QTL Mapping of Traits Associated with Dual Resistance to the African Stem Borer (Busseola fusca) and Spotted Stem Borer (Chilo partellus) in Sorghum (Sorghum bicolor)

Sorghum (Sorghum bicolor (L.) Moench) is an important food crop in semi-arid tropics. The crop grain yield ranges from 0.5 t/ha to 0.8 t/ha compared to potential yields of 10 t/ha. The African stem borer Busseola fusca Fuller (Noctuidae) and the spotted stem borer Chilo partellus Swinhoe (Crambidae), are among the most economically important insect pests of sorghum. The two borers can cause 15% - 80% grain yield loss in sorghum. Mapping of QTLs associated with resistance traits to the two stem borers is important towards marker-assisted breeding. The objective of this study was to map QTLs associated with resistance traits to B. fusca and C. partellus in sorghum. 243 F9:10 sorghum RILs derived from ICSV 745 (S) and PB 15520-1 (R) were selected for the study with 4,955 SNP markers. The RILs were evaluated in three sites. Data was collected on leaf feeding, deadheart, exit holes, stem tunnels, leaf toughness, seedling vigour, bloom waxiness, and leaf glossiness. ANOVA for all the traits was done using Genstat statistical software. Insect damage traits and morphological traits were correlated using Pearson's correlation coefficients. Genetic mapping was done using JoinMap 4 software, while QTL analysis was done using PLABQTL software. A likelihood odds ratio (LOD) score of 3.0 was used to declare linkage. Joint analyses across borer species and sites revealed 4 QTLs controlling deadheart formation; 6 controlling leaf feeding damage; 5 controlling exit holes and stem tunneling damages; 2 controlling bloom waxiness, leaf glossiness, and seedling vigour; 4 conditioning trichome density; and 6 conditioning leaf toughness. Joint analyses for B. fusca and C. partellus further revealed that marker CS132-2 colocalised for leaf toughness and stem tunneling traits on QTLs 1 and 2, respectively; thus, the two traits can be improved using the same linked marker. This study recommended further studies to identify gene(s) underlying the mapped QTLs.


Introduction
Sorghum (Sorghum bicolor (L.) Moench) is an important food crop in drought prone areas in the tropics. The crop is cultivated by more than 500 million resource-challenged smallholder farmers, mostly women [1]. The crop is an important industrial crop in East Africa and has untapped potential in bioenergy production [2]. The African stem borer, Busseola fusca Fuller (Noctuidae) and the spotted stem borer Chilo partellus Swinhoe (Crambidae), are among the most economically important insect pests of sorghum and maize in Eastern and South Africa [3]. The two stem borer species cause leaf feeding, deadheart formation, exit holes, and stem tunneling damages in cereals [3]. Stem borers are associated with grain yield loss of 15% -80% depending on borer species population and variety phenological stage at the time of attack [4]. Management approaches such as cultural practices and use of synthetic chemical pesticides have yielded little success in the management of the stem borers. Host plant resistance can reduce sorghum grain yield losses to stem borers and thus improve food security among rural households [5]. Breeding for host plant resistance is part of integrated pest management that can contribute in an economic way in the management of insect pests in indigenous cereals cultivated by resource-challenged smallholder farmers [5]. Breeding for stem borer resistance has been slow partly due to the inadequate understanding of inheritance of traits conditioning resistance to the pests. Stem borer resistance is a polygenic trait controlled by many genes of small effects [6]. Grain sorghum with dual resistance to B. fusca and C. partellus has been reported [7]. There is no information about genomic regions associated with dual resistance to B. fusca and C. partellus in grain sorghum. Traits such as leaf toughness, trichomes, bloom waxiness, leaf glossiness, and seedling vigour have been reported to condition resistance to borer and foliar insect resistance in cereals [5,[8][9][10][11].
Identification and mapping of quantitative trait loci (QTLs) associated with sorghum resistance to stem borers could enhance the efficiency and effectiveness in markerassisted breeding [12]. QTL mapping enhances the biological understanding of inheritance of quantitative traits, and the markers identified can be used to select for a complex trait [13]. QTL mapping using recombinant inbred lines (RILs) increases the power of QTL detection compared to the F 2:3 population because of complete homozygosity at QTLs and marker loci [14]. Marker-assisted selection for stem borer resistance can enhance breeding for B. fusca and C. partellus since traditional breeding has been unsuccessful. Genetic linkage maps are essential for localization of genes conferring resistance/tolerance to stem borer damage in sorghum [15]. Single nucleotide polymorphism (SNP) markers have become widely accepted as a tool for understanding complex genetic traits and evolution [16]. SNPs have been found to be the most efficient genetic marker for gene identification since they are codominant, highly polymorphic, and have good reproducibility [16]. SNPs represent the finest resolution of a DNA sequence [17]. The genotyping-by-sequencing approach has been employed in whole-genome sequencing to discover SNPs for mapping studies in crop plants [18,19]. The objective of this study was to identify and characterize QTLs associated with resistance traits to B. fusca and C. partellus in sorghum.

Materials and Methods
2.1. Development of Mapping Population. 243 F 9:10 recombinant inbred lines (RILs) derived from a cross between the stem borer-susceptible cultivar ICSV 745 and stem borerresistant PB 15520-1 were selected for the mapping study. The progenies and their parents had been developed following a single seed descent approach in ICRISAT, Patancheru, India [20]. The two parents were crossed, and the resulting F 1 seeds were advanced to F 2 by selfing of single F 1 plants. The F 2 seeds were selfed and the resulting F 3 population grown in progeny rows [20]. Single sorghum plants in each of the progeny row were selfed and the process continued up to the F 9 generation [20]. Seeds from the F 9 sorghum plants of each row were bulked to produce the 243 F 9:10 RILs used in this study.

Experimental Design, Site, and Source of Stem Borer
Larvae. The 243 RILs along with their parents were imported from the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Patancheru, India to Kenya. The sorghum plants were phenotyped at Embu and Kabete for resistance against B. fusca for one season in each environment and at Kiboko for resistance against C. partellus for two seasons thus totaling to four environments. Each experiment in each site was laid out in a 25 × 10 alpha-lattice design consisting of twenty-five plots in ten blocks, replicated twice. Each plot consisted of 2 m rows with plants spaced at 0:75 m × 0:25 m inter-and intra-rows, respectively. Firstinstar neonates of the two borer species were obtained from the International Centre of Insect Physiology and Ecology (ICIPE), Nairobi, Kenya. At 30 days after sowing, five plants in each row were tagged and artificially infested with five larvae of the respective stem borer species using a camel hair brush as described by Singh et al. [21] in all the three sites.

Phenotypic Data Collection and Data
Analysis. Data on stem borer leaf feeding damage, deadheart incidence, number of exit holes, and stem tunnel length were scored as described by Muturi et al. [7]. Morphological traits measured included leaf toughness, seedling vigour, bloom waxiness, leaf glossiness, and total grain yield as described by Kumar et al. [22]. Data were subjected to analysis of variance using the residual maximum likelihood model (ReML) in Genstat Version 14 statistical package. The predicted means for each genotype were estimated with genotypes as fixed and reps as random effects in individual and across environment analyses. Phenotypic and genetic correlation coefficients were calculated from adjusted entry means across environments for each parameter. Estimates of variance components, including genotypic variance (σ 2 g), genotype × environment interaction (σ 2 g × e), and residual (σ 2 ) were calculated by equating the mean squares to their expected values as described by Shimelis and Shiringani [23]. Heritability (H) was estimated using Meta-R software version 6 using H = ð∂ 2 g Þ/ð∂ 2 g + ∂ 2 g×e/ðrÞ + ∂ 2 e/ðr×envÞ Þ, where ∂ 2 g = genotypic variance, ∂ 2 g×e = genotype by environment variance, ∂ 2 e = error variance, r = number of replications, and env = number of environments as described by Hallauer and Miranda, [24]. Direct and indirect effect analyses were conducted using path analysis to study interrelationship among resistance parameters and their relationship to grain yield reduction [25]. Genotypic and phenotypic correlation coefficients were calculated following the method described by Holland [26].
2.4. DNA Extraction and Quantification. Leaf tissue from the parental lines and the 243 RILs were harvested from 10-dayold sorghum seedlings. DNA was extracted according to [27]. The quality of DNA in each sample was checked using 0.8% agarose gel stained with ethidium bromide. Each well of the agarose gel was loaded with 5 μl of the sample, and the gel was allowed to run at 100 V for 5 minutes. After  Figure 1). Segregation of each marker was tested with a chi-square goodness of fit test to the expected Mendelian segregation ratio (1 : 1) of parental configuration. The names of the markers were coded for ease of analysis (Supplementary  Material Table 14). The markers that did not conform to the expected segregation ratio were excluded from the analysis. Three-point linkage analysis was performed for each linkage group, and the most saturated linkage was adopted as described by Ooijen [28]. After the addition of each loci, a ripple was used to verify local locus orders [28].
2.6. QTL Analysis. QTL analyses were performed on morphological data and on leaf feeding, deadheart, exit holes, and stem tunneling damage under infestation with B. fusca at the Kabete and Embu sites and the same traits under infestation with C. partellus at the Kiboko site. Joint analysis was also done using averaged means across environments and insect species. Composite interval mapping (CIM) was performed on the data using PLABQTL software, version 1.2 [29]. Whole-genome scan with CIM was conducted using an automatic cofactor selection model to determine additive effects at individual QTL, and the F value of ten options were selected. Cofactors were chosen by step-wise regression and Akaike's information criterion (AIC) of 3.0 with the "cov" statement in PLABQTL. Dependence of QTL estimation on sampling effects was estimated by a five-fold cross validation by dividing the genotypes into five subsets. The LOD threshold for declaring a putative QTL was set to 3.0 after performing 1,000 permutation tests (type I error level α = 10%). All QTLs identified in this study explained ≥10% of the total phenotypic variation and were thus classified as major QTLs [30].

Phenotypic Analysis
3.1.1. Results for Combined Analyses of Variance across Environments. The combined results for the two borer species suggested that plant damage (deadheart formation, leaf feeding damage, exit holes, and stem tunnels) and agro-morphological traits (seedling vigour and bloom wax-iness) evaluated in this study were influenced significantly (P < 0:01) by borer species tested, genotype, and the environment where the experiment was conducted ( Table 1).
Results of mean squares for leaf feeding, deadheart, exit holes and morphological traits at Kiboko are presented in Table 2. There were significant differences observed for all the traits measured except for leaf feeding damage. Exit holes and stem tunneling damages were the only plant damage traits that were significantly (P < 0:01) influenced by genotype by season interaction ( Table 2). Results of mean squares for leaf feeding, deadheart, exit holes, and morphological traits at Embu are presented in Table 3. There were significant (P < 0:01) genotypic differences on all the traits evaluated at Embu site for B. fusca ( Table 3).
Results of mean squares for leaf feeding, deadheart, exit holes, and morphological traits at Kabete are presented in Table 4. Genotype significantly (P < 0:01) influenced all the traits evaluated except leaf feeding damage (Table 4). Heritability estimates and their standard errors for the sorghum damage and morphological traits across the three environments are presented in Table 5. It was noted that heritability estimates for C. partellus at Kiboko were generally low for most of the traits compared to the two other sites where B. fusca was tested.
The contribution of B. fusca damage parameters to total grain yield through leaf feeding, deadheart, stem tunnels, and number of exit hole damages at Embu are presented in Table 6. Damage effects were partitioned into direct and indirect associations through path coefficient analysis, and grain yield was used as the resultant variable (Table 6). Deadheart, exit holes, leaf feeding, and stem tunneling damages had a negative correlation with grain yield. Deadheart had a significant direct positive effect on grain yield supported by an indirect positive effect through stem tunneling and indirect negative effects through the number of exit holes and leaf feeding damages. Exit holes had a direct positive effect on grain yield supported by indirect positive effects through deadheart and leaf feeding damages and an indirect negative effect through stem tunneling. Leaf feeding had a direct positive effect on grain yield supported by indirect positive effects through deadheart and the number of exit holes and an indirect negative effect through stem tunneling. Stem tunneling had a direct positive effect on grain yield supported by indirect positive effects through deadheart, number of exit holes, and leaf feeding damages. The contribution of B. fusca damage parameters to total grain yield through leaf feeding, deadheart formation, stem tunnels, and number of exit hole damages at Kabete is presented in Table 7.
Deadheart, exit holes, leaf feeding, and stem tunneling damages had a negative correlation with grain yield. Deadheart had a direct positive effect on grain yield supported by indirect positive effects on exit holes, leaf feeding, and stem tunneling damages. The number of exit holes had a direct positive effect on grain yield supported by indirect positive effects through deadheart, leaf feeding, and stem tunneling damages. Leaf feeding damage had a direct positive effect on grain yield supported by an indirect positive effect through deadheart damage and indirect negative effects through exit hole and stem tunneling damages. Stem   International Journal of Genomics     5 International Journal of Genomics tunneling had direct positive effects on grain yield supported by indirect negative effects through deadheart, the number of exit holes, and leaf feeding damages. The contribution of C. partellus damage traits to total grain yield through leaf feeding, deadheart, stem tunneling, and the number of exit hole damages at Kiboko are presented in Table 8.
Deadheart, exit holes, leaf feeding, and stem tunneling damages had a negative correlation with grain yield (Table 8). Deadheart had a direct positive effect on grain yield supported by an indirect positive effect through stem tunneling and an indirect negative effect through exit holes and leaf feeding damages. Exit holes had direct positive effects on grain yield supported by an indirect positive effect through leaf feeding and stem tunneling damages and an indirect positive effect through deadheart damage. Leaf feeding had a direct positive effect on grain yield supported by indirect negative effects through deadheart, exit holes, and stem tunneling damages. Stem tunneling had a direct positive effect on grain yield supported by indirect positive effects through deadheart, exit holes, and leaf feeding damages.
The results for correlations of mean for C. partellus damage and agromorphological traits at Kiboko, Kenya are presented in Table 11. Stem tunneling had a significant positive correlation with exit holes (r = 0:68, P ≤ 0:001). Days to 50% flowering had a positive correlation with leaf damage (r = 0:17, P = 0:008) and with plant height (r = 0:13, P = 0:04). Panicle length had a significant positive relationship with    Table 12. Three QTLs for deadheart damage due to B. fusca were detected on chromosomes 1, 3, and 5 (flanked by markers SB261_1, DB103_3, and AD323_5) at Embu ( Table 12) that explained 3.9% of the phenotypic variation, with individual QTLs accounting for 7.5% -11.2%. A simultaneous fit with all the three QTLs based on cross validation explained 28.3% of the adjusted genetic variance. The additive gene action ranged from -7.79 to 35.13. Two of the QTLs on chromosomes 1 and 5 (flanked by markers SB261_1 and AD323_5) were from the susceptible parent, while the remaining QTLs on chromosome 3 (marker DB103_3) came from the resistant parent (PB 15520-1). The resistant parent contributed to less deadheart damage, which is a putative trait associated with resistance to stem borers in sorghum.
Five QTLs for leaf feeding damage due to B. fusca at Embu site were detected on chromosomes 2 and 10 (flanked by markers CS115_2, CS190_2, CS389_2, CS414_2, and CPS162_10, respectively) ( Table 12). These explained 10.5% of the phenotypic variation, with individual QTLs accounting for 7.1% -9.1%. A simultaneous fit with all the five QTLs based on cross validation explained 40.6% of the adjusted genetic variance. The additive gene action ranged from -6.48 to 8.13. Three of the QTLs on chromosomes 2 and 10 (flanked by markers CS190_2, CS389_2, and CPS162_10) came from the susceptible parent. The rest of the QTLs on chromosome 2 (flanked by markers CS115_2 and CS414_2) came from the resistant parent (PB 15520-1), which contributed to less leaf feeding damage, which is a trait associated with resistance to stem borers in sorghum.
One QTL for exit hole damage due to B. fusca at Embu site was detected on chromosome 7 flanked by marker GH89_7 (Table 12). This explained 6.1% of the phenotypic variation, with individual QTL accounting for 9%. A simultaneous fit with the single QTL based on cross validation explained 9% of the adjusted genetic variance. The additive gene action was -1.42. This QTL came from the resistant parent (PB 15520-1) which contributed to less exit hole damage, a trait associated with resistance to stem borers in sorghum.
Two QTLs for stem tunneling damage by B. fusca at Embu site were detected on chromosomes 7 and 9, respectively (flanked by markers GH87_7 and BF123_9) ( Table 12). These explained 5.9% of the phenotypic variation, with individual QTLs accounting for 8.1% -8.2%. A simultaneous fit with the two QTLs based on cross validation explained 16.3% of the adjusted genetic variance. The additive gene action ranged from -2.39 to -7.33. The two QTLs came from the resistant parent (PB 15520-1), an indication of the transfer of genes associated with resistance to stem tunneling by borers in sorghum.
Four QTLs for deadheart damage due to C. partellus at Kiboko site were detected on chromosomes 2, 6, and 9 (flanked by markers CS369_2, CS389_2, EF322_6, and BF152_9) (Table 12). These QTLs explained 2.7% of the phenotypic variation with individual QTLs accounting for 8.3% -12.6%. A simultaneous fit with all the four QTLs based on cross validation explained 41.2% of the adjusted genetic variance. The additive gene action ranged from -8.46 to 3.32. One of the QTLs on chromosome 2 (flanked by marker CS369_2) came from the susceptible parent (ICSV 745), while the other three QTLs came from the resistant parent (PB 15520-1) which contributed to less deadheart damage.
Three QTLs for leaf feeding damage due to C. partellus at Kiboko site were detected on chromosomes 2 and 6 (flanked by markers CS133_2, CS397_2, and EF184_6) ( Table 12). The markers explained 4.1% of the phenotypic variation with individual QTLs accounting for 7.2% -8.7%. A simultaneous fit with all the three QTLs based on cross validation explained 24.5% of the adjusted genetic variance. The additive gene action ranged from -3.07 to 2.76. All the three QTLs came from the resistant parent (PB 15520-1) which contributed to reduce leaf feeding damage.
Seven QTLs for exit hole damage due to C. partellus at Kiboko site were detected on chromosomes 2, 3, 4, 5, 6, and 7 (flanked by markers CS414_2, DB153_3, DB208_3, BC222_4, m05/015.6, EF255_6, and GH66_7) ( Table 12). The QTLs explained 1.4% of the phenotypic variation with individual QTLs accounting for 7.2% -10.3%. A simultaneous fit with all the seven QTLs based on cross validation explained 58.5% of the adjusted genetic variance. The additive gene action ranged from -8.04 to 0.61. Two of the QTLs on chromosomes 2 and 3 (flanked by markers CS414_2 and DB153_3) came from the susceptible parent (ICSV 745). The other five QTLs came from the resistant parent (PB 15520-1), an indication of the transfer of the resistance genes to the progenies.
Three QTLs for stem tunneling damage due to C. partellus at Kiboko site were detected on chromosomes 3 and 7 (flanked by markers DB152_3, GH70_7, and GH118_7) ( Table 12). These explained 0.29% of the phenotypic variation, with individual QTLs accounting for 7.8% -11.4%. A simultaneous fit with all the three QTLs based on cross validation explained 28.7% of the adjusted genetic variance. The additive gene action ranged from -3.56 to 2.64. One of the QTLs on chromosome 3 flanked by marker DB152_3 came from the susceptible parent (ICSV 745), while the other two QTLs came from the resistant parent (PB 15520-1).
Colocalization of markers for some of the traits evaluated was detected on site analysis for B. fusca (at Embu and Kabete), and C. partellus (at Kiboko) ( Table 12). For example, it was observed that the region flanked by markers (DB172_3, EF184_6, and BF123_9) on chromosomes 3, 6, and 9 conditioned trichome density, deadheart damage, and stem tunneling damage at Kabete. Marker DB169_3 on chromosome 3 conditioned leaf glossiness and deadheart damage at Kabete. Marker CS403_2 on chromosome 2 conditioned seedling vigour and leaf damage at Kabete and Embu, respectively. Marker AD323_5 on chromosome 5 conditioned resistance to deadheart formation at Embu as well as seedling vigour at Embu. Marker CS259_2 on chromosome 2 conditioned resistance to deadheart formation at Kabete as well as bloom waxiness in Kiboko. Marker GH108_7 on chromosome 7 conditioned resistance to seedling vigour at Kiboko as well as stem tunneling in Kabete. Marker AD323_5 on chromosome 5 conditioned resistance to seedling vigour and deadheart damage at Embu. Marker CS389_ 2 conditioned resistance to leaf damage and deadheart damage at Embu and Kiboko. Marker CS414_2 on chromosome 2 conditioned resistance to exit holes at Kiboko and leaf damage at Embu. Marker CS190_2 on chromosome 2 conditioned resistance to leaf feeding at Embu and exit holes at Kabete.
Joint analysis of B. fusca and C. partellus damage and morphological traits across the three sites is presented on Table 13. Four QTLs due to deadheart damage were detected chromosomes 4, 5, 8, and 10 (flanked by markers BC242-4, m05/023.7, JK399-8, and CPS160-10, respectively) ( Table 13). These explained 2.3% of the phenotypic variation, with individual QTLs accounting for 7.5% -9%. A simultaneous fit with all the four QTLs based on cross validation explained 24.8% of the adjusted genetic variance. The additive gene action ranged from 3.32 to 16.64. One of the QTLs on chromosome 8 (flanked by marker JK399-8) came from the resistant parent (PB 15520-1), while the other three QTLs came from the susceptible parent (ICSV 745).
Six QTLs for leaf feeding damage were detected on markers on chromosomes 1, 3, 6, 7, and 8 (flanked by markers SB255-1, DB160-3, DB169-3, EF222-6, GH70-7, and JK290-8) Table 13). These explained 2.25% of the phenotypic variation, with individual QTLs accounting for 7.6% -17.4%. A simultaneous fit with all the six QTLs based on cross validation explained 56% of the adjusted genetic variance. The additive gene action ranged from -2.02 to 5.070. Two of the QTLs on chromosomes 3 and 7 (flanked by markers DB169-3 and GH70-7, respectively) came from the 12 International Journal of Genomics 13 International Journal of Genomics resistant parent (PB 15520-1) that contributed to reduce leaf feeding damage, while the other four QTLs came from the susceptible parent (ICSV 745).
Six QTLs for stem tunneling damage were detected on chromosomes 1, 2, 3, 4, and 9 (flanked by markers SB466_ 1, CS116-2, CS132-2, DB36_3, BC140_4, and BF140_9) ( Table 13). These explained 5.37% of the phenotypic variation, with individual QTLs accounting for 10.1%-14.7%. A simultaneous fit with all the five QTLs based on cross validation explained 65.5% of the adjusted genetic variance. The additive gene action ranged from -3.4 to 4.4. Two of the QTLs on chromosomes 2 and 9 (flanked by markers CS132-2 and BF140_9, respectively) came from the resistant parent (PB 15520-1), while the other four QTLs came from the susceptible parent (ICSV 745). The analyses further revealed that marker CS132-2 was colocalised for leaf toughness and stem tunneling traits on QTL 1 and 2, respectively.

Discussion
Heritability is used to predict response to selection and to help plant breeders know if it is more efficient to improve traits of economic importance through selection [31]. Heritability estimates thus explain the level to which genes control expression of the trait of interest. Heritability estimates for deadheart formation, leaf feeding damage, exit holes, stem tunneling, bloom waxiness, and seedling vigour tested for B. fusca at Embu and Kabete were low to high and ranged between 0.21 and 0.65. On the other hand, heritability estimates for the same aforementioned traits evaluated for C. partellus at Kiboko were low and ranged between 0.15 and 0.32. This observation suggested that genetic gain may not be realized in selection for some of the traits evaluated for the two borer species in the three sites. These traits that had low heritabilities are as follows: leaf feeding damage at Kabete and Kiboko; deadheart formation, bloom waxiness, and seedling vigour at Kiboko; and exit holes at all the three sites. Therefore, there is a need to consider other alternatives such as DNA marker technology to fix the desired genes. This observation is supported by other researchers. For example, Oloyede-Kamiyo et al. [31] worked on variability for resistance to the pink stem borer (Sesamia calamistis Hampson) and the sugarcane borer (Eldana saccharina Walker) in two tropical maize populations and observed that narrow-sense heritability was low to moderate and ranged from 1.45% for leaf feeding to 40.6% for stalk breakage.
Path coefficient analyses revealed that leaf feeding, deadheart, exit holes, and stem tunneling had a direct negative effect on grain yield. This observation suggested that either of the damage traits could lead to direct grain yield loss. Singh et al. [6] observed that direct effects of stem tunneling by C. partellus on grain yield loss were greater than leaf feeding and deadhearts in sorghum. However, Starks and Doggett [32] reported that leaf feeding damage by C. partellus in sorghum is a weak pointer of grain yield as production of new fresh leaves compensate for the leaf damage. Sandoya et al. [33] reported that improvement of maize stem tunneling resistance to the Mediterranean corn borer (Sesamia nonagrioides Lef) and the European corn borer (Ostrinia nubilalis Hbn) led to increased grain yield. The significant positive correlations observed between leaf damage and deadheart, and between stem tunneling and exit holes, implied that there is a direct relationship between the two traits, respectively, and either of the traits can be used to predict the other in the sorghum population studied. Generally, tall vigorous plants with bloom waxiness produced longer panicle length and high yield with less stem damage. There was a negative relationship between stem tunneling and bloom waxiness and between seedling vigour and deadheart formation implying that sorghum can be bred with either of the morphological traits for reduced damage with high grain yield to the two stem borers.
Colocalization of the markers for some of the traits evaluated was detected on site analysis for B. fusca (at Embu and Kabete) and C. partellus (at Kiboko). For example, it was observed that the region (flanked by markers DB172_3, EF184_6, and BF123_9) on chromosomes 3, 6, and 9, respectively, conditioned trichome density, deadheart damage, and stem tunneling damage at Kabete for B. fusca. This observation implies that the aforementioned traits can be improved by using the same linked markers. Marker DB169_3 on chromosome 3 conditioned leaf glossiness and deadheart damage at Kabete implying that both traits can be improved by using the same linked marker for B. fusca. Marker CS403_2 on chromosome 2 conditioned seedling vigour and leaf damage for B. fusca at Kabete and C. partellus at Embu, respectively, implying that both traits can be improved by using the same linked marker. Marker AD323_5 on chromosome 5 conditioned resistance to deadheart formation as well as seedling vigour at Embu (for B. fusca) suggesting that both traits can be improved by using the same linked marker. Marker CS259_2 on chromosome 2 conditioned resistance to deadheart formation at Kabete (for B. fusca) as well as bloom waxiness in Kiboko (for C. partellus) implying that the two traits can be improved by using the same linked marker. Marker GH108_7 on chromosome 7 conditioned resistance to seedling vigour at Kiboko (for C. partellus) as well as stem tunneling in Kabete (for B. fusca) implying that both traits can be improved by using the same linked marker. Marker AD323_5 on chromosome 5 conditioned resistance to seedling vigour and deadheart damage at Embu (for B. fusca) implying that both traits can be improved by using the same linked marker. Marker CS389_2 conditioned resistance to leaf damage and deadheart damage at Embu (for B. fusca) and Kiboko (for C. partellus) implying that both traits can 14 International Journal of Genomics be improved by using the same linked marker. Marker CS414_2 on chromosome 2 conditioned resistance to exit holes at Kiboko (for C. partellus) and leaf damage at Embu (for B. fusca) implying that both traits can be improved by using the same linked marker. Marker CS190_2 on chromosome 2 conditioned resistance to leaf feeding at Embu and exit holes at Kabete implying that both traits can be improved by using the same linked marker for B. fusca. This study identified and mapped genomic regions conditioning dual resistance to B. fusca and C. partellus in sorghum. The study utilized a large number of RILs and highdensity evenly spaced genetic markers scored to ensure high power and precision in QTL mapping as described by Bekele et al. [34]. A five-fold cross validation analysis was run in this study as described by Gowda et al. [30] to confirm the frequency of QTL detection that gave an estimation of the precision of QTL localization. Identification of QTLs that influence dual resistance to B. fusca and C. partellus would increase the efficiency of selection and breeding for resistance of these two cereal stem borer species. Majority of the identified loci were highly significant, and they accounted for a substantial amount of phenotypic variation for B. fusca and C. partellus. This implied that fixing these traits (leaf feeding, deadheart formation, exit holes, and stem tunneling damages) would lead to the development of resistant cultivars and thus improve food security since the losses to stem borers will be lowered. A number of reports on linkage and QTL mapping for borer resistance traits have been published in maize [35][36][37]. QTL mapping for C. partellus resistance using the same sorghum mapping population utilized in this study is reported by Vinayan [20]. QTLs governing more than one resistance trait (pleiotropic QTLs) were identified in the present study. For example, it was observed that markers DB172_3, EF184_6, and BF123_9 on chromosomes 3, 6, and 9 conditioned trichome density, deadheart damage, and stem tunneling damage at Kabete implying that all these traits can be improved using the same linked markers. Pleotropic QTLs conditioning resistance to borers have been reported in the maize population [38][39][40].

Conclusion and Recommendation
This study identified and mapped genomic regions that conditioned expression of dual resistance to B. fusca and C. partellus in sorghum. QTLs identified in this study can be used in marker-assisted selection in the development of B. fuscaand C. partellus-resistant sorghum cultivars. Marker CS132-2 colocalised for leaf toughness and stem tunneling on QTLs 1 and 2, respectively; thus, both traits can be improved for B. fusca and C. partellus with the help of the same linked marker. There is a need for further studies to identify gene(s) underlying the mapped QTLs.

Data Availability
Some of the data used to support the findings of this study are included in the article. Additional data are available from the corresponding author upon request.