Influência do modelo de análise estatística e da forma das parcelas experimentais na seleção de clones de Eucalyptus spp
Carregando...
Data
2007-11-23
Autores
Scarpinati, Edimar Aparecido [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Foram instalados três testes clonais de Eucalyptus spp em delineamentos de blocos completos ao acaso em diferentes tamanhos de parcelas experimentais, com 18 tratamentos (18 clones) repetidos em cada experimento. Experimento 1:-Teste clonal em delineamento de blocos ao acaso, com 6 repetições de parcelas retangulares com 42 plantas (6 linhas x 7 plantas); Experimento 2:-Teste clonal em delineamento de blocos ao acaso, com 6 repetições de parcelas lineares de 10 plantas; Experimento 3:-Teste clonal em delineamento de blocos ao acaso, com uma planta por parcela (single tree plot - STP), com 20 repetições (blocos). Aos três anos de idade foi calculado o IMA (incremento médio anual de volume). Os dados do IMA foram avaliados para os valores genotípicos e os componentes de variância para os três delineamentos através de duas metodologias de análise: a) tradicional (ANOVA) e b) modelo misto (REML/BLUP). As estimativas de interesse foram obtidas utilizando os procedimentos GLM e MIXED do software estatístico SAS (2004). Foi efetuada análise de covariância para correção do efeito de competição testando três covariáveis: índice de Heigy, Falha, Média, para os três experimentos e nas duas metodologias de análise. A metodologia REML/BLUP, foi ligeiramente superior à metodologia GLM, em todas as análises. O índice de HEIGY não foi eficiente em nenhuma metodologia de análise para os três experimentos estudados para a característica IMA aos 3 anos de idade. As covariáveis Falha e Média não melhoraram as análises no experimento de parcelas retangulares, mas contribuíram para uma melhora sutil nos experimentos em linha e de planta única. Houve alteração no ordenamento dos genótipos de um delineamento para outro em todas as metodologias de análise. A alocompetição foi a grande causadora de erro experimental entre parcelas...
Three Eucalyptus spp cloning tests were designed in plots at randomized blocks, under different arrays and experimental parcels size, having 18 treatments (18 clones), replicated in each experiment. Experiment 1: Clonal test in randomized blocks with 6 replications in rectangular parcels with 42 plants (6 lines x 7 plants); Experiment 2: Clonal test in randomized blocks, with 6 replications in linear parcels having 10 plants; Experiment 3: Clonal test in randomized blocks in Single Tree Plot – STP, having 20 replications (blocks). After three years, the average of volume annual increment – IMA was calculated. IMA data were evaluated to genotypes value and the variance components to the three designs trough two analysis methodologies: a) Traditional (ANOVA) and b) mixed model (REML/BLUP). Interested estimative were obtained using GLM and MIXED procedures from SAS (2004). Covariance analysis was done to correct the competition effect testing three covaries: Heigy index, Falha, Mean, to the three experiments and in the two analysis methodology. The REML/BLUP methodology was slightly better than GLM methodology in all analysis. The HEIGH index was not apply to none of the three evaluated methodology to IMA characteristic after three years. The Falha 8 and Mean 8 covaries did not improve the design analysis of rectangular parcels, but they contributed to a subtle improve in the line and only plant designs. There was alteration from a design genotype order to other in all analysis methodology. The alocompetition was the main responsible of experimental error between parcels. The rectangular experiment was more efficient than STP, being this one more efficient than the linear parcels to the actual commercial crop system (mono clonal planting). The STP experiment presented higher medium experimental productivity as well higher averages predicted in almost all genotypes tested ...(Complete abstract, click electronic access below)
Three Eucalyptus spp cloning tests were designed in plots at randomized blocks, under different arrays and experimental parcels size, having 18 treatments (18 clones), replicated in each experiment. Experiment 1: Clonal test in randomized blocks with 6 replications in rectangular parcels with 42 plants (6 lines x 7 plants); Experiment 2: Clonal test in randomized blocks, with 6 replications in linear parcels having 10 plants; Experiment 3: Clonal test in randomized blocks in Single Tree Plot – STP, having 20 replications (blocks). After three years, the average of volume annual increment – IMA was calculated. IMA data were evaluated to genotypes value and the variance components to the three designs trough two analysis methodologies: a) Traditional (ANOVA) and b) mixed model (REML/BLUP). Interested estimative were obtained using GLM and MIXED procedures from SAS (2004). Covariance analysis was done to correct the competition effect testing three covaries: Heigy index, Falha, Mean, to the three experiments and in the two analysis methodology. The REML/BLUP methodology was slightly better than GLM methodology in all analysis. The HEIGH index was not apply to none of the three evaluated methodology to IMA characteristic after three years. The Falha 8 and Mean 8 covaries did not improve the design analysis of rectangular parcels, but they contributed to a subtle improve in the line and only plant designs. There was alteration from a design genotype order to other in all analysis methodology. The alocompetition was the main responsible of experimental error between parcels. The rectangular experiment was more efficient than STP, being this one more efficient than the linear parcels to the actual commercial crop system (mono clonal planting). The STP experiment presented higher medium experimental productivity as well higher averages predicted in almost all genotypes tested ...(Complete abstract, click electronic access below)
Descrição
Palavras-chave
Eucalipto, Competição (Biologia), Modelo misto, Competição intergenotípica, Teste clonal, Mixed model, Intergenotipic competition, Clonal test
Como citar
SCARPINATI, Edimar Aparecido. Influência do modelo de análise estatística e da forma das parcelas experimentais na seleção de clones de Eucalyptus spp. 2007. xiii, 58 f. Dissertação (mestrado) - Universidade Estadual Paulista, Faculdade de Ciências Agrárias e Veterinárias, 2007.