Utilização da geoestatística como ferramenta auxiliar em programas de melhoramento genético de milho

• Ano de defesa: 2017
• Autor(a) principal: Silveira, Marco Tulio de Paiva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Programa de Pós-Graduação em Agronomia/Fitotecnia; UFLA; brasil; Departamento de Agricultura
• Palavras-chave em Português: Milho – Melhoramento genético; Solos – Atributos físicos – Distribuição espacial; Geoestatística; Corn – Breeding; Soils – Physical attributes – Spatial distribution; Geostatistics; Fitotecnia
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/15630

Maize (Zea mays) is among the three main cereal species grown in the world, and Brazil occupies a pride of place in world production. The economic importance of corn is characterized by the different forms of its use, ranging from animal feed to the high technology industry. Important technological changes in the crop, obtained through intense genetic plant breeding, have resulted in better adaptability to unfavorable conditions and significant gains in productivity. In plant breeding programs, at certain stages of the research project, the independence between observations based on the concepts of classical statistics can be violated, as there are usually a large number of treatments and little availability of genetic material for propagation, requiring the use of reduced installments. In this way, the homogeneity of the blocks is impaired if the spatial variability of the soil is present. Soil heterogeneity is the main cause of experimental error in field experiments. The objective of this work was to study and describe the influence of the spatial variability of soil chemical attributes in the experimental areas and to evaluate geostatistics as a tool of support in a process of selection of maize genotypes as a function of productivity. The experiment was conducted in two experimental areas at Vitorinha´s Farm, in the city of Lavras – MG. The experiment was in randomized complete blocks design, with 17 treatments and 3 replications, totalizing 51 plots for each area. On the area 1 where soil sampling grid were used with 60 collection points spaced 5,5 x 4,0 m lines and in the area 2 where grid were tested with 65 sampling points spaced 9,0 x 3,0 m lines. It was installed in both areas a grid with 51 points for the plant yield, each point located in the central area of the plot. Geostatistics was used to characterize the spatial dependence and predict attribute values in unsampled locations. It was verified that all the attributes studied in area 1 presented spatial dependence and for Potassium and potential CTC was verified the dependence for area 2. It was observed by isolines maps of the soil quimics parameters that the productivity of some hybrids was affected by soil spatial variability, confirming an association. The spatial dependence found and the similar behavior among the variables allowed to infer that, if an experiment were installed that required homogeneity of the area, the variability found would interfere in the responses of the treatments. The use of geostatistics techniques allowed the identification of superior genotypes in certain nutrient contents.