TA percentages were quantified with a Metrohm Robotic Titrosampler System from 1 to 5 ml of the defrosted homogenized fruit juice . SSC was measured from approximately 200 ml of juice on an RX-5000a-Bev Refractometer . Total AC was measured from a 25 ml sample of juice in 200 ml 1% HCl in methanol by reading absorption at a wavelength of 520 nm on a Synergy HTX platereader equipped with Gen5 software . A standard curve was calculated for quantifying AC using a dilution series of pelargonidin from 0 to 300 mg/ml in 50 mg/ml increments, where y was absorption readings for the perlagonin dilution series, s was the slope, x was the concentration of perlagonin inthe dilution series, and i was the intercept. AC was estimated by ðA iÞ=s, where A was the absorption reading.Gray mold incidence ranged from 0.0% to 2.7% among cultivars at 14 dph, a typical postharvest storage duration for LSL cultivars. The five day-neutral cultivars were screened out to 21 dph to develop insights into the post harvest storage limits for modern LSL cultivars . Although the fruit were still marketable at 14 dph, 30 litre plant pots they became marginally marketable or unmarketable by 17–18 dph .
We observed an exponential increase in gray mold incidence beyond 17–18 dph for every cultivar with means ranging from 10.3% to 36.7% among day-neutral cultivars at 21 dph. These studies showed that gray mold was ubiquitous and eventually rendered the fruit unmarketable but that the natural incidence of gray mold was negligible on fruit of LSL cultivars grown in coastal California within the 14 dph storage window . From common knowledge and earlier surveys of phenotypic diversity for resistance to gray mold , we hypothesized that the low incidence of gray mold on commercially produced fruit of LSL cultivars might be genetically correlated with fruit firmness and other fruit quality traits affecting shelf life . Although phenotypic correlations have been reported , genetic correlations have not. The fruit of LSL cultivars are typically much firmer than the fruit of SSL cultivars commonly grown for local or direct-market consumption, as exemplified by Earlimiss, Madame Moutot, and Primella in the present study . The latter are sweeter, softer, and perish more rapidly than “Royal Royce” and other LSL cultivars under normal post harvest storage conditions . To explore how these phenotypic differences affect resistance to gray mold, we developed a training population for GS studies by crossing “Royal Royce,” one of the LSL cultivars assessed for natural infections , with four SSL cultivars , in addition to crossing a pair of LSL parents with differences in fruit firmness and AC .
These full-sib families were phenotyped for resistance to gray mold using an artificial inoculation protocol and genotyped with a 50K Axiom SNP array .Natural infections are too inconsistent and unreliable for analyses of the genetics of resistance to gray mold in strawberry. To overcome this problem, we developed a highly repeatable artificial inoculation protocol for gray mold resistance phenotyping that involved puncturing fruit with a 3-mm probe, propagating spores of a single B. cinerea strain , introducing a known concentration of spores into the wound site, and monitoring disease development on individual fruit stored undisturbed under high humidity . Two quantitative B. cinerea disease symptoms were recorded on multiple fruits harvested from training population individuals: water-soaked LD in mm and the number of dpi when EM was observed on the surface of the fruit. We found that incubating artificially inoculated fruit at 10 C and 95% humidity in the dark yielded highly repeatable results with minimal contamination from other post harvest decay pathogens. LD and EM were recorded daily from 1 to 14 dpi . This protocol produced highly reproducible results with repeatability estimates in the 0.66–0.83 range for LD and 0.68–0.71 range for EM . Although critical for maximizing repeatability, this protocol produced more severe disease symptoms than those commonly observed from natural infection, especially on non-wounded fruit of firm-fruited LSL cultivars .
To study the genetics of resistance to gray mold in strawberry, artificially inoculated fruit of individuals in multifamily and Royal Royce Tangi populations were phenotyped daily for LD and EM over 14 days in cold storage . The speed of fungal development and symptom severity differed among individuals in both populations . The phenotypic extremes we observed are illustrated in time-series photographs of four individuals from the upper and lower tails of the LD and EM distributions in the multifamily population . Lesions became visible and had enlarged to 10.0 mm by 5 dpi in one of the most susceptible individuals , whereas lesions were not visible until 8 dpi and developed the slowest in one of the least susceptible individuals . Lesions spanned the entire fruit surface of the most susceptible individuals by 8 dpi, thereby resulting in significant post harvest fruit deterioration and fungal decay . Consequently, our genetic analyses of LD were applied to phenotypes observed 8 dpi, the last day in the study that resistance phenotypes could be observed for every individual. As expected, our analyses confirmed that resistance to gray mold is genetically complex in strawberry, a finding consistent with observations in other hosts . Although statistically significant differences were observed among individuals for LD and EM in both populations , every individual was susceptible and the phenotypic ranges were comparatively narrow . LDs were approximately normally distributed and ranged from 7.0 to 33.5 mm at 8 dpi in the multifamily population and 7.0 to34.0 mm at 8 dpi in the Royal Royce Tangi population , a few in Supplementary Figure S2. Similarly, the speed of appearance of teresting candidate gene associations were identified when short mycelium on the surface of the fruit was approximately nor- QTL-associated haploblocks were searched in the reference gemally distributed and ranged from 4.0 to 12.5 dpi in the multifam- nome for genes with biotic stress and disease resistance annotaily population and 5.5 to 12.5 dpi in the Royal Royce Tangi tions . A cluster of 11 tandemly population . The repeatabil- duplicated genes encoding pathogenesis-related proteins ities for LD and EM among individuals in these populations sug- were found in close proximity to the most significant SNP associated with a QTL on chromosome 4A . observed for gray mold resistance was genetically caused These genes share sequence homology to FcPR10, an Fragaria chi- . Narrow-sense genomic heritability estimates ranged loensis ribonuclease encoding gene previously predicted to reduce from 0.38 to 0.71 for LD and 0.39 to 0.44 for EM, 25 liter pot plastic which suggested the severity of gray mold disease in strawberry . The other QTL-associated candidate genes thatmight warrant further study encode peroxidases reported to modulate reactive oxygen species levels and inhibit fungal growth during B. cinerea infections and transcription factors reported to signal pathogen-triggered immunity, e.g., WRKY and AP2/ERF , that might target pathogenicity factors, e.g., chitinases and protease inhibitors . While these genes are worthwhile candidates for further study , the effects of the associated QTL were too small and insignificant for direct selection . This was, nevertheless, a first attempt to identify loci underlying resistance to B. cinerea in strawberry through a genome-wide search for genotype-to-phenotype associations in the octoploid genome . Royal Royce, the firm-fruited LSL parent, was more resistant to gray mold than the soft-fruited SSL parents .
Royal Royce was the more resistant parent for both traits in the four full-sib families with that parent . For the 05C197P002 16C108P065 full-sib family, 05C197P002 was more resistant than 16C108P065 for LD and vice versa for EM. The LD and EM differences were highly significant with individuals transgressing the phenotypic ranges of the parents . Transgressive segregation was primarily bidirectional for both traits; however, the EM distributions for Royal Royce Tangi and 05C197P002 16C108P065 were right-skewed toward more resistance and lacked individuals in the lower tails distal to the more susceptible parent . These results suggested that favorable alleles were transmitted by both parents for both traits and that favorable alleles for different loci segregated in most of the families.Genomic prediction accuracies for different WGR methods ranged from 0.28 to 0.47 for LD and 0.37 to 0.59 for EM when estimated by cross-validation from 100% of the subsamples . The differences in accuracy among WGR methods for each trait-population combination ranged from 0.00 to 0.07. RKHS produced the highest accuracy for two of the trait-population combinations and was equal in accuracy to G-BLUP for the other two trait-population combinations. SVM often perfomed intermediate to both G-BLUP and RKHS. The prediction accuracy was greater for LD than EM in the Royal Royce Tangi population, whereas the reverse was observed in the multifamily population. Using cross-validation with 100% of the subsamples, clear differences in prediction accuracy and shrinkage were observed between disease symptoms within and between populations . The prediction accuracy for LD was markedly different between the multifamily and Royal Royce Tangi populations. The GEBV range for LD in the multifamily population was half as wide and the kernel density was flatter and more vertical than that observed in the Royal Royce Tangi population . Notably, the LD phenotypes of the most resistant individuals in the RR Tangi population were well predicted. Their EM phenotypes, however, were not as well predicted—the GEBV range for EM was half that of the phenotypic range and the kernel density distribution was flatter and more vertical . One of the challenges of breeding for resistance to gray mold and other postharvest traits is phenotyping throughput. Collectively, 2563 fruit were harvested and individually stored,tracked, and phenotyped in our study . Our expectation was that multiple fruit/individual was needed to more accurately estimate EMMs and GEBVs and nominally increase heritability. To assess the effect of subsamples on prediction accuracy and explore the feasibility of applying selection for resistance to gray mold from a single subsample/individual, GEBVs and prediction accuracies were estimated from a single randomly selected subsample/individual. We observed a significant decrease in narrowsense genomic heritability for LD and EM in the single subsample analyses, e.g., h ^2 decreased from 0.38 to 0.13 for LD and 0.39 to 0.16 for EM in the multifamily population . Naturally, prediction accuracies plummeted in the single subsample analyses too . This is clearly illustrated by the kernel density distributions for GS accuracy estimated for G-BLUP, RKHS, and SVM by cross-validation with a single subsample/individual . GEBV ranges were narrower and kernel density distributions were flatter and more vertical for the single subsample vs multiple subsample analyses for LD and EM in both populations . Hence, we concluded that breeding values cannot be accurately predicted without subsampling fruit.One of our working hypotheses was that selection for increased fruit firmness and other shelf life-associated fruit quality traits pleiotropically decreased susceptibility to gray mold in strawberry. The additive genetic correlations support this hypothesis and highlight between family differences driven by breeding history, the phenotypic diversity of the parents, and transgressive segregation . The pairwise breeding value distributions further highlight the family structure and phenotypic diversity within and among families. The fruit size, firmness, and TA by LD and EM breeding value distributions for the only elite elite family in our study were distinct from the four elite exotic families . LD and EM were weakly negatively genetically correlated and weakly to strongly genetically correlated with fruit quality traits in directions predicted by our hypotheses. Because gray mold resistance increases as LD decreases and EM increases, signs of the additive genetic correlations have different interpretations for LD and EM and can be antagonistic or synergistic. The interpretation depends on the specific phenotypes targeted for a particular market, e.g., SSL vs LSL. LD was negatively genetic correlated with titratable acidity and positively genetically correlated with pH ; hence, LD increased as titratable acidity decreased and pH increased . The effect of titratable acidity on resistance phenotypes was the motivation for screening additional individuals from the Royal Royce Tangi family, which had a significant genetic variation for TA and yielded more accurate genomic predictions for LD than were observed in the multifamily population . EM was more strongly positively genetically correlated with fruit size and firmness than LD and negatively genetically correlated with total soluble solids .