THEORETICAL ANALYSIS OF THE EFFICIENCY OF THE LEAST SQUARE METHOD IN THE MASS SELECTION

Abstract

Soil heterogeneity has been universally recognized as one of the main factors that mask the true genotypic value when some type of selection is undertaken. This problem is magnified when the genotypes are evaluated on the basis of the phenotypic value of a single plant. For this situation, there are several selection procedures designed to control, at least partlally, the soil heterogeneity.
One of such procedures involves the selection on the basis of adjusted phenotypic values. This adjustment consists of subtracting from the actual phenotypic values the difference between the phenotypic mean of the plants in the corresponding block and the predicted mean of this block on the basis of a regression model of the phenotypic values on the coordinates of the bidimensional plane.

In this study such method is analyzed, from a theoretical point of view. Main results indicate that: 1) This method does not control the soil variation among blocks and 2) The method makes linear the block effects on the coordinates. A modification to this method is proposed and its relative efficiency is discussed.