The population has phenotypic diversity for each trait. We will compare the QTLs detected in prior studies to the QTLs detected in this study. For the work, a large number of high-quality SNP markers, the use of high-throughput phenotyping, and a parent with blood orange in the pedigree were used. A unique aspect of this research is the possible detection of QTLs associated with red color in the peel derived from the blood orange grandparent.The pollinations were performed on April 11, 2011, using Kiyomi as the female parent and Amoa 8 as the male. 18 pollinations were made. This resulted in 10 fruit that set seed. These 10 fruit produced 232 seeds that were planted in seed cones in early 2012. Approximately 175 seeds germinated and were grafted by Spring 2013. 131 grafted trees were planted in Field 11B on June 24, 2014, vertical tower for strawberries and 39 grafted trees were planted in Field 12C on July 9, 2015. Trees that were planted in 2015 were held back because they were too small to plant the prior year or had to be repropagated. Any grafted trees not planted were likely discarded because they failed to thrive.
Some of the trees in 11B began to set fruit as early as 2016. Both Carrizo and C-35 citranges were used as rootstocks. These trees were about 6 years old when the phenotyping was done.Fruits were collected from 159 Kiyomi x Amoa 8 hybrids at the UCR Citrus fields during the 2019/2020 season in December to identify loci of interest that could be contributing to variation in fruit quality traits for QTL mapping. First, the identities of the trees were written on each bag, and then the fruits were picked according to their tree identities. 50 fruits per tree were collected in the labeled bags for phenotypic evaluation. Before being sent for phenotypic analysis, each bag was washed in detergent water, rinsed, and left to dry under sunlight to reduce the risk of HLB contamination . Each bag was checked for the presence of materials that may carry a contamination risk, such as leaves or stems, which were then removed. All fruit bags were then transportedfor analysis of phenotypic features to the UC Lindcove Research & Extension Center .The other packline trait is the fruit color measured by InVision® software. For each pixel, they assign it to one of ten color categories. They do this for all pixels in the image. They assign each image a value for each of the ten colors. The value is the percentage of pixels that are a certain color. Therefore, each fruit has a value for the color.
Colors were defined and referenced from color sample standards from The Royal Horticultural Society’s Colour Chart. Each color sample was matched to actual fruit colors chosen according to the color chart used by the UC research community.From the 50 fruit per tree analyzed using the packline, 12 fruit were randomly selected for destructive sampling. Then, these twelve fruits per tree were destructively sampled to obtain measurements for the additional traits. The fruits were prepared for destructive data analysis by cutting them in half at the equatorial region. The number of seeds for each fruit of twelve fruits per tree was counted as the number of seeds visible in an equatorial cut. Then, the total number of seeds of these twelve fruits was divided by the number of twelve fruits and the average seeds number was calculated. Fruit peel thickness was determined by measuring the flavedo portion of twelve fruits per tree by a digital caliper and then the values were transferred to the computer automatically. Average peel thickness data was obtained by dividing the total peel thickness value per fruit by the total fruit number. After measuring the seed and peel thickness of the fruits, a hydraulic fruit press was used to extract fruit juice for further measurements.
For each tree, juice from twelve fruit was combined. Then, the juice weight of each juice sample was measured with a precision balance and expressed in grams. The trait specified as the sugar content is the soluble solids content of the fruit. OBrix value was obtained by measuring the juice from twelve fruits per tree using a refractometer . The total OBrix values per tree were divided by the total number of fruits and the sugar ratios of the tree identities were calculated asOBrix.For Packline data analysis, an average of 50 fruits were analyzed. Then, 12 fruits were selected randomly from 50 fruit bags for destructive data analysis, and the destructive analysis was completed. The data of the fruits whose analysis was completed were submitted as two separate files. After taking the averages of each fruit, both data files were merged according to tree identities to create a phenotype file. The data were transformed using the best Normalize package in R Studio Software for each of the fruit characters. The best Normalize package determined which method could be performed to normalize data for each trait. The distribution of the transformed data was examined. In addition, Spearman correlation coefficients were performed to find the relationship among the phenotypic traits of the fruits. The correlation was visualized using the R package corrplot .Linkage analysis was conducted using a pseudo-test cross strategy in the R/qtl package. SNPs were divided into three categories according to their segregation patterns: AB × AA, AA × AB , and AB × AB . Although some individuals were sampled for DNA extraction, they could not be analyzed for phenotype data because they did not have any fruit. For this reason, the individuals were filtered to keep the ones with both phenotype and genotype data. There were 93 samples with complete genotype and phenotype data for QTL analysis. To perform QTL mapping, the R/qtl package was used to build maternal and paternal genetic maps and identify QTL. A total of 4491 markers are informative for mapping QTL derived from Amoa 8 and 7303 markers are informative for mapping QTL derived from Kiyomi . The recommended quality control procedures were followed according to to filter markers prior to QTL mapping. This includes removing distorted markers, identifying markers incorrectly placed, removing duplicated and switched markers, estimating the recombination fraction between them, calculating a LOD score for the test of r = 0.5 for each pair of markers, and counting crossing over numbers. SNPs with any missing data were removed.Fruit quality traits have a significant effect on consumers in the global industry. Identifying the genetic bases of important fruit quality traits is essential for Citrus breeding. Understanding how mandarin fruit quality traits are genetically regulated and correlated is the first step toward improving marker-assisted breeding programs. The distribution of the offspring and the parents of the population in terms of fruit quality characteristics measured was examined as well as the correlation between these features. The fruit quality traits were analyzed for Kiyomi, Amoa 8, and their outcross F1 mandarin progenies. The phenotypic trait distributions of the parents and progenies were shown as non-transformed in Figure A1, A2, and transformed data in Fig. 9, Fig. 10, Fig. 11, and Figure A3. The distributions were transformed using the Best-normalization package in R, container vertical farming and the best transformation method for each phenotypic trait . Trait distributions were typically unimodal with transgressive segregation evident in most of the measured traits. Transformed phenotypic distributions of destructive traits, which were JW, JV, TA, pH, SC, AC, ASN, and APT, in the populations, transgress beyond the two parents . For example, the average sugar content from 12 fruits of Amoa 8 was 14.6 oOBrix while that of Kiyomi was 11.9 OBrix. The minimum and maximum values of hybrids are 10 and 17.8 OBrix, respectively. The value of Amoa 8 was measured as 3.62 mm and Kiyomi was 4.57 mm for APT . The TA, AC, and pH values, which were responsiblefor the acidity of the fruit, and TA, AC, and pH values of the parents were closer to each other. FW, FV, MajorFD, and MinorFD are traits related to fruit size.
Looking at the distribution histograms of these features, it is seen that these features were transgressively segregated beyond two parents . MajorFD is the measurement of the average diameter of the fruits from the equatorial region. Amoa 8’s MajorFD measured 37.9 cm and 68.58 cm for Kiyomi. The min value for MajorFD was measured as 1.77 cm and the max value was 84.21 cm in the population . The distributions of fruit characters, which are ELG, TEX, OVR, FLT, STA, STS, CAS, SMT, RS, and SS that contribute to the difference in fruit shape and size were also examined . In addition, it is seen that the frequency distributions of the hybrids for these fruit characteristics were often substantially broader than those of their parents . The progenies and the parents were also evaluated for fruit color traits. The distribution of ten features, CMR, CR, CRO, CDO, CO, COY, CY, CYG, CG, and CDG, related to fruit color, which is one of the most important fruit quality traits related to the external appearance of the fruit, was shown in Figure 11. Transgressive segregation was observed in all the color traits, but to varying degrees. For CMR, CRO, CG, and CYG distributions, parental values span the range of progeny values with less transgression. Most progeny values were similar to parental values in CMR, CRO, CG, and CYG distributions. On the other hand, parental values are similar with extensive transgression in progenies in the distribution of CR, CDO, COY, and CY traits. Since the value of Kiyomi was very low, it was not seen in the CDG distribution histogram.The first component separated individuals within the population based on their values for the CY and CO color traits. The second component separated individuals within the population based on their values for the CMR and CG color traits. In addition, PC2 separated the parents , which are known to differ in the level of external red fruit color. The same PC plot, but variations in specific colors are highlighted with the yellow-purple shading. In other words, individuals of the population were separated with CY and CMR colors. Phenotypic variation in color was summarized using PC1 and PC2 values and these values were used in all downstream analyses.Fruits were collected from field-grown trees that were subject to local weather conditions. The aim is to understand how robust traits measurements were across years and focused on two traits – SC and MajorFD. To measure the stability of genetic effects across years, according to the measurements for 2020, twenty individuals were selected to represent the phenotypic extremes for two traits – SC and MajorFD. For each trait, 10 individuals with the highest trait values and 10 individuals with the lowest trait value were identified. SC and MajorFD were re-measured in fruit collected in 2021 for these 40 individuals. The measurements for each trait showed consistency between two years for each group, and hence genetic effects across years are stable . However, differences between the high and low groups were smaller in year 2 and during the year 1 when the groups were identified. This is consistent with an environmental contribution to variation in these traits.The recommended quality control procedures to filter markers prior to QTL mapping were followed according to two references , Broman . These procedures included removing distorted markers, identifying markers incorrectly placed, and removing duplicated and switched markers. After completing these procedures, the recombination fraction was estimated and a LOD score for the test of r = 0.5 for each pair of markers was calculated, and then crossing over numbers were counted. According to the plot of estimated pairwise recombination fractions, there were not any problematic markers. It was most probably that markers were placed in their accurate position on chromosomes and. There were no estimated recombination fractions that were r> 0.5, and no large recombination fractions with large LOD scores. There was a distribution between 0-30 crossing overs among the 96 individuals. Although there were some departures from 1:1 segregation on chromosomes 1 and 9, the estimated genetic maps were good to start constructing QTL mapping.In previous QTL studies associated with fruit size, 8 QTLs for fruit weight and 3 QTLs for fruit diameter were determined by Yu et.al . These QTLs on the scaffolds 4, 5, and 8 of the Clementine reference genome explained phenotypic variance from 15.03 % to 24.6%.