The optimum plot size in experiments with rubber trees

• Autor(a) principal: Pimentel-Gomes, Frederico
• Data de Publicação: 2014
• Outros Autores: Rossetti, Adroaldo Guimarães, Viégas, Rosemary Moraes Ferreira
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Pesquisa Agropecuária Brasileira (Online)
• Texto Completo: https://seer.sct.embrapa.br/index.php/pab/article/view/18411
• topic: intraclass correlation coefficient; guard rows; variance of treatment mean; coeficiente de correlação intraclasse; bordaduras; variância da média de um tratamento

The mathematical-statistical study of the problem of optimum plot size in experiments with rubber trees, by the methodology of the intraclass correlation coefficient (p), was carried out with the aid of data from four Brazilian experiments. The methodology applied defines as optimum plot size the number (k) of test trees per plot that minimizes the variance of a treatment mean, for a fixed total number (N) of trees, per treatment. In the case of a complete guard row, it led to recommend of 6 to 12 test trees in two rows, or 9 to 18 in three rows. Plots with one half guard row should have 4 to 10 test plants in two rows, or 6 to 15 in three rows. The usual number of replications in experiments with rubber trees (3, in most cases) is not enough, in general. When small plots are used, greater is the need to increase the number of replications, which should not be less than 5. But even with more replications, trials with small plots allow a considerable reduction in the experimental area, without loss of precision.