Marcadores espectrais no visível e infravermelho próximo pra identificação de sementes em progênies de Ricinus communis L.

• Autor(a) principal: Ferreira, Daniel da Silva
• Data de Publicação: 2015
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFPB
• Texto Completo: https://repositorio.ufpb.br/jspui/handle/123456789/16033
• topic: Ricinus communis L.; Classificação de sementes; Espectroscopias no infra-vermelho próximo-NIR; Espectroscopia no visível; CNPQ::CIENCIAS AGRARIAS::AGRONOMIA

The production and availability of seeds of castor bean (Ricinus comunnis L.) with physical, genetic and physiological quality, constitutes a challenge to the productive chain, because, despite their own variability concerning seeds, in many cases it is not possible to distinguish visually some genotypes that which can promote the loss of quality. With the aim of establishing spectral markers based on identification of castor seeds already stable and cultivars generations involving different progenitors using spectroscopy, is that this study was developed. Spectroscopy consists of an analytical method, where it is studied the interaction of electromagnetic radiation with the molecules being explored in this study the visible region VIS-NIR, with approximate absorption range between 400 to 750nm and near shortwave infrared band SWNIR between 750 to 1100nm. In the first study, four cultivars were evaluated: BRS Nordestina, BRS Paraguaçu, BRS Energy, BRS Gabriela and the lineage CNPAM 2009-7. The reflectance spectra were recorded in the range of 750 to 1100nm. For Classification models that were used 50 seeds of each material, being selected for the test suite 36 seeds each. The second study evaluated two genitors being: CNPAM 2012-137 and CNPAM 93-168, and their reciprocal generations of F1, F2, F3 and F4. The reflectance spectra were recorded in the visible band VIS-NIR of 400 to 750nm. For classification models were used 20 seeds of each material, and for the test set 10 seeds of each. The readings were made with the seeds always in the same position. The spectra were preprocessed using the algorithm Savitzky-Golay with window of 11 points and first derivative for baseline correction. By PCA (Principal Component Analysis) was established that both the SWNIR band as on VIS-NIR band in study, the results allowed the distinction between the cultivars and reciprocal generations. The SIMCA model demonstrad promising results for classification to the significance levels of 1.5 and 10% in both tracks of study. The SPA-LDA (Sucessive Projections Algorithm-Linear Discriminant Analysis) was efficient classifing correctly all the samples, both in the SWNIR and in the VIS-NIR band. The methods in study have shown to be promising for the phenotypic differentiation of castor bean seeds and their reciprocal generations in a fast, practical and non-destructive.