Autor(es): Almeida, Raquel Portanova
Data: 2013
Identificador Persistente: http://hdl.handle.net/10400.5/5739
Origem: Repositório da UTL
Assunto(s): grapevine; genetic selection; mixed models; genotypic correlation
Autor(es): Almeida, Raquel Portanova
Data: 2013
Identificador Persistente: http://hdl.handle.net/10400.5/5739
Origem: Repositório da UTL
Assunto(s): grapevine; genetic selection; mixed models; genotypic correlation
Mestrado em Biologia Funcional - Instituto Superior de Agronomia
The ancient Portuguese viticulture led to the accumulation of a large amount of genetic variability within native grapevine varieties. This variability is a high raw material for selection. When performing genetic selection concerning several traits, it is very important to know the genetic correlation between them. Our main purpose was to study whether there are any phenotypic and genotypic correlations between yield and quality traits. The study was conducted in six Portuguese grapevine varieties, in different site/year combinations. Two different approaches were used. Firstly an univariate mixed model was fitted to the data of different traits. When significant genetic variability was detected, the phenotypic correlations between pairs of traits were evaluated. Secondly, two multivariate mixed models were fitted. One considering independence between traits (model MM1) and the other considering dependence between traits (modelo MM2). The efficiency of model MM2 over model MM1 was calcuted using the predicted standard errors of the empirical best linear unbiased predictors of genotypic values of the traits. For the six varieties, the genotypic correlations between traits obtained with model MM2 were always higher than the phenotypic correlations. Model MM2 had always a better fit than model MM1, leading to a more precise and efficient selection.