A cluster based approach to genotype selection using BLUPs

DI RIENZO, J.; BALZARINI, M.; GONZALEZ, L.; TABLADA, M.; BRUNO, C.

Cairns

Conferencia; XXII International Biometric Conference; 2004

International Biometric Society

In agriculture research, the look for new varieties results in extensive comparisons of several genetic materials. The evaluation is usually carried out by means of multiple comparison methods of yield means. A common strategy to select the most promising genotypes is to use Fisher´s LSD multiple comparison procedure. The mean comparison approach relies on the framework of fixed effect models. However, during breeding program stages in which the experimental material is obtained by random selection of a larger genotype population, the genotype effects might be regarded as random variables instead of fixed effects. Under a random effects model for genotypes, comparison of yield means is not the best statistical approach to select genotypes. The best linear unbiased predictors (BLUPs) of the genotype effects are instead recommended. This presentation shows a convenient way to order genotypes BLUPs based on a hierarchical cluster analysis using the matrix of standardized pairwise BLUP differences. We present a method to choose, in the resulting dendrogram, the cluster that contains, with high probability, the group of best genotypes.