RESEARCHOpen Access icon OPENOpen Access license

Evaluating Quantitative Trait Locus Resistance in Tomato to Multiple Xanthomonas spp.

    Authors and Affiliations
    • Eduardo Bernal
    • Debora Liabeuf
    • David M. Francis
    1. Department of Horticulture and Crop Science, The Ohio State University, Wooster, OH 44691


    Bacterial spot of tomato is a foliar disease caused by four Xanthomonas species. Identifying genetic resistance in wild tomatoes and subsequent breeding of varieties has been a strategy to reduce the loss from this disease because control using pesticides has been ineffective. Three independent sources of resistance have been identified with quantitative trait loci (QTL) mapping to the centromeric region on chromosome 11. These sources are derived from Hawaii 7998 (QTL-11A), PI 114490 (QTL-11B), and LA2533 (QTL-11C). To determine which QTL introgression from chromosome 11 provides the greatest resistance to multiple species, we developed near-isogenic lines (NILs) using marker-assisted backcrossing. In parallel, we developed an NIL that contains Rx-4/Xv3, which provides major gene resistance to Xanthomonas perforans. Additionally, we combined Rx-4/Xv3 resistance with QTL-11A. These sources of resistance were independently introduced into the susceptible parent, OH88119. During a 3-year period from 2016 to 2018, we evaluated backcross-derived families and NILs from each source in independent field trials inoculated with X. perforans, X. euvesicatoria, or X. gardneri. Our results suggest that both QTL-11C and QTL-11A combined with Rx-4/Xv3 provide effective genetic resistance against multiple Xanthomonas species. In addition, we provide evidence for additive to dominant genetic action for the QTL introgressions.

    Bacterial spot is a foliar disease of tomato (Solanum lycopersicum L.) caused by Xanthomonas euvesicatoria, X. perforans, X. vesicatoria, and X. gardneri. The disease causes significant yield losses throughout the world, especially in areas with high humidity and rainfall (Potnis et al. 2015). The symptoms include scab-like lesions on the fruit; black and chlorotic spots on the stem, leaves, peduncles; defoliation; and blossom abortion (Yang and Francis 2005). To control the spread of the disease, copper compounds are commonly used, although efficacy is poor under high disease pressure (Louws et al. 2001; Marco and Stall 1983). Additionally, Xanthomonas populations tend to be copper insensitive (Liao et al. 2019). Genetic resistance remains a strategy that may reduce losses caused by bacterial spot.

    The Xanthomonas species causing bacterial spot can be loosely aligned to historical race designations that define specific bacterial effector and resistance gene-mediated responses. Xanthomonas isolates with T1 phenotypes are currently considered X. euvesicatoria (Scott et al. 1995). Race T1 strains are defined by an avrRxv-mediated hypersensitive response (HR) on the tomato accession Hawaii 7998 (Whalen et al. 1993). Hawaii 7998 possesses three loci capable of eliciting an HR when challenged with strains containing the avrRxv effector (Wang et al. 1994; Yu et al. 1995). These loci fall on the top and bottom of chromosome 1 (Rx1 and Rx2) and chromosome 5 (Rx3), with Rx3 providing the most effective resistance in the field (Yang and Francis 2005). Races T3 and T4 tend to be X. perforans, with T3 strains defined by an avrXv3-mediated HR on Hawaii 7981, PI 128216, and several other Solanum pimpinellifolium accessions (Scott et al. 2001). The resistance locus Rx-4/Xv3 maps to chromosome 11 and defines the T3 interaction (Pei et al. 2012; Robbins et al. 2009; Scott et al. 1996; Wang et al. 2011). Race T4 is defined by xopJ4-mediated HR on LA0716, with resistance mapping to chromosome 6 (Sharlach et al. 2013). Because race T2 strains are defined by the lack of HR on key differential germplasm, these may be either X. vesicatoria or X. gardneri.

    In addition to major gene resistance, quantitative resistance that seems to be effective against multiple species has been identified. For example, PI 114490 has been described as resistant to T1, T2, and T3 (Scott et al. 2003). Furthermore, genetic studies have defined quantitative trait loci (QTL) that are effective against more than one Xanthomonad. For example, PI 114490 (S. lycopersicum var. cerasiforme) LA2533 (S. pimpinellifolium), and Hawaii 7998 all possess QTL in the centromeric region of chromosome 11 (Hutton et al. 2010; Liabeuf 2016; Sim et al. 2015). For Hawaii 7998 and PI 114490, the QTL region on chromosome 11 explained 14.6 to 63.8% of the phenotypic variation for resistance within data derived from T2, T3, and T4 strains (Yang et al. 2005). In a separate study, Hawaii 7998 also displayed QTL associated with resistance on chromosome 11 against T1 and T3 (Sim et al. 2015). For LA2533, the region from 7.8 to 54.8 Mb on chromosome 11 accounted for 38% of the phenotypic variation associated with resistance against X. gardneri (Liabeuf 2016). Previous studies have not evaluated all three sources or progeny against the same pathogen strains in common experiments.

    To address the relative effectiveness of the QTL on chromosome 11, we implemented a backcross breeding program with marker-assisted selection (MAS) and background genome selection to independently introgress chromosome 11 QTL from different sources to create near-isogenic lines (NILs). Coupling MAS to select marker-linked traits and selecting genomic regions of the recurrent parent using background genome selection (Hospital 2009) allowed us to efficiently develop and evaluate germplasm with QTL of interest in a uniform genetic background. In tomato, >7,720 validated single-nucleotide polymorphisms (SNPs) have been developed using next generation sequencing technology (Sim et al. 2012). We took advantage of these SNPs for selection in three backcross populations. The main objectives were to (i) use MAS and background genome selection to introgress three independent sources of a QTL found in the same region of chromosome 11, (ii) determine the effectiveness of the resistance from each source against multiple Xanthomonas species, and (iii) evaluate homozygous and heterozygous allele genetic effects across QTL introgressions.

    Materials and Methods

    Plant material.

    Multiple tomato backcross populations were developed based on three distinct sources of resistance (Hawaii 7998, PI 114490, and LA2533) as the origin of a QTL on chromosome 11 that confers resistance to bacterial spot. Advanced lines from established pedigrees were used as donors of the QTL, because these lines already possessed a greater percentage of elite processing cultivar genetic background. Parent OH087663 is a line derived from a complex population (Sim et al. 2015) and contains a QTL on chromosome 11 inherited from Hawaii 7998 via Fla. 7600 (Hutton et al. 2010; Scott et al. 2003). OH087663 also contains Rx-4/Xv3 on chromosome 11 inherited from PI 128216 (Robbins et al. 2009; Wang et al. 2011). Parent 01-BR-7087 is a line containing a QTL on chromosome 11 inherited from PI 114490. This line is derived from a cross between PI 114490 and FLa. 7600 (F1), which was backcrossed twice to OH9242 (Scott et al. 2003). Parent FG12-433E-43 is a line containing a QTL on chromosome 11 inherited from LA2533 derived from two backcrosses to OH2641 (Liabeuf et al. 2015). In all cases, OH88119 (Berry et al. 1995) was used as the susceptible recurrent parent. For ease of naming, each introgression has been designated with a unique name: QTL-11A (Hawaii 7998), QTL-11B (PI 114490), and QTL-11C (LA2533).

    NIL development.

    Development of NILs started in summer 2015 with crosses between each resistant parent containing QTL-11A, QTL-11B, or QTL-11C and our recurrent parent line OH88119 to produce BC1 populations. Ohio OH88119 was selected because of its susceptibility to all Xanthomonas species causing bacterial spot (Yang and Francis 2005). Progeny from BC1 populations were sown in 288 Square Deep Plug Trays (Hummert International) with PRO-MIX (Hummert International) and covered with a thin layer of vermiculite in the greenhouse. After 3 weeks, emerged seedlings were processed as described below.

    DNA isolation.

    A small leaflet from seedlings was sampled into 96-well cluster tubes (Corning Inc.). A hexadecyltrimethylammonium bromide (CTAB) method was scaled to 96-well tubes for DNA extraction as described (Sim et al. 2015). Samples were macerated in 150 µl of extraction buffer (0.35 M sorbitol, 0.1 M Tris buffer, 5 mM EDTA, and 25 mM sodium bisulfite at pH 7.5), 150 µl of CTAB nuclei lysis buffer (0.2 M Tris, 0.05 M EDTA, 2 M NaCl, and 2% CTAB at pH 7.5), and 60 µl of 5% Sarkosyl. Two 4-mm metallic beads were added to each tube, and a GenoGrinder (OPS Diagnostics) was used to grind tissue at 200 strokes per minute for 2 min. Tubes were incubated at 65°C for 20 min; then, they were set at room temperature for 10 min before adding 350 μl of chloroform isoamyl alcohol (24:1) followed by centrifugation at 5,000 × g for 10 min. The aqueous phase was transferred to a 96-well round-bottom microtiter plate (Corning Inc.) containing 110 μl of isopropanol and then centrifuged at 5,000 × g for 15 min for DNA precipitation. Plates were air dried for 30 min before resuspending the DNA with 100 μl of Tris EDTA buffer (10 mM Tris and 0.1 mM EDTA).


    Approximately 192 individuals were genotyped from each BC1 population using two PCR-based DNA markers: Marker Sli_1831 (forward primer: 5′-GGAAGCTTGGATTAAAGGGG-3′, reverse primer: 5′-CAGTCGCTTAGGAAACCGAG-3′) and Sli_1901 (forward primer: 5′-CGCGTTTCATCTTTTTCCTC-3′, reverse primer: 5′-TCACCTGATAGCAGTGACGTAG-3′). These markers broadly flank the known location of the QTL on chromosome 11 at physical positions 20 and 48 Mb, respectively, based on tomato reference genome (SL 3.0). Additionally, indel marker PCC12 (Pei et al. 2012) at physical position 53 Mb was used to select recombinants with and without Rx4/Xv3 within the QTL-11A BC1 population. PCR was conducted in 20-μl reactions consisting of 2 μl of buffer A (10 mM Tris-HCl, 50 mM KCl, and 1.5 mM MgCl2), 0.8 μl of 1.25 mM dNTP, 0.2 μl of 10 μM forward and reverse primers, 12.4 μl of ddH2O, 0.4 units of Taq DNA polymerase, and 4 μl of DNA template using the following cycling conditions: (1) 94°C for 3 min, (2) 94°C for 1 min, (3) 54°C for 30 s, (4) 72°C for 1 min, and (5) 72°C for 5 min. Steps 2 to 4 were repeated 35 times before the final annealing step. PCR products were run on a 2% agarose gel at 180 V for 1 h and 30 min and detected using ethidium bromide fluorescence.

    Individuals heterozygous for the QTL markers were selected for background genome selection. Markers for background selection were chosen from a set of high-quality validated SNPs based on distribution in the genome and polymorphic information content in germplasm described previously (Blanca et al. 2015; Sim et al. 2012). For each BC1 population, polymorphic markers spread across the genome, and distinguishing parents were selected (Table 1). SNPs were assayed using the Kompetitive Allele Specific PCR assay (Semagn et al. 2014).

    Table 1. Single-nucleotide polymorphic (SNP) markers used for background genome selection in BC1 populations

    Selected progeny from BC1 populations developed from QTL-11A, QTL-11B, and QTL-11C were self-pollinated. BC1S1 progeny were genotyped and selected for resistant and susceptible alleles at the chromosome 11 QTL introgression. Homozygotes were self-pollinated again (BC1S2) and paired as resistant or susceptible members of the same family before transplant in summer 2016 field trials. Family structure was maintained such that background genome effects could be tested based on the original BC1 selections. At the end of the trials, BC1S3 seed from the homozygous resistant progeny was collected from the field for future studies. Beginning in summer 2016, individuals that were homozygous resistant from BC1S2 progeny were advanced through two more generations of backcrossing. After the self-pollination step to produce BC3S1 seed, selection was imposed for the QTL and/or Rx4/Xv3. DNA isolation and selection occurred within BC3S1 before transplanting in summer 2017 trials. Individual plants were grouped based on homozygosity and heterozygosity for QTL introgression alleles to test for gene action. BC3-derived families were evaluated in both summer 2017 and summer 2018 trials.

    Bacterial strains, inoculum preparation, and inoculation.

    Five Xanthomonas strains (X. euvesicatoria race T1 [Xcv110C and Xcv767], X. gardneri [SM775-12 and SM605-11], and X. perforans race T3 [Xcv761]) were used. Strains were grown separately on nutrient yeast broth agar media at 28°C for 48 to 72 h and subsequently resuspended in autoclaved ddH2O. The optical density of each suspension was measured and standardized to a 10-mm path length at an absorbance of 600 nm = 0.15 (∼3.0 × 108 CFU/ml). Xanthomonas strains from the same species were then mixed to obtain a uniform concentration. Suspensions were sprayed to runoff with a compressed air sprayer (Preval sprayer) in a greenhouse. Inoculum was prepared and sprayed ∼1 week before field transplanting.

    Disease evaluation.

    Disease ratings were conducted on a per plot basis, each having eight plants of the same genotype in 10-foot plots. The severity of bacterial spot was measured using the Horsfall–Barratt scale (Horsfall and Barratt 1945), where 1 = no disease present and 12 = complete defoliation. Susceptible and resistant controls were used to monitor the progression of the disease over the growing season. Two ratings for each field were conducted at time points corresponding to when 80% of the fruits in plots were at the mature green (early) or ripe fruit (late) stage of maturity as described previously (Robbins et al. 2009).

    Experimental design.

    Field trials of germplasm with distinct QTL-11 introgressions were conducted in summer 2016, 2017, and 2018. Three separate fields were used in each year, and each field had backcross-derived families (2016), lines (2017), or NILs (2018); QTL-11 resistant parents; and susceptible control OH88119. Fields were inoculated with X. perforans, X. euvesicatoria, or X. gardneri. In 2016 and 2017, plots were evaluated for disease severity using a randomized complete block design (RCBD) with three blocks and eight-plant plots. In 2016, several BC1S2 families were advanced from each QTL-11 introgression, with each family having individuals grouped by resistant and susceptible QTL introgressions. In 2017, BC1S4 resistant genotypes and BC3S1 families were evaluated in the same locations inoculated with the same pathogens. BC3S1 families were selected and grouped based on homozygosity and heterozygosity for markers flanking the QTL introgressions. During 2016 and 2017, fields containing X. perforans and X. gardneri were located in Wooster, Ohio, and the X. euvesicatoria trial was at The Ohio State University’s North Central Agricultural Research Station (NCARS) in Fremont, Ohio. In 2018, BC3S3 NILs were evaluated for disease severity in an RCBD with four blocks and eight-plant plots in three fields. X. euvesicatoria and X. gardneri trials were located in Wooster, Ohio, and the X. perforans trial was located at NCARS in Fremont, Ohio.

    Data analysis.

    The data for QTL-11 BC1S2, BC1S4, and BC3S1 populations were analyzed independently each year using the fixed effect linear model for analysis of variance (ANOVA):


    where yijk is the disease rating for the ith introgression within QTL-11 introgression (Qi) for the jth family within QTL-11 introgression (Fj) in the kth block (Bk). The family term, Fj, was dropped from the model in the analysis of data for BC3S3 populations, because individual selections were advanced for each QTL-11 introgression and genetic backgrounds were uniform. Before conducting ANOVA, the disease rating data for individual plots were converted to mean percentages (Bock et al. 2009; Redman et al. 1964). Data for 24 comparisons were then evaluated for normality of the residuals and homogeneity of the variance using the Horsfall–Barratt scale, estimated mean percentages, and estimated mean percentages with an arcsin transformation. Data were tested using the Shapiro–Wilks test and Bartlett test in R using the functions “shapiro.test” and “bartlett.test” (R Core Team 2017). Tukey’s honest significant difference (HSD; α = 0.05) was used after a significant F test in ANOVA to examine individual introgression effects. Statistical analyses were conducted using the linear model function “lm” in the R Core Package and Tukey’s HSD function “HSD.test” in the agricolae package using R version 3.4.1 (de Mendiburu 2017; R Core Team 2017).


    Background genome selection and recovery of recurrent parent.

    Genome-wide SNP markers were applied to three distinct BC1 populations with the intent of selecting individuals with a higher percentage of OH88119 background. The average proportion of recurrent parent genome for 241 individuals across the three BC1 populations was 76.7% (Fig. 1). Selections within each BC1 population with at least 87.5% of the recurrent parent genome were identified. Eight, seven, and five selections were advanced for QTL-11A, QTL-11B, and QTL-11C BC1 populations, respectively. Background genome selection allowed us to effectively select BC1 individuals with a recurrent parent genome equivalent to a BC2. These individuals were both self-pollinated and backcrossed for evaluation in Xanthomonas trials.

    Fig. 1.

    Fig. 1. The histogram represents the distribution and proportion of OH88119 recurrent parent genome within quantitative trait loci 11 (QTL-11) BC1 populations. Individuals from each QTL-11 population having a high percentage OH88119 were selected and advanced.

    Download as PowerPoint

    Inbred backcross evaluation.

    Disease pressure across all field trials every year was uniform and severe. Disease severity allowed us to easily discriminate between the resistant and susceptible controls, and data for backcross selections displayed a distribution that approached normal. The mean Horsfall–Barratt scales for our susceptible parent, OH88119, against X. perforans, X. gardneri, and X. euvesicatoria were 7.8, 8.9, and 5.7, corresponding to mean percentages of 75.7, 86.1, and 37.1%, respectively. By comparison, resistant parents rated 3.1, 3.9, and 2.1 for OH087663, corresponding to mean percentages of 5.0, 10.8, and 2.9% when challenged with X. perforans, X. gardneri, and X. euvesicatoria, respectively. Ratings for 01-BR-7087 were 4.0, 5.4, and 3.3, corresponding to 10.8, 28.7, and 6.9%, respectively. Horsfall–Barratt ratings for FG12-433E-43 were 3.0 for X. perforans, 3.1 for X. gardneri, and 3.9 for X. euvesicatoria, corresponding to 4.7, 5.5, and 9.6%, respectively.

    The distribution of data and the normality of the residuals were evaluated using Horsfall–Barratt scaled data, Horsfall–Barratt scale converted to mean percentage disease, and the arcsin of the mean percentage disease. Results of this evaluation suggested that either the original Horsfall–Barratt scale or the arcsin of the percentage disease was best suited for additional analysis. For both Horsfall–Barratt scale and the arcsin of the percentage disease, 21 of 24 of the models met the assumptions of normality. In contrast, data converted to the mean percentage violated the assumptions of normality in 10 of 24 models. The general conclusions regarding rank, mean separation, and effectiveness of QTL-11 introgressions did not change significantly despite the data conversion and transformation. We, therefore, present analysis of the original Horsfall–Barratt scaled data, because they better fit the assumptions of the ANOVA and provide a more intuitive interpretation relative to the arcsin transformation of mean percentage disease.

    ANOVA was used to determine significant differences between our main effects, QTL introgression, family (independent selections of QTL introgressions), and blocks (within field variation). Across most trials, significant differences were detected owing to QTL introgression at P ≤ 0.0001 (Table 2). These effects are maintained through the breeding process as material was advanced from BC1 families to BC3 selected lines. Family was significant for some BC1 families (BC1S2 and BC1S4), but the importance of family differences decreases, becoming insignificant, as material was advanced to BC3S1 (Table 2). The block effect was insignificant in most trials, suggesting that fields were uniform and that environmental effects were minimal compared with genetic effects.

    Table 2. Analysis of variance of Horsfall–Barratt disease ratings for Xanthomonas gardneri, X. perforans, and X. euvesicatoria for 2016, 2017, and 2018 field trials


    QTL-11A from source OH087663 (tracing to Hawaii 7998) was effective against all three Xanthomonas species but displayed a higher level of resistance when combined with Rx4/Xv3 (Fig. 2). In 2016, both QTL-11A and QTL-11A with Rx4/Xv3 displayed strong significant differences compared with OH88119, QTL-11B, and QTL-11C NILs (Fig. 2A, E, and I). In 2017, BC1S4 families with QTL-11A or QTL-11A with Rx4/Xv3 were significantly more resistant compared with OH88119 (Fig. 2B, F, and J), but differences were less striking compared with other QTL-11 NILs. In 2017, BC3S1 QTL-11A families were significantly more resistant compared with OH88119 in X. perforans (Fig. 2G) and X. euvesicatoria (Fig. 2K) trials. However, only BC3S1 QTL-11A with Rx4/Xv3 displayed a strong level of resistance in the X. gardneri trial (Fig. 2C). In 2018 BC3S3, the QTL-11A combined with Rx4/Xv3 selection displayed the lowest mean disease rating comparable with QTL-11C (Fig. 2D, H, and L).

    Fig. 2.

    Fig. 2. Boxplots illustrate the average disease rating and distribution for each of the quantitative trait loci 11 (QTL-11) introgressions. Different letters above each boxplot represent significant differences between genotypes at P = 0.05 using the Tukey’s honest significant difference test. Data are shown for A to D, Xanthomonas gardneri, E to H, X. perforans, and I to L, X. euvesicatoria at different stages in the development of near-isogenic lines (A, E, and I, BC1S2; B, F, and J, BC1S4; C, G, and K, BC3S1; and D, H, and L, BC3S3).

    Download as PowerPoint


    QTL-11B, originating from PI 114490, was effective but provided the least control against multiple Xanthomonas species. In 2016, BC1S2 families displayed statistically significant differences for disease ratings compared with OH88119 in X. perforans and X. euvesicatoria trials (Fig. 2E and I), but they were insignificant in the X. gardneri trial (Fig. 2A). In 2017, BC1S4 families were significantly more resistant than the recurrent parent, OH88119, in the X. gardneri trial; however, the mean disease rating was the highest compared with QTL-11A and QTL-11C (Fig. 2B). In 2017, BC3S1 families were not different from the recurrent parent control in X. euvesicatoria and X. gardneri trials (Fig. 2C and K). In 2018, BC3S3 selections with QTL-11B introgressions displayed significantly more resistance relative to OH88119 in all trials, and they were comparable with QTL-11A but had a higher mean disease rating than QTL-11C (Fig. 2D, H, and L).


    Among QTL-11 NILs, QTL-11C from source LA2533 was effective across all Xanthomonas species and provided the most control in 2018 trials at the BC3S3 stage (Fig. 2). In 2016, BC1S2 families displayed statistically significant differences for disease ratings compared with OH88119 in X. perforans and X. euvesicatoria trials (Fig. 2E and I), although differences were insignificant for trials with X. gardneri (Fig. 2A). Conversely, in 2017, BC1S4 families were significantly different from OH88119 in the X. gardneri trial (Fig. 2B) and the X. perforans trial (Fig. 2F), but they were not significantly different in X. euvesicatoria trials (Fig. 2J). The nonsignificant results reflect variability of introgression performance in different family backgrounds, which decreases during the breeding process (Table 2). In 2017 and 2018, BC3S1 and BC3S3 selections displayed resistance to all Xanthomonas species (Fig. 2C, D, G, H, K, and L). For all trials, QTL-11C displayed a lower disease rating compared with OH88119.

    Gene action for QTL introgressions and Rx-4/Xv3.

    Gene action was evaluated by comparing the mean disease ratings between homozygous introgressions and heterozygous genotypes within BC3S1 families and BC3S3 selections. We compared sources of the QTL-11 introgression and the additional presence of the major gene Rx4/Xv3 in some combinations. In 2017 and 2018, QTL-11B displayed no statistically significant differences for disease ratings between homozygous and heterozygous genotypes, suggesting dominance (Fig. 3). In 2017, QTL-11C displayed dominant gene action, although statistically significant differences were detected between heterozygous and homozygous resistant genotypes in 2018 trials (Fig. 3B, D, and F), suggesting that they acted with additive to dominant action (Fig. 3). In 2017 and 2018, QTL-11A displayed dominant gene action. In 2017 trials, the combination of QTL-11A and Rx-4/Xv3 also displayed a similar disease rating between homozygous and heterozygous genotypes, suggesting dominance. However, in the 2018 X. euvesicatoria trial, QTL-11A in combination with Rx-4/Xv3 displayed additive gene action (Fig. 3F). In 2018, the introgression of Rx-4/Xv3 without QTL-11A was also evaluated. In the X. perforans trial, lines with only Rx-4/Xv3 performed better than the QTL-11A introgression (Fig. 3D). This significant reduction of disease severity provides field-based evidence for the effectiveness of Rx-4/Xv3 against X. perforans T3.

    Fig. 3.

    Fig. 3. Boxplots illustrate the average disease rating and distribution for each of the quantitative trait loci 11 (QTL-11) introgressions, heterozygous genotypes, and near-isogenic lines (NILs) lacking QTL-11 introgressions. A only, B, and C refer to NILs that have only a QTL-11 introgression in the homozygous state. A only Het, B Het, and C Het refer to NILs having a QTL-11 introgression in the heterozygous state. A(+)Xv3 refers to NILs that has both QTL-11A and Xv3 in the homozygous state, and A(+)Xv3 refers to NILs having both loci in the heterozygous state. Xv3 only refers to an NIL that only contains Xv3 without QTL-11. Different letters above each boxplot represent significant differences between genotypes at P = 0.05 using the Tukey’s honest significant difference test. Data are shown for A and B, Xanthomonas gardneri, C and D, X. perforans, and E and F, X. euvesicatoria at different stages in the development of NILs (A, C, and E, BC3S1 and B, D, and F, BC3S3).

    Download as PowerPoint


    As sources of resistance are identified and incorporated into breeding programs and varieties, Xanthomonas populations have also changed (Jones et al. 1998). When these populations gain or lose effectors that interact with the host, resistance may become ineffective (Abrahamian et al. 2018; Stall et al. 2009). To combat this problem, a strategy to combine major resistance genes and quantitative resistance loci that can target multiple Xanthomonas species has been proposed (Sim et al. 2015). Genetic sources Hawaii 7998 (QTL-11A), PI 114490 (QTL-11B), LA2533 (QTL-11C) were previously reported as providing quantitative resistance to control bacterial spot of tomato. In all cases, a major QTL on chromosome 11 was implicated in resistance. It had been unclear which QTL-11 provided the best control against multiple Xanthomonas species.

    This study developed nearly isogenic resources, determined the best introgression for resistance to several Xanthomonas species, and assessed gene action to facilitate the development of advanced breeding material. NILs were developed using a backcross strategy coupled to marker-assisted background genome selection. At the BC1 generation, we were able to select individuals in each QTL-11 population having a BC2 equivalent proportion of the recurrent parent genome. We then evaluated NILs in field trials against X. gardneri, X. perforans (T3), and X. euvesicatoria. In general, all QTL-11 selections and introgressions displayed a higher level of resistance compared with the susceptible control OH88119.

    In general, we found that QTL-11C and QTL-11A combined with Rx4/Xv3 provided the best control for all three Xanthomonas species with the most advanced selections (BC3S3). For example, QTL-11C reduced the percentages of foliar infection to 11.4, 1.9, and 3.8% relative to OH88119, which were 79.1, 91.0, and 72.9% when challenged with X. gardneri, X. perforans, and X. euvesicatoria, respectively. QTL-11A combined with Rx4/Xv3 reduced infection to 11.4, 1.2, and 5.3% when challenged with X. gardneri, X. perforans, and X. euvesicatoria, respectively. For QTL-11A and QTL-11B introgressions, statistically improved disease resistance was also noted, although QTL-11B displayed less efficacy against X. gardneri and X. euvesicatoria in the 2017 BC3S3 trials. A surprising finding was the high level of resistance against X. gardneri imparted by the combination of the gene Rx4/Xv3 and the QTL-11A introgression. By itself, Rx4/Xv3 provides resistance against X. perforans race T3 but not against X. gardneri. QTL-11A provides moderate resistance against X. gardneri. However, the combination of Rx4/Xv3 and QTL-11A displayed the strongest level of resistance to X. gardneri. Additionally, QTL-11A in combination with Rx4/Xv3 displayed strong resistance against X. euvesicatoria. The mechanism of this resistance is unclear, because X. gardneri and X. euvesicatoria lack an AvrXv3 effector (Timilsina et al. 2016). We prefer not to speculate on the underlying biology of this enhanced resistance given the tendency of tomato resistance genes to be clustered and the large number of genes underlying the coupling phase recombination (Andolfo et al. 2014).

    For gene action comparisons, all QTL-11 introgressions displayed additive to dominant effects for resistance. In 2017, gene effects seemed dominant for all three QTL introgressions. In contrast, in 2018, QTL-11A and QTL-11B again displayed dominant gene action, whereas QTL-11C displayed additive effects. It is unclear whether these differences reflect distinct mechanism for each QTL source or variability in the evaluation data. From a breeding standpoint, this information provides knowledge important for the development of hybrid lines.

    Developing tomato cultivars with genetic resistance to bacterial spot has been an ongoing focus for many years. Breeding for resistance to bacterial spot is difficult owing to the diversity of Xanthomonas species, weather conditions that promote disease, and the time that it takes to develop and release new varieties. It is important that new sources of resistance are constantly explored and evaluated against present Xanthomonas populations. The nearly isogenic resources described here suggest that some QTL introgressions provide effective resistance against multiple Xanthomonas species. These resources can be developed rapidly using MAS and background genome selection, allowing for new loci or combinations to be evaluated rapidly. Furthermore, combinations of specific resistance genes and QTL may enhance resistance.

    The author(s) declare no conflict of interest.


    The authors thank Jiheun Cho, Troy Aldrich, Bruce Williams, and Matt Hofelich for their help in field and greenhouse management of the research.

    Literature Cited

    The author(s) declare no conflict of interest.

    Funding: This research was supported by U.S. Department of Agriculture, National Institute of Food and Agriculture award 2014-67013-22410 and Specialty Crop Research Initiative subaward 2015-51181-24312 from the University of Florida.