A carregar...
Publicação
Modelos estatísticos espaciais para ensaios de populações vegetais
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.36 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Strategies for controlling spatial variation, for quantification of genetic variability and for an
efficient genetic selection in large trials of vegetatively propagated perennial plants were studied.
The mixed model was carefully studied, mainly concerning the variance parameters, the best
linear unbiased predictors of genotypic effects and the heritability.
Methodological studies revealed that when yield data from trials of vegetatively propagated
plants were analysed (case study - grapevine clones), the decomposition of the spatial model-fit
residuals in two components is possible: a spatially independent component that explain 40 to
60% of the total residual variance, and a spatially dependent component, characterized as an
anisotropic power correlation function.
With the aim of identifying the most suitable experimental designs for trials of plant populations,
a simulation a study was carried out in order to assess the comparative efficiency of various
experimental designs, fully replicated and unreplicated. The results obtained with fully replicated
designs indicated a greater efficiency for row-column designs. The use of unreplicated trials as a
way of preserving and quantifying the genetic variability of plant populations was feasible,
mainly when collections are organized according to an alpha-alpha design containing over 250
genotypes and a check plot frequency higher than 33%.
Descrição
Doutoramento em Matemática e Estatística - Instituto Superior de Agronomia
Palavras-chave
grapevine mixed models spatial model experimental design unreplicated trials genetic variability