Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Scott Milligan

Second Advisor

Manjit Kang


The plant breeding process begins with the selection of parents and crosses. Promising progeny from these crosses progress through a series of selection stages that typically culminate in multi-environment trials. I evaluated best linear unbiased predictors (BLUP), other predictors and prediction models at the initial (cross prediction), early replicated testing and late (multi-location) stages of a sugarcane breeding selection cycle. Model and predictor accuracy was assessed in the first two stages by using cross-validation procedures. I compared statistical models of progeny test data in their ability to predict the cross performance of untested sugarcane crosses. Random parental effect predictors and a random cross effect predictors were compared to mid-parent values (MPV) derived from a fixed female-male parental effect model. The cross effect model was evaluated with and without incorporating the genetic relationships among tested crosses into the BLUP derivation. Models with BLUP-based predictors showed smaller mean square prediction error and higher fidelity of top cross identification than the MPV for all traits evaluated. The MP-BLUP was consistently the best one. Prediction of per se (genotype) performance is needed during the selection process and requires combining information from different trials. The study investigated three mixed models involving three versions of BLUPs estimated under different strategies, a fixed least squares genotype means model, and four check-based methods for combining information at early replicated stages. BLUP-based predictors were superior to the currently used predictor (average percent of check cultivar). In addition, BLUP accuracy was not dependent on check values. In later selection stages, when few and highly selected genotypes are evaluated, genotype effects may be assumed fixed. By assuming genotype-by-environment interaction effects as random, the modeling of the covariance matrix allowed direct estimation of stability and genotype-by-environment measures. Closely related mixed models involving covariance parameters related with genotype-by-environment interaction were estimated. The covariance structure of the observations under the mixed models adjusted the genotype mean separation. Stability parameters were integrated into broad (across environment) and narrow (environment specific) inferences about genotype yield performances. A procedure to obtain visual representation of the genotype-by-environment interaction (BIPLOT) under a mixed AMMI model was also derived.