Tomato breeding line NC 1CELBR was developed at North Carolina State University (NCSU). It is a large-fruited fresh-market tomato line with determinate growth habits and is resistant to EB. The line was developed by multiple crosses involving wild species S. habrochaites and S. pimpinellifolium (Gardner and Panthee, 2010). Dr. Jay Scott, University of Florida, kindly provided the seed of the susceptible cultivar Fla. 7775. Despite other similar characteristics, contrasting EB reactions in NC 1CELBR and Fla. 7775 provided ideal materials to develop a population for genetic mapping studies. Crosses were made in the fall of 2009 at the Mountain Horticultural Crops Research and Extension Center (MHCREC), (NCSU), Mills River, North Carolina (NC). The F₂ seeds were produced in the spring of 2010 by selfing the F₁. Subsequently, 174 F_2:3 families were developed and used for phenotypic evaluation in the field and greenhouse, SNP marker analysis, and QTL mapping.

To evaluate plants for resistance to EB in the field, the experiment was conducted in 2011. Seeds were planted in 72 cell flats (56 × 28 cm²) in potting mix in the first week of May, and transplants at about six weeks from seed were planted. In the first week of June 2011, greenhouse-grown seedlings of the 174 F₂ and F₁ hybrid (NC 10175), susceptible controls (Fletcher, NC123S and NC 30P), resistant controls (NC 2CELBR and Mountain Merit), and resistant and susceptible parents (NC 1CELBR and Fla. 7775) were planted at the Mountain Research Station, Waynesville, NC. Spacing was 45 cm between plant-to-plant and 150 cm between row-to-row. The soil was a clay-loam texture, and the natural daylight photoperiod was about 14/10 hr, with temperatures averaging 25-30°C at their high and 14-16°C at their low. This field site was chosen because A. linariae inoculum naturally occurs each year almost three weeks after transplanting. Parents and F₁ were planted as a control to make sure that the disease developed well in the susceptible parent and that the resistant parent was healthy even under high inoculum pressure. No fungicide application was made to control the EB whereas late blight and Septoria leaf spot-specific fungicides were applied to control those diseases by spraying Presidio every week in combination with others as per the fungicide spray guide in NC (Ivors, 2011). Each plant was assessed for EB symptoms six weeks after planting to the field using a Horsfall and Barratt (1945) rating scale of 1 to 11, where 1 indicates no EB symptoms on the leaf surface, and 11 indicates complete defoliation. Humid and warm conditions favor A. linariae development, which was conducive to EB development in 2011.

Seeds of the 174 F_2:3 population and resistant and susceptible parents (NC 1CELBR and Fla. 7775) were surface-sterilized and sown in the greenhouse at Mills River. Seeds were sown in 4P soil mixture (Fafard^®, Florida, USA) in flatbed metal trays in a standard seeding mix (2:2:1 v/v/v) peat moss: pine bark: vermiculite with macro- and micro-nutrients (Van Wingerden International Inc., Mills River, NC) in March 2015. After ten days, seedlings were transplanted to 24-cell flats (56 cm x 28cm). Three plants per genotype were planted with two replications, and the experiment was conducted in a completely randomized design. Plants in the greenhouse study were fertilized using a 20:20:20 ratio of nitrogen, phosphorus, and potassium, respectively. Standard greenhouse pesticide application was used for possible insect and bacterial disease control. A single-spore isolate of A. linariae Sorauer collected from naturally infected tomato plants in Hendersonville, NC was used in this study. The fungus was isolated from infected leaf tissues and grown on potato dextrose agar (PDA, 39 g of Difco PDA, Becton, Dickinson and Company, Sparks, MD) in 10-cm Petri dishes and incubated at 23° C under white fluorescent lamps with a 12-h photoperiod. This isolate collected from the field was confirmed as A. solani using microscopic examination and PCR-based assays (Gannibal et al., 2014). After 10-12 days, conidia were harvested by flooding the plates with sterile distilled water. The inoculum concentration was adjusted to 1 × 10⁷ conidia mL^-1 using a hemocytometer. Before inoculation, a drop (~10 µL) of Tween 20 (Polyoxyethylene-20-sorbitan monolaurate) was added to the inoculum suspension to facilitate uniform spore deposition onto leaves. Nine-week-old plants were artificially inoculated using a hand sprayer (R & D Sprayers Inc., Opelousas, LA, USA). After inoculation, plants were placed in the dark for 24 h and covered entirely with white plastic to create a relative humidity of > 95%. Each inoculated plant was scored for EB symptoms using a Horsfall-Barratt rating scheme (Horsfall and Barratt, 1945) at 14 and 21 days after inoculation, as described above. Average disease scores were used to measure resistance to EB and to identify QTL in the greenhouse trials.

Genomic DNA of young leaf tissues of each parent and individual plant from F₂ generation was extracted using the DNeasy Plant Mini Kit (Qiagen Inc., Valencia, CA). A NanoDrop (Model ND-2000, Thermo Scientific Inc., Wilmington, DE) was used to quantify each DNA sample. Approximately, 50 ng/µl of DNA was prepared from each sample for SNP genotyping. We used an optimized subset of 384 SNPs markers that were derived from the 7,725 SNP array developed by the Solanaceae Coordinated Agricultural Project (SolCAP) (Sim et al., 2012a; Sim et al., 2012b). The subset of markers was selected based on polymorphism rates among six fresh market tomato accessions, including Fla.7776, Fla. 8383, NC33EB-1, 091120-7, Fla. 7775, and NC 1CELBR. Also, the genetic position in the genome based on recombination (Sim et al., 2012a) and the physical position was considered important selection criteria to ensure genome coverage. These 384 SNPs were analyzed using the Kompetitive Allele Specific PCR (KASP) genotyping platform (LGC Genomics, Beverly, MA).

The visual illustration of the correlation matrix and principal component analysis (PCA) was done by using the R language v3.2.3 coupled with the RStudio interface v1.0.143 and R packages (“FactoMineR”, “factoextra”, “ggplot2”, “ggplots”, “corrplot”), respectively (R Core Team, 2018; Amanullah et al., 2022). The summary statistics and normal probability plots were calculated using the UNIVARIATE procedure of SAS. The heritability was estimated for each environment by calculating variance components using the ‘ASYCOV’ function in PROC MIXED in SAS (SAS Institute Inc., 2012).

Broad-sense heritability (H² ) was estimated using the following variance components from the F₂ population (Nyquist, 1991; Falconer and Mackay, 1996):

Narrow-sense heritability (h² ) was determined using a regression analysis of offspring on parent approach, using data from the F₂ and F₃ generations as has been used by Ohlson and Foolad (2015) and as follows (Nyquist, 1991; Falconer and Mackay, 1996):

Where, H= broad-sense heritability, h² = narrow-sense heritability, VG=genetic variance, VP=phenotypic variance, VA = additive variance, VD= dominance variance, VE=error variance, VF₂ = Variance at F₂ generation, VF₃ = Variance at F₃ generation, and Cov (F₃xF₂) = Covariance of individuals at F₂ and F₃ generations.

Of the 384 SNP markers tested, 375 were polymorphic between the two parental lines, NC 1CELBR and Fla. 7775, that were used for genetic map construction (Meng et al., 2015). The linkage map was constructed using JoinMap 4.0 (van Ooijen, 2006). The grouping mode was set as the autonomous limit of detection, the mapping algorithm was used to perform regression mapping (limit of detection > 2.5, recombination frequency< 0.4, and jump = 5) (Asekova et al., 2021). The Kosambi mapping function was used to convert recombination frequencies into map distance (Kosambi, 1943). Independent limit of detection and maximum likelihood algorithms were used for grouping and ordering of markers, respectively. The ordering of the markers within each chromosome was based on the recombination events between the markers. Linkage groups were compared with published tomato linkage maps.

QTL analysis was conducted using windows QTL Cartographer v 2.5 (Wang et al., 2010) software. The Composite Interval Mapping (CIM) method was used with the default parameters (model 6). A backward regression was used to perform the CIM analysis to enter or remove background markers from the model. The walking speed was set at one cM for the detection of QTL. The additive effect and the proportion of the phenotypic variation (R²) for each QTL were also obtained using this software. A 1000 permutation option was chosen to determine the likelihood of an odd (LOD) score threshold to identify the presence of QTL in both environments (Li et al., 2007; Li et al., 2008; Meng et al., 2015). We used 5 cM scanning steps for the detection of QTL. The coefficient of variance (R² -value), the relative contribution of genetic components, was calculated and described as the proportion of genetic variance explained by the QTL out of the total phenotypic variation. QTLs explaining more than 10% of the phenotypic variance were considered major QTLs, and QTLs found in at least two environments were considered to be consistent.

To designate each QTL, the letter ‘q,’ followed by an abbreviation of EB resistance (EBR) was used as ‘qEBR.’ Additionally, each QTL was classified by the chromosome in which a QTL was detected and then categorized by QTL number. Any QTL within a 5 cM distance on the same chromosome was regarded as a single QTL.

The disease symptoms of infected tomato plants in the greenhouse experiment varied from chlorotic and necrotic areas of leaves with concentric rings to defoliation and death. The two parental lines exhibited distinguished responses to EB, with NC 1CELBR being consistently resistant (disease score 3.0), and Fla. 7775 being susceptible (disease score 9.0) ( Figure 1 ). The inoculated plants were scored for EB symptoms using a Horsfall-Barratt rating scheme (Horsfall and Barratt, 1945) at 14 and 21 days after inoculation. In field experiments, higher disease severities (6 to 11) were observed in 2011 ( Figure 1A ). There was a significant variation among F_2:3 lines for visual disease rating ( Figure 1 and Table 1 ). Distribution of both field and greenhouse phenotypic data was continuous, indicating quantitative and polygenic control of EB resistance in tomatoes ( Figure 1 ).

The minimum and maximum EB development in 2011 in the population was 6 and 11, respectively, with an average of 8.1. In 2015, the minimum and maximum disease developments in this population were 1 and 11, with an average of 6.8 ( Figure 1 and Table 1 ). These basic statistics over the years indicated that there was a good distribution of EB resistance in this population. The broad-sense heritability estimate for phenotypic data was 28.3%, and 25.3% for 2011, and 2015 disease evaluations, respectively. The disease score values showed a negative correlation between the years 2011 and 2015 ( Figure 2A ). The PCA bi-plot showed the possible association and high percentage of phenotypic variability was observed between the data sets of EB resistance in both environments ( Figure 2B ). The dimension of the first PC (Dim1) broadly outlined and explained 51.8% of the phenotypic variability for EB resistance in 2011 ( Figure 1B ). The dimension of the second PC (Dim2) also distinguished the 48.4% of phenotypic variability for EB resistance in 2015 at opposite angles of the PCA biplot ( Figure 2B ). This data also showed that EB resistance is controlled by multiple genes.

A total of 375 SNP markers were polymorphic between the parents. Those markers were used to genotype the population. A linkage map was constructed with these markers which covered approximately 737.17 cM genetic distance. The map results yielded a total of 12 linkage groups which are comparable with other tomato linkage maps and the number of tomato chromosomes. The Individual chromosomes had 18 to 65 markers with lengths ranging from 42.04 to 88.87 cM ( Figure 3 ). Nearly 65 SNP markers were mapped on chromosome 4, followed by 42 SNP markers on chromosome 12 ( Figure 3 ).

We identified QTLs for EB resistance using 174 F_2:3 derived lines and the SNP-based linkage map in two environments ( Figure 4 and Table 2 ). In total, 6 QTLs, including major and minor effects, common for both environments were identified across the genome, explaining phenotypic variation (R²) ranging from 3.8 to 21.0% ( Figure 4 and Table 2 ). The QTLs on chromosomes 2, 8, and 11 (qEBR2011-2, qEBR2011-8, and qEBR2011-11) were detected in 2011, respectively. The QTLs qEBR2011-2 (LOD: 4.2), qEBR2011-8 (LOD: 4.2), and qEBR2011-11 (LOD: 4.0) explained 3.8%, 12.1% and 11.7% of total phenotypic variations ( Figure 4 and Table 2 ). The QTLs on same chromosomes were detected in 2015 as well ( Figure 4 and Table 2 ). The QTLs qEBR2015-2 (LOD: 5.0), qEBR2015-8 (LOD: 5.2), and qEBR2015-11 (LOD: 9.1) explained 21%, 11.4% and 19.8% of total phenotypic variations ( Figure 4 and Table 2 ). We used the linked markers of the resistant QTLs to compare the resistance levels and allelic effects in the mapping population ( Figure 5 ). As shown in the box plots, the homozygous resistant genotypes BB were associated with enhanced resistance compared to the homozygous susceptible genotype AA for all the QTLs in both environments ( Figure 5 ). It also confirmed that all the resistant alleles in mapping population were inherited from NC 1CELBR. These results indicated that multiple genes/QTLs are contributing to EB resistance.