Abordagem multivariada, perfil composicional e construção de curva de calibração para predição do teor de óleo em Jatropha curcas L

• Ano de defesa: 2014
• Autor(a) principal: Corrêa, Aretha Arcenio Pimentel [UNESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Pinhão-manso; Espectroscopia de infravermelho; Cromatografia a gás; Acidos graxos; Análise multivariada; Infrared spectroscopy
• Link de acesso: http://hdl.handle.net/11449/110318

The aim of this study was to establish a calibration curve for prediction of oil content by NIR and the compositional profile by gas chromatography, both of Jatropha curcas, such as analyze multivariate techniques for purposes of selection of promising accessions for breeding programs. For the development of calibration curve, were used seeds of 142 plants, originated of 56 accessions, which had oil extracted by destructive method of seed to obtain reference values, which ranged from 29.32 to 49.69% of oil. The spectra were collected from intact seed’s kernels. Each spectrum was taken with an average of 64 scans with resolution of 16 cm-1.The regions established for the calibration were 8344-7736 cm-1, 6072-4968 cm-1 and 4520-3976 cm- 1. To evaluate the accuracy of the curve, was calculated the correlation coefficient (R2) which was 71.26% and the standard error of calibration (RMSECV) which was 0.138 between the data obtained by the reference method and the NIR. The results indicate that the reference method was satisfactory, showing correlation with the spectral information, making the curve reliable, requiring however, the addition of new data to increase the robustness and extent. Gas chromatography showed that fatty acids with the highest percentages were: oleic (43.45%), linoleic (35.13%), palmitic (13.44%) and stearic (6.30%), concluding that the compositional profile of Jatropha curcas oil is predominantly unsaturated. For multivariate approach data were obtained from the traits: oil content, stem diameter, plant height, length and width of seed. The data were submitted to principal component analysis and cluster analysis by the hierarchical method using Euclidean distance and connection between groups by Ward method. In principal components analysis, three components explained 74.1% of the variance contained in the original variables and were characterized by the variable oil ...