Abstract:

The objective of this work was to identify the most effective method to estimate the coefficient of repeatability in progenies of half-sib families of physic nut (Jatropha curcas), and to predict the minimum number of measurements required for some agronomic traits of this oilseed crop. Eighteen half-sib families, in a randomized complete block design with three replicates, were evaluated from 2009 to 2013. In this period, the following traits were evaluated: plant height, stem diameter, number of branches per plant, crown projection on the line, crown projection on the space between rows, and grain yield. The repeatability coefficient (r) was estimated considering different strategies: analysis of variance; principal component analysis (PCA) based on the correlation matrix (PCCOR); PCA based on the matrix of phenotypic variance and covariance (PCCOV); and structural analysis based on the correlation matrix. The PCCOV strategy provides the most accurate estimates of the repeatability coefficient and of the number of measurements (minimum of four), which are required for an accurate evaluation (minimum 80% accuracy) of the traits, due to the cyclical behavior of the progenies.

Index terms:
Jatropha curcas; multivariate analysis; selection efficiency; experimental precision; repeatability.