of the phenotypic variation at 3 and 6 MAP stages, respectively. It was detected between markers cXII:5237169 and cXII:5330694 (Fig 3). Besides the CMD2 locus, the position of SSR marker NS198 linked with the CMD3 locus [41] was identified in interval cXII:1106091 –cXII:1275959 on v5.1 on the draft cassava genome sequence and was not found to be associated with any QTL on chromosome XII in this study (Fig 3).
Interestingly, both parents used in this study are thought to possess the dominant CMD2 locus, yet segregation in the F1 progeny was still observed. The CMD2 locus was mapped in Albert by Masumba et al. [26], and AR40-6 was selected for the CMD2 locus using SSR markers at CIAT. Wolfe et al. [37] found that the CMD2 locus accounted for between 30 to 66% of genetic resistance and Rabbi et al. [36] found that a single marker within the CMD2 locus accounted for 74% of the variation. Wolfe et al. [37] and Masumba et al. [26], also found, in addition to the major QTL in the vicinity of the CMD2 locus, a number of other QTL of small effect, distributed across many chromosomes. Likewise, other influencing QTL such as qCMDc10.2AA and qCMDc9AA in Chambezi and qCMDc10.1AA in Naliendele were found in this study. Although the CMD2 locus clearly has a large effect, and thus explains the skewness observed in CMD at 3 and 6 MAP, it does not explain all the variation, hence possibly explaining some of the segregation observed.
Two peroxidases (Manes.12G076200 and Manes.12G076300), a class of protein implicated in whitefly-mediated geminivirus infection in tomato [81], were identified in both CMD resistant and susceptible varieties at the CMD2 locus defined in part by the major QTL of Wolfe et al. [37] and Kuon et al. [39]. Manes.12G076300 encodes a protein disulfide-isomerase-like 2–3 (PDI) which catalyzes the correct folding of proteins and prevents the clustering of precursors. In addition, a similar protein disulfide-isomerase-like 2–2 ortholog, a thioredoxin (PDIL2-2) was found within the secondary GWAS peak found by Wolfe et al. [37] and by Kuon et al. [39]. In barley, an ortholog of PDIL2-2 was identified as a major virus susceptibility factor with loss of function contributing resistance to bymoviruses [82]. Under this peak a Ubiquitin-conjugating enzyme E2 ortholog (UBC5) gene was also identified. Similar genes in the ubiquitinylation pathway have been found to influence plant virus infection response [37,83]. Kuon et al. [39] also highlight the possible involvement of a Suppressor of Gene Silencing 3 (SGS3) gene, 1.71 Mb downstream of the CMD2 locus. They also noted that two genes present in the variety TME3 (containing CMD2 resistance elements) were missing in two CMD susceptible genotypes AM560-2 and 60444. These were both short gene models of unknown functions.
QTL associated with tolerance to CGM
SSR markers NS1009 and NS346 have been reported for utilization in MAS for CGM resistance at CIAT as part of validation studies [73]. Macea-Choperena et al. [53] also displayed the association of previously reported marker NS346 and new marker NS1099 for CGM resistance in the individuals of different families of cassava. In the present investigation, a total of five consistent QTL for CGM resistance were identified, but all these QTL showed minor effects and were specific to either 3 or 6 MAP stages. Interestingly, marker NS346 linked with CGM resistance was found in the QTL qCGMc18AR region (cXVIII:3265545 –cXVIII:3701705) that accounted for a maximum of 8.6% of phenotypic variance for CGM resistance at 3 MAP stage. In addition, QTL qCGMc9AR located on chromosome IX accounted for between 1.76%– 7.44% of PVE and was the only QTL that was consistently detected in all four environments (Table 1).
In order to protect from disease and pest attack, plants evolved a wide range of defense mechanisms such as alteration in metabolic pathways and signaling molecules which stimulate defense-related processes [84]. The minor QTL qCGMc18A for CGM resistance identified in the present investigation harbors three leucine-rich repeat receptor-like protein kinase (LRR RLK) genes (Manes.18G053900, Manes.18G054000, and Manes.18G054100). LRRs have an important role for recognition specificity and these domains are present in the majority of R proteins which have been described as being effective against disease and insect pests [85,86].
In this study we identify 17 consistent QTL explaining between 1.35%– 45.97% of the phenotypic variance associated with CBSD, CMD, and CGM resistance. To achieve the goal of MAS, further resolution of QTL is necessary and the identification of markers accounting for maximum phenotypic variation. This could include fine-mapping or primers could be designed for specific polymorphisms across the identified QTL regions using the cassava genome sequence, and used to genotype segregating populations or contrasting genotypes, possibly using a bulk segregant analysis approach. Linkage of specific alleles with resistance or susceptibility could then be determined. Interestingly none of the QTL (except qCGMc9AR) were consistent across all environments. This has implications for the utility of linked markers derived from these QTL which may not be robust across environments.
Conclusions
This study identified QTL for multiple disease and pest resistance in cassava. Consistent with earlier studies, we find that root necrosis and foliar symptoms of CBSD are to some extent under different genetic control, and QTL appear to be influenced by genotype-by-environment interactions. We identify a new putative QTL for CBSD root necrosis on chromosome XIV from a variety with approximately 30.25% of its genome of South American origin (AR40-6). In addition, QTL on chromosomes XI and XVIII for CBSD foliar symptoms corroborate the findings from other studies. From our population, in which both parents are thought to possess the CMD2 locus, we provide further evidence for the existence of two QTL or multi-allelic variants influencing CMD resistance within the locality of the CMD2 locus. Similarly, the co-localization of our QTL qCGMc18AR for CGM resistance with previously identified SSR marker NS346 adds credence to the validity of this QTL. The growing evidence corroborating the effect of certain QTL associated with pest and disease resistance in cassava, and the germplasm in which these QTL exist, helps to progress the application of molecular markers in cassava.
Supporting information
S1 Fig. The climate of experimental sites (Chambezi and Naliendele) in Tanzania.
https://doi.org/10.1371/journal.pone.0236674.s001
(PPTX)
S2 Fig. Description of genetic linkage map of F1 full-sib family of cassava derived from a cross between AR 40–6 and Albert using 2125 SNP markers.
The genetic distance between markers is given in centimorgans.
https://doi.org/10.1371/journal.pone.0236674.s002
(DOCX)
S1 Table. Analysis of variance (ANOVA) for field disease and pest resistance in the AR40-6 x Albert mapping population.
https://doi.org/10.1371/journal.pone.0236674.s003
(DOCX)
S2 Table. Correlation coefficients (r) of CBSD foliar and root necrosis symptoms, foliar CMD and CGM symptoms of the AR40-6 x Albert mapping population across four environments.
https://doi.org/10.1371/journal.pone.0236674.s004
(DOCX)
S3 Table. Trait distributions of full-sib population of AR40-6 x Albert based on pooled mean of four environments.
https://doi.org/10.1371/journal.pone.0236674.s005
(DOCX)
S4 Table. Summary of single nucleotide polymorphism (SNP) markers information of linkage map of AR40-6 x Albert derived clonal F1 population.
https://doi.org/10.1371/journal.pone.0236674.s006
(DOCX)
S5 Table. Favourable alleles at QTL flanking markers detected for field disease and pest resistance in the AR40-6 x Albert mapping population.
https://doi.org/10.1371/journal.pone.0236674.s007
(DOCX)
S6 Table. Comparison of previously reported markers with QTL identified in present investigation for disease and pest resistant loci in cassava.
https://doi.org/10.1371/journal.pone.0236674.s008
(DOCX)
S7 Table. Genes underlying the consistent QTL (qCGMc18) for resistance/tolerance to cassava green mite.
https://doi.org/10.1371/journal.pone.0236674.s009
(XLSX)
Acknowledgments
Authors are thankful to Dan S. Rokhsar (University of California, Berkeley, California, USA) and Steve Rounsley (Genus, Basingstoke, UK) for their kind support during the experiment.
References
- 1. Olasanmi B, Akoroda MO, Egesi C, Okogbenin E, Fregene M. Cross-compatibility among six improved cassava (Manihot esculenta Crantz) varieties. J Root Crops. 2014;40(1):15–22.
- 2. Mahungu NM, Dixon AGO, Kumbira JM. Breeding cassava for multiple pest resistance in Africa. Afr Crop Sci J. 1994;2(4):539–552.
- 3. Calvert LA, Thresh JM. The viruses and virus diseases of cassava. In: Hillocks RJ, Thresh JM, Bellotti AC. (eds) Cassava: Biology, production and utilization. CABI. 2002;10:237–260.
- 4. Legg JP, Kumar PL, Makeshkumar T, Tripathi L, Ferguson M, Kanju E, et al. Cassava virus diseases: biology, epidemiology, and management. Adv Virus Res. 2015a;91:185–142.
- 5. Bellotti AC, Smith L, Lapointe SL. Recent advances in cassava pest management. Annu Rev Entomol. 1999;44:343–370. pmid:9990720
- 6.
Legg JP, Thresh JM. Cassava virus diseases in Africa. In “Plant Virology in Sub-Saharan Africa Conference Proceedings” (J. d’A. Hughes and B. O. Odu eds), IITA, Ibadan, Nigeria. 2003;517–552.
- 7. Alicai T, Omongo CA, Maruthi M, Hillocks RJ, Baguma Y, Kawuki R, et al. Re-emergence of cassava brown streak disease in Uganda. Plant Dis. 2007;91:24–29. pmid:30781061
- 8. Storey HH. Virus diseases of East African plants. VI-A progress report on studies of the disease of cassava. East Afric J. 1936;2:34–39.
- 9. Nichols RFW. The brown streak disease of cassava: distribution, climatic effects and diagnostic symptoms. East Afr Agricul J. 1950;15:154–160.
- 10. Legg JP, Raya M. Survey of cassava virus diseases in Tanzania. Int J Pest Manage. 1998;44:17–23.
- 11. Hillocks RJ, Raya MD, Thresh JM. Distribution and symptom expression of cassava brown streak disease in southern Tanzania. Afr J Root Tuber Crops. 1999;3:57–62.
- 12. Bigirimana S, Barumbanze P, Ndayihanzamaso P, Shirima R, Legg JP. First report of cassava brown streak disease and associated Ugandan cassava brown streak virus in Burundi. New Dis Rep. 2011;24:26.
- 13.
FAO. Cassava virus on verge of epidemic in East Africa: experts urge funding, swift action to protect staple food crop. Rome: Food and Agricultural Organisation. 2011.
- 14. Mulimbi W, Phemba X, Assumani B, Kasereka P, Muyisa S, Ugentho H, et al. First report of Ugandan cassava brown streak virus on cassava in Democratic Republic of Congo. New Dis Rep. 2012;26:11.
- 15. Tomlinson KR, Bailey AM, Alicai T, Seal S, Foster GD. Cassava brown streak disease: historical timeline, current knowledge andfuture prospects. Mol Plant Pathol. 2018:19(5);1282–1294. pmid:28887856
- 16. Roux-Cuvelier M, Teyssedre D, Chesneau T, Jeffray C, Massé D, Jade K, et al. First report of Cassava brown streak disease and associated Ugandan cassava brown streak virus in Mayotte Island. New Dis Rep. 2014:30(28).
- 17.
IITA. Combating the threat of CBSD. http://r4dreview.org/2011/04/combating-the-threat-of-cbsd/. Cited on 12/12/2011. 2011.
- 18.
Cuambe C, Zacarias AM, Mutaca AG, Valentim R, Amisse J, Avijala M, et al. Evaluation of cassava root yield during four crop season in high CBSD incidence zone in Mozambique. In: Mahungu NM, Mkumbira J. (eds) Root and tuber crops for poverty alleviation through science and technology for sustainable development. 10th ISTRC Symposium held at Centro Internacional Joaquin Chissano, 8–12 October 2007, Maputo, Mozambique. 2007.
- 19. Ephraim N, Yona B, Evans A, Titus A. Effect of cassava brown streak disease (CBSD) on cassava (Manihot esculenta Crantz) root storage components, starch quantities and starch quality properties. Int J Plant Physiol Biochem. 2015;7(2):12–22.
- 20. Patil BL, Legg JP, Kanju E, Fauquet CM. Cassava brown streak disease: a threat to food security in Africa. J Gen Virol. 2015;96:956–968. pmid:26015320
- 21.
Asiedu R, Ng SYC, Bai KV, Ekanayake IJ, Wanyera NMW. Genetic improvement. In: Orkwor GC, Asiedu R, Ekanayake IJ. (eds) Food yams: advances in research. IITA and NRCRI, Ibadan, Nigeria. 1998. 63–104 p.
- 22. Jennings DL, Iglesias C. Breeding for crop improvement. In: Hillocks RJ, Thresh JM, Bellotti AC. (eds) Cassava: Biology, production and utilization. CABI. 2002;10:149–166.
- 23.
Zacarias AMS. Breeding potential of cassava (Manihot esculenta Crantz) in Mozambique. PhD thesis, Bloemfontein, University of the Free State, South Africa. 2008.
- 24. Jennings DL. Further studies in breeding cassava for virus resistance. East Afr Agric J. 1957;22:213–219.
- 25. Kulembeka HP, Ferguson M, Herselman L, Kanju E, Mkamilo G, Masumba E, et al. Diallel analysis of field resistance to brown streak disease in cassava (Manihot esculenta Crantz) landraces from Tanzania. Euphytica. 2012;187:277–288.
- 26. Masumba EA, Kapinga F. Mkamilo G, Salum K, Kulembeka H, Rounsley S, et al. QTL associated with resistance to cassava brown streak and cassava mosaic diseases in a bi-parental cross of two Tanzanian farmer-varieties, Namikonga and Albert. Theor Appl Genet. 2017;130(10):2069–2090. pmid:28707249
- 27. Nzuki I, Katari MS, Bredeson JV, Masumba E, Kapinga F, Salum K, et al. QTL mapping for pest and disease resistance in cassava and coincidence of some QTL with introgression regions derived from Manihot glaziovii. Front Plant Sci. 2017;8:1168. pmid:28785268
- 28. Kawuki RS, Kaweesi T, Esuma W, Pariyo A, Kayondo IS, Ozimati A, et al. Eleven years of breeding efforts to combat cassava brown streak disease. Breeding Sci. 2016;66(4):560–571.
- 29. Sheat S, Fuerholzner B, Stein B, Winter S. Resistance against cassava brown streak viruses from Africa in cassava germplasm from South America. Front Plant Sci. 2019;10:567. pmid:31134114
- 30. Legg JP, Owor B, Sseruwagi P, Ndunguru J. Cassava mosaic virus disease in East and Central Africa: epidemiology and management of a regional pandemic. Adv Virus Res. 2006;67:355–418. pmid:17027685
- 31. Rey C, Vanderschuren H. Cassava Mosaic and Brown Streak Diseases: current perspectives and beyond. Annu Rev Virol. 2017;4:429–52. pmid:28645239
- 32. Hahn S.K., Terry E.R. & Leuschner K. Breeding cassava for resistance to cassava mosaic disease. Euphytica. 1980;29:673–683.
- 33. Ceballos H, Rojanaridpiched C, Phumichai C, Becerra LA, Kittipadakul P, Iglesias C, et al. Excellence in cassava breeding: Perspectives for the future. Crop Breed Genet Genom. 2020;2(2):e200008.
- 34. Akano O, Dixon A, Mba C, Barrera E, Fregene M. Genetic mapping of a dominant gene conferring resistance to cassava mosaic disease. Theor Appl Genet. 2002;105:521–525. pmid:12582500
- 35. Rabbi IY, Hamblin MT, Gedil MA, Kulakow PA, Ferguson M, Ikpan AS, et al. Genetic mapping using genotyping-by-sequencing in the clonally propagated cassava. Crop Sci. 2014a;54:1–13.
- 36. Rabbi IY, Hamblin MT, Kumar PL, Gedil MA, Ikpan AS, Jannink JL. et al. High-resolution mapping of resistance to cassava mosaic geminiviruses in cassava using genotyping-by-sequencing and its implications for breeding. Virus Res. 2014b;186:87–96.
- 37. Wolfe MD, Rabbi IY, Egesi C, Hamblin M, Kawuki R, Kulakow P, et al. Genome-wide association and prediction reveals genetic architecture of cassava mosaic disease resistance and prospects for rapid genetic improvement. Plant Genome. 2016;9(2):1–13.
- 38. Beyene G, Chauhan RD, Wagaba H, Moll T, Alicai T, Miano D, et al. Loss of CMD2-mediated resistance to cassava mosaic disease in plants regenerated through somatic embryogenesis. Mol Plant Pathol. 2016;17(7):1095–1110. pmid:26662210
- 39. Kuon JE, Qi W, Schläpfer P, Hirsch-Hoffmann M, von Bieberstein PR, Patrignani A, et al. Haplotype-resolved genomes of geminivirus-resistant and geminivirus-susceptible African cassava cultivars. BMC Biol. 2019;17(1):1–15. pmid:30616566
- 40. Fregene M, Puonti-Kaerlas J. Cassava biotechnology. In: Hillocks RJ, Thresh JM, Bellotti AC. (eds) Cassava: Biology, production and utilization. CABI. 2002;10:179–207.
- 41. Okogbenin E, Egesi CN, Olasanmi B, Ogundapo O, Kahya S, Hurtado P, et al. Molecular marker analysis and validation of resistance to cassava mosaic disease in elite cassava genotypes in Nigeria. Crop Sci. 2012;52(6):2576–2586.
- 42. Bondar G. Notas entomológicas da Bahia. III.–Revista Entomologia, Rio de Janeiro. 1938;9:441–445.
- 43. Campo BVH, Hyman G, Bellotti A. Threats to cassava production: known and potential geographic distribution of four key biotic constraints. Food Sec. 2011;3(3):329–345.
- 44. Machi AR, Esteca FCN, Arthur PB, Gava MA, Arthur V. A Review on Mononychellus tanajoa (Bondar, 1938) pest of cassava in Brazil. Aust J Basic Appl Sci. 2014;8(3):342–348.
- 45.
Leuschner K. Pest control for cassava and sweet potato. In: Root crops in eastern Africa: Proceedings of a workshop held in Kigali, Rwanda. 23–27 Nov. 1980. Int Dev Res Ctr. Ottawa, ON. 1982.
- 46.
Yaninek S, Hanna R. Cassava green mite in Africa: A unique example of successful classical biological control of a mite pest on a continental scale. In: P. Neuenschwander, et al., editors, Biological control in IPM systems in Africa. CABI, Wallingford, UK. 2003; 61–76.
- 47. Onzo A, Hanna R, Sabelis MW. Biological control of the cassava green mite in Africa: impact of the predatory mite Typhlodromalus aripo. Entomologische Berichten. 2005;65(1):2–7.
- 48. Chalwe A, Melis R, Shanahan P, Chiona M. Inheritance of resistance to cassava green mite and other useful agronomic traits in cassava grown in Zambia. Euphytica. 2015;205:103–119.
- 49.
Bellotti AC. Arthropod pests. In: Hillocks R.J. and Thresh J.M., editors, Cassava: Biology, production and utilization. CABI, Wallingford, UK. 2002;209–235.
- 50. Ezenwaka L, Del Carpio DP, Jannink JL, Rabbi I, Danquah E, Asante I, et al. Genome-wide association study of resistance to cassava green mite pest and related traits in cassava. Crop Sci. 2018;58(5):1907–18.
- 51. Jaramillo G, Morante N, Perez JC, Calle F, Ceballos H, Arias B, et al. Diallel analysis of cassava adapted to the mid-altitude valleys environment. Crop Sci. 2005;45:1058–1063.
- 52. Ezenwaka L, Rabbi I, Onyeka J, Kulakow P, Egesi C. Identification of additional/novel QTL associated with resistance to cassava green mite in a biparental mapping population. Plos one. 2020;15(4):e0231008. pmid:32240258
- 53. Macea-Choperena EDP, Ospina C, Fregene M, Montoya-Lerma J, Bellotti AC. Microsatellite identification associated with resistance to the mite Mononychellus tanajoa (Acari: Tetranychidae). Rev Colomb Entomol. 2012;38(1):70–75 (in spanish).
- 54. Bredeson JV, Lyons JB, Prochnik SE, Wu GA, Ha CM, Edsinger-Gonzales E, et al. Sequencing wild and cultivated cassava and related species reveals extensive interspecific hybridization and genetic diversity. Nature Biotechnol. 2016;34(5):562–570.
- 55. Kaweesi T, Kawuki R, Kyaligonza V, Baguma Y, Tusiime G, Ferguson ME. Field evaluation of selected cassava genotypes for cassava brown streak disease based on symptom expression and virus load. Virol J. 2014;11(1):1.
- 56.
Lynam J. The dynamic tension between public and private plant breeding in Thailand. A report for the global partnership initiative for plant breeding capacity building. FAO Plant Production and Protection Paper 208. ISSN 0259-2517. 2011;1-27p.
- 57.
Kulembeka HPK. Genetic linkage mapping of field resistance to cassava brown streak disease in cassava (Manihot esculenta Crantz) landraces from Tanzania. PhD Thesis, University of the Free State, Bloemfontein South Africa. 2010. 1–281 p.
- 58. Mohammed IU, Abarshi MM, Muli B, Hillocks RJ, Maruthi MN. The symptom and genetic diversity of cassava brown streak viruses infecting cassava in East Africa. Adv Virol. 2012;795697. pmid:22454639
- 59.
IITA. Cassava in tropical Africa: A reference manual. International Institute of Tropical Agriculture, Ibadan, Nigeria. 1990. 1–176 p.
- 60. Shirima RR, Maeda DG, Kanju EE, Tumwegamire S, Ceasar G, Mushi E, et al. Assessing the degeneration of cassava under high-virus inoculum conditions in coastal Tanzania. Plant Disease 2019;103:2652–2664. pmid:31322490
- 61. Maruthi MN, Hillocks RJ, Mtunda K, Raya MD, Muhanna M, Kiozia H, et al. Transmission of cassava brown streak virus by Bemisia tabaci (Gennadius). J Phytopath. 2005;153:307–312.
- 62. Knapp SJ, Stroup WW, Ross WM. Exact confidence intervals for heritability on a progeny mean basis. Crop Sci. 1985;25:192–194.
- 63. Johnson HW, Robinson HF, Comstock RE. Estimates of genetic and environmental variability in soybean. Agron J. 1955;47:314–318.
- 64. Dellaporta SL, Wood J, Hicks KB. A plant DNA minipreparation: Version II. Plant Mol Biol Rep. 1983;1(4):19–21.
- 65. ICGMC. High-resolution linkage map and chromosome-scale genome assembly for cassava (Manihot esculenta Crantz) from 10 populations. G3 Genes Genom Genet. 2015;5(1):133–144.
- 66. Van Ooijen JW. JoinMap 4, Software for the calculation of genetic linkage maps in experimental populations. Wageningen, Kyazma BV. 2006;33:10–1371.
- 67. Kosambi DD. The estimation of the map distance from recombination values. Ann Eugen. 1944;12:172–175.
- 68. Zhang L, Meng L, Wu W, Wang J. GADC: Integrated software for genetic analysis in clonal F1 and double cross populations. J Hered. 2015;106(6):741–744. pmid:26503825
- 69. Yu Y, Zhang J, Shi Y, Song Y, Wang T, Li Y. QTL analysis for plant height and leaf angle by using different populations of maize. J Maize Sci. 2006;14:88–92.
- 70. Collard BCY, Jahufer MZZ, Brouwer JB, Pang ECK. An introduction to markers, quantitative trait loci (QTL) mapping and marker-assisted selection for crop improvement: the basic concepts. Euphytica. 2005;142:169–196.
- 71. Lander ES, Botstein D. Mapping mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics. 1989;121(1):185–199. pmid:2563713
- 72. Kumar N, Kulwal PL, Balyan HS, Gupta PK. QTL mapping for yield and yield contributing traits in two mapping populations of bread wheat. Mol Breed. 2007;19(2):163–177.
- 73. Ceballos H, Okogbenin E, Perez JC, Lopez-Valle LAB, Debouck D. Cassava. In: Bradshaw I, John E. (ed) Root and tuber crops. Springer Science & Business Media. 2010;7:53–96.
- 74. Legg JP, Shirima R, Tajebe LS, Guastella D, Boniface S, Jeremiah S, et al. Biology and management of Bemisia whitefly vectors of cassava virus pandemics in Africa. Pest Manag Sci. 2014;70(10):1446–1453. pmid:24706604
- 75. Legg JP, Jeremiah SC, Obiero HM, Maruthi MN, Ndyetabula I, Okao-Okuja G, et al. Comparing the regional epidemiology of the cassava mosaic and cassava brown streak virus pandemics in Africa. Virus Res. 2011;159(2):161–170. pmid:21549776
- 76. Wiesner-Hanks T, Nelson R. Multiple disease resistance in plants. Annu Rev Phytopathol. 2016;54:229–252. pmid:27296142
- 77. Frei A, Blair MW, Cardona C, Beebe SE, Gu H, Dorn S. QTL mapping of resistance to Thrips palmi Karny in common bean. Crop Sci. 2005;45(1):379–87.
- 78. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 2012;9:671–675. pmid:22930834
- 79. Lokko Y, Danquah EY, Offei SK, Dixon AGO, Gedil MA. Molecular markers associated with a new source of resistance to the cassava mosaic disease. Afr J Biotechnol. 2005;4:873–881.
- 80. Okogbenin E, Porto M.C.M, Egesi C, Mba C, Espinosa E, Santos L.G, et al. Marker-assisted introgression of resistance to cassava mosaic disease into Latin American germplasm for the genetic improvement of cassava in Africa. Crop Sci. 2007;47(5):1895–1904.
- 81. Dieng H, Satho T, Hassan AA, Aziz AT, Morales RE, Hamid SA, et al. Peroxidase activity after viral infection and whitefly infestation in juvenile and mature leaves of Solanum lycopersicum. J Phytopathol. 2011;159(11–12):707–712.
- 82. Yang P, Lüpken T, Habekuss A, Hensel G, Steuernagel B, Kilian B, et al. PROTEIN DISULFIDE ISOMERASE LIKE 5–1 is a susceptibility factor to plant viruses. Proc Natl Acad Sci USA. 2014;111:2104–2109. pmid:24481254
- 83. Becker F, Buschfeld E, Schell J, Bachmair A. Altered response to viral infection by tobacco plants perturbed in ubiquitin system. Plant J. 1993;3(6):875–881.
- 84. Suzuki N, Rivero RM, Shulaev V, Blumwald E, Mittler R. Abiotic and biotic stress combinations. New Phytol. 2014;203:32–43. pmid:24720847
- 85. Gururani MA, Venkatesh J, Upadhyaya CP, Nookaraju A, Pandey SK, Park SW. Plant disease resistance genes: Current status and future directions. Physiol Mol Plant Path. 2012;78:51–65.
- 86. Smith CM, Clement SL. Molecular bases of plant resistance to arthropods. Annu Rev Entomol. 2012;57:309–328. pmid:21910639