Multitrait Mixed Modeling and Categorical Data Analyses of Phenotypic Variances.

Quantitative and categorical data digitally recorded, measured or scored on whole canopies; single plants, leaves, and siliques; and on random seed samples of 224 genotypes in a phenotyping nursery of Brassica napus were used to (1) develop a pyramiding phenotyping model based on multitrait field and laboratory characterization and evaluation data, and (2) account for fixed and random sources of variation and interpret components of phenotypic variance. A photothermal quotient, the soil series identified in the experimental area, and soil apparent electrical conductivity were measured for the whole nursery or for each plot and, in addition to a systematic check variety, were used as covariates in estimating and adjusting for soil spatial variation. We identified a minimum set of traits at different stages of plant ontogeny with maximum discriminating power between genotypes, partitioned total variance for each trait into its components, developed a reliable field phenotyping protocol; and identified and selected genotypes adapted to the short-growing season in the upper Midwest having 2-5% oil content larger than the check variety.