Combining resistances to major bacterial, fungal, and oomycete diseases of Northeast-adapted tomato

Breeding and testing of Bacterial resistances:

Transfer and field testing of Bacterial Spot Resistances on Chromosome 11.

We successfully used tandem foreground and background marker-assisted selection to rapidly introgress the CH11 resistances from the OH parent OH087663 into the background of the CU fresh market tomato line CU151095-146, creating a set of chromosome 11 lines that combine broad spectrum resistances to Xanthamonas sp., narrow-sense resistance to X. perforans (T3), and strong resistances to fungal/oomycete blights including late blight, stem, foliar, and fruit phases of EB, and Septoria leaf spot. We have now performed two years of performance testing of these lines, with uninoculated trials in New York to evaluate horticultural quality/utility, and in inoculated trials in Ohio (in cooperation with Dr. Francis of OSU) to evaluate the disease resistance of the new lines. Denser genotyping on the CH11 materials, performed by semi-random and targeted quantitative sequencing, revealed a small stretch of heterozygosity on chromosome 1 for some lines that did not appear to be correlated with modifications of important performance-related phenotypes.

There have now been two years of field trials of the chromosome 11 introgression lines. A

Marketable yield of the chromosome 11 lines significantly increased over that of the recurrent parent with the introduction of the QTLs on chromosome 11, going from 14.4 Kg of marketable produce per 6-plant plot to 21.5 Kg (p = 0.03), while there was no significant impact of the introgression on average weight of marketable fruit, the percentage of marketable fruit, or days to maturity (defined as number of days from planting until 50% of the fruit are red-ripe). As expected, there was a significant reduction in fruit size for the subset of chromosome 11 lines that carry an extraneous chromosome 2 donor introgression. In all other traits evaluated, such as foliar coverage and uniformity of set and ripening, the chromosome 11 lines performed at least equivalently to the recurrent parent CU151095-146, indicating the successful introduction of these resistance without negatively impacting quality (linkage drag). Chromosome 11 lines also showed a dense plant growth habit with shorter internodes that may be a challenge to field-based growing environments but could be a benefit to high-tunnel production due to the reduction in plant growth area, shorter internodes, and denser fruit clusters.

Replicated and inoculated field trials in cooperation with Ohio State University revealed a strong and significant effect (Figure 2); P << 0.001, Type II SS, df = 3,27) of the chromosome 11 introgression from OH7663 at reducing disease symptoms caused by X. perforans (T3). Lines which had functional copies of Rx4 and QTL-11 fared significantly better than the recurrent parent check, susceptible checks, and sister lines that had lost the QTL due to recombination and/or independent assortment. This was as expected, due to the nearly complete control of a true R-gene, characterized by a gene-for-gene interaction between a pathogen effector and a plant pathogen receptor protein. Every line that carried Rx4, and several of the recombinant lines that may have retained Rx4 (The marker density was not high enough to specify where the recombination occurred between the markers flanking the two QTL on chromosome 11 due to a lack of polymorphism between the parents that inhibited the creation of additional descriptive genetic markers), displayed infection severity below the population mean, confirming the successful transfer of an effective bacterial spot resistance gene into the Cornell fresh market tomato background.

Results of the inoculated field trials in Fremont OH were less clear against X. euvesicatoria (T1) and X. gardneri (T2). We had not yet introduced the Rx3 locus (on chromosome 5) into the chromosome 11 lines, so resistance to these pathogens would have been derived from the partial resistance conferred by QTL-11 and any marginal effect of the Rx4 gene against non-target pathogen races. Materials that were homozygous for QTL-11 were in fact significantly more resistant (P<0.05, Type II SS, df = 3,27) to X. euvesicatoria than the Cornell recurrent parent, but curiously were not significantly different from two sister lines that had apparently lost the QTL-11 during recombination and/or independent assortment. This could be due to the small sample size (there were only two of these lines entered into the replicated trial). Alternatively, this could occur from the not-unlikely scenario that the QTL was present in these lines despite the genotypes indicated by the flanking markers. Due to our poor knowledge regarding the exact location of this centromeric QTL, and to the absence of nearby and tightly-linked flanking markers because of low polymorphism between the OH and CU parents on chromosome 11, we cannot rule out this possibility. Indeed, these are the challenges of working within elite, cultivated materials that already have high levels of quantitative resistance and that lack genetic polymorphism due to the significant genetic bottleneck that occurred during domestication.

Finally, there was no significant difference among the genotype classes at the 5% Type I error rate for resistance to X. garneri (P = 0.463, Type II SS, df = 3,27). Reasons for this result may include the relatively high level of quantitative resistance already present in the Cornell materials, the lower trait heritability due to the QTL, or the relatively low disease pressure, and hence smaller phenotypic range, in the X. garneri trial.

Transfer of Bacterial Spot and Speck Resistances on Chromosome 5

We successfully transferred the chromosome 5 resistances from OH7536 into the elite fresh-market tomato line CU151095-146. This creating a set of chromosome 5 lines that combine X. euvesicatoria (T1) bacterial spot resistance from the Rx-3 locus, pto-derived resistance to Pseudomonas syringae p.v. tomato, the causative agent of bacterial speck of tomato, and the previously outlined fungal and oomycete resistances in CU151095-146. There are two types of chromosome 5 lines:  those with the full chromosome 5 introgression and therefore have Rx3 and pto but lack SLS-2, and other with recombinant introgressions so that they maintained SLS-2 and now have Rx3, but not Pto. These lineages were one generation behind the transfer of the chromosome 11 bacterial resistance genes to the Cornell blight resistant parental line, and so have only undergone one field tested for horticultural and disease resistance performance in summer 2019 (see Table 2). Denser genotyping on the CH5 materials did not indicate any residual heterozygosity not identified by low-density genotyping throughout the breeding process, indicating the usefulness of the marker-assisted backcross breeding approach used in this work to more rapidly generate near-isogenic lines. We perceived time-saving gains of ~3 backcross generations with the implementation of this tandem MAS approach, compared to MAS backcross breeding with only foreground markers about the target QTL, and traditional MAS breeding based on phenotypic selection alone.

The chromosome 5 introgression lines displayed a larger, more vigorous growth habit with internode lengths longer by approximately 20% over the recurrent parent line CU151095-146. Chromosome 5 lines also displayed later maturity (Table 2), on approximately 14-21 days later than the CH11 lines. We consider this to be significant linkage drag that will have to be broken using the recombinant materials developed during this breeding project (see methods). Table 2 shows that the chromosome 5 introgression lines from the 2019 field were lower yielding than the CH11 lines and hybrids, though the result is not statistically significant in most cases. This trend toward lower marketable yield is likely a reflection of the later maturity of the chromsome 5 lines, as fruit of these entries remain on the vine until the first frosts, and these fruit would not have been included in the marketable yield trial. These lines may also have slightly smaller fruit than the other entries in the trial (Table 2), though once again the result was not statistically different at the 95% confidence level.

Combination of Resistances

We are currently analyzing the yield trial data from the recent 2019 field season on the CH5/CH11 combined lines and so cannot yet provide accurate statistical results; these will be presented in the 2019 VBI report and in an upcoming publication on this work that will reference funding from SARE (Winter 2020). What follows are anecdotal observation on these lines from the field trials. There was little apparent difference in fruit set or size among the lines, however the later maturity of lines homozygous for the CH5 resistance region meant some fruit were still on the vine at the end of the season when the first frosts hit. Because such late-maturing fruit were not harvested, it is likely that lines with the chromosome 5 introgression will show reduced yields by our measures. Such late production may jeopardize field-based tomato production in the northern tier, where field seasons are short. However, these late season varieties may be acceptable under the extended field seasons offered by high-tunnel production systems. The F1 hybrids heterozygous for these resistances fared better, maturing approximately 7 days later than the recurrent parent. This is encouraging, as derived commercial hybrid varieties will likely be heterozygous for all four resistances. Among the 24 plants that that had brought together all of the resistances, we identified four plants which appeared superior to our eyes in terms of yield, fruit set and quality, and plant morphology. These are presently being crossed in the greenhouse to other breeding lines from the CU, NCSU, and OSU breeding programs to develop test hybrids for commercial licensing. This downstream work is funded by USDA specialty-crop grants that were made possible by the preliminary data generated in this work.

Mapping and validation of Early Blight Resistances.

Large effect EB resistance QTL, for both stem and foliar EB infection phases, were identified in our population and mapped to three chromosomes. Several additional minor effect QTL were also identified, but these were not prioritized for future work. The impacts of these QTL were confirmed in the 2018 confirmation populations. These QTL are now in the process of being fine-mapped and have been narrowed to regions of approximately 3Mbp in size. Current QTL intervals are enough for marker assisted selection as performed in this work, but refined intervals will allow plant pathologists to study this system further and elucidate the underlying resistance mechanism.

The three prioritized large-effect QTL were located on the tomato chromosomes 1, 5, and 9, respectively. The chromosome 1 QTL reduced defoliation when fixed for the CU alleles. Regression of Foliar AUDPC against the chromosome 1 QTL alleles gave a statistically significant result both years (p < .01), and reduced defoliation on average by about ~25%. The chromosome 5 QTL also impacted defoliation. When fixed for the OH allele, defoliation was reduced on average ~40% across the two years, and the QTL regression results were statistically significant (p < .01) both years. The chromosome 9 QTL mostly contributed stem lesion resistance, reducing stem lesion AUDPC by an average ~75% across both years. The effect of this QTL was also statistically significant across both years (p < 0.01). There was some indication of QTL x QTL interactions, but these were not significant at the genome-adjusted p-value threshold. Further details regarding these QTL, including a full characterization of the QTL effects and information on predictive SNPs and variant sequences will be published in an upcoming academic journal publication (Winter 2020).

Data from growers helped make breeding decisions.

As part of our grower outreach program we presented the tomato lines and fruit to grower in LI and HV. Growers were also given anonymized tomato samples to taste from the finished CH5/CH11 lines (Figure 3) that were used to help make selections. In addition, we performed two years of on-farm trials on grower fields. Grower field days offered valuable input on the traits they liked and disliked about our breeding selections.

Current and future line releases.

Table 1 below summarizes the first breeding line releases from this project.

Table 1. Performance data on the best lines bacterial spot resistant lines that are now available for licensing. Data are from a 2018 yield trial in Freeville, NY. These lines have the bacterial spot resistance gene combinations Rx-4 and QTL-11 as well as resistance to early blight, late blight (Ph2 + Ph3), Septoria (SLS-2), Verticillium (Ve) and Fusarium (I + I2). Adjusted means for horticultural traits (with Tukey letter groupings, α=0.05) and best linear unbiased predictors (BLUPs) for performance-related traits in the lines currently available for release, as well as for the previously released Cornell recurrent parent (italicized). Negative disease BLUP values indicates that a plant is more resistant; positive BLUP value indicates more susceptible.

*181065-180 has an extraneous introgression on chromosome 2 that reduces fruit size.

Table 2. Preliminary performance data on lines and hybrids developed during this project. Except for the lines also in Table 1, these breeding materials have not yet been released to the breeding community. Data are from a 2019 yield trial in Freeville, NY. The CH5 lines have the bacterial speck (pto) and spot (Rx-3) resistance genes. The CH11 lines have additional bacterial spot resistances (Rx-4 and QTL-11). The hybrids are heterozygous for all four bacterial resistance genes. All lines have resistance to early blight, late blight (Ph2 + Ph3), Septoria (SLS-2), Verticillium (Ve) and Fusarium (I + I2). Adjusted means are for horticultural traits (with Tukey letter groupings, α=0.05) as well as for the previously released Cornell recurrent parent (italicized) and a commercial benchmark (italicized).

*Maturity slope is calculated by fitting a linear model prop_harvest ~ date + e, where "prop_harvest" is the proportion of the total harvest for an entry collected on a particular harvest date, and "date" is the dates of harvest (in order) over the course of the season.

Table 3. Farmer taste test results. The following entries were included in a farmer taste test at the Long Island Research Station in summer 2019. Below are the average scores for the entries. The favorite, in terms of fruit size and shape, was the CH5 x CH11 hybrid 181241 x 181246, while it appears that growers preferred the internal characteristics and taste of the recurrent parent control line, suggesting some linkage drag associated with the resistance introgressions. Further breeding is now being done to improve fruit flavor and texture characteristics.

*1 is least and 5 most preferred.

Future releases planned.

Chromosome 5 lines: Development of lines that combine the Pto (bacterial speck) and Rx-3 (bacterial spot) resistances was completed fall of 2018. The lines were trialed during the summer of 2019, so that we can select lines for release by early 2020.

Combined Chromosome 5/Chromosome 11 lines: Lines that combine all the bacterial speck and spot resistance genes and QTL (on chromosomes 5 and 11) with the fungal/oomycete resistances are being developed. The first F₁ hybrids with all the bacterial resistances combined in the heterozygous state were trialed during the 2019 field season, and we anticipate having homozygous lines available for release by the fall of 2020, or perhaps sooner.