Evaluation location: Minnesota, United States
In September 2019, leaves were sampled from 100 seedlings randomly selected from each population (two year old plants) or from 10 plants per cultivar. One leaf was sampled per seedling and three leaves were sampled per cultivar plant, all of which were growing in 7.57-L pots in Chanhassen, MN. Plants were exposed to naturally occurring Xanthomonas campestris inoculum under container nursery conditions including overhead irrigation. Populations were completely randomized within the nursery to ensure equivalent exposure to inoculum. The fourth fully expanded leaf was chosen for sampling based on the observation that the leaf spot symptoms were less severe on the top ofthe canopy, whereas leaves further down in the canopy were severely senesced or otherwise deteriorated. Leaves were immediately transported to the laboratory for imaging against a white background with uniform, cool white fluorescent lighting. Images were obtained with a Samsung SM-G950U1 camera (Samsung Electronics Co., Seoul, South Korea) with 4032- × 3024-pixel resolution, a 4.25-mm focal length and saved in the .jpg file format. Images were analyzed using Food Color Inspector software (v4.0; http://www.cofilab.com/portfolio/food-color-inspector/) to determine the percent leaf area affected by lesions and secondary symptoms. This software allows the user to define each of the categories in a training set; then, it assigns all pixelsin the image to one of the categories based on the color using a Bayesian algorithm. One leaf per population was used as the initial training set for the population, which was updated iteratively as needed for each leaf to ensure accurate classifications. Pixels were assigned to four categories: background; healthy leaf tissue; necrotic leaf tissue; and discolored leaf tissue. Discoloration was either chlorosis or other impacted tissue surrounding or immediately adjacent to necrotic tissue; it typically represented the leading edge of the lesion. The percent lea farea affected by leaf spot (necrotic tissue and discoloration) was calculated for each leaf by dividing the number of pixels classified as each of the diseased categories by the number of nonbackground pixels and then multiplying by 100. Population and cultivar means were assessed with an ANOVA and Bonferroni-corrected for multiple comparisons. Each leaf was considered a replicate; therefore, there were 100 replicates for each population and 30 replicates for each cultivar. Several representative infected leaves were submitted to the University of Minnesota Plant Disease Clinic to verify the identity of the pathogen. The pathogen was determined to be Xanthomonas campestris, which was consistent with the expectation. No additional pathogens were detected. See citation for additional details.