Repeatability of traits evaluated in a split-plot or factorial experiment

Abstract

In this paper repeatability expressions are derived and their respective ANOVA estimators obtained by using split-plot and factorial models, both in a randomized complete block design. The paper also considers different fixed and random effect models and their assumptions and restrictions. Repeatability estimates, such as the correlation between successive measurements (over time) of the same genotype, always have the same value regardless of the model used, and this allows repeatability to be calculated using models based on the mean of the experimental units (mean of blocks) of each genotype in each time. This independence was not observed for repeatability estimates at the upper limit of broad-sense heritability based on the mean of successive measurements (over time) of the same genotype. The repeatability of traits evaluated in experimental trials of different designs is also discussed.