Aplicação da espectroscopia NIR e quimiometria para estudos de fenotipagem e de genótipos de Agave spp.

• Ano de defesa: 2023
• Autor(a) principal: Dias Junior, Joabel Freire
• Orientador(a): Simões, Simone da Silva
• Banca de defesa: Diniz, Paulo Henrique Gonçalves Dias , Gondim, Tarcísio Marcos de Souza
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso embargado
• Idioma: por
• Instituição de defesa: Universidade Estadual da Paraíba
• Programa de Pós-Graduação: Programa de Pós-Graduação em Química - PPGQ
• Departamento: Pró-Reitoria de Pós-Graduação e Pesquisa - PRPGP
• País: Brasil
• Palavras-chave em Português: Espectroscopia; Fibra de sisal; Quimiometria
• Palavras-chave em Inglês: Spectroscopy; Chemometrics
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::QUIMICA
• Link de acesso: http://tede.bc.uepb.edu.br/jspui/handle/tede/4734

The sisal fiber, Agave sisalana, can be used for various purposes, including primarily industrial applications and crafts. However, only 3% to 5% of the leaf material is utilized as fiber, while co-products such as liquid fraction, mucilage, and pulp are discarded. After leaf processing, the residues undergo rapid fermentation, causing an increase in acidity in the liquid fraction, which hinders value addition. For the liquid extracts, acidity determination is a critical variable due to fermentation, mass loss, color changes, pH, and conductivity alterations. With the aim of enabling new applications for the sisal liquid fraction, this study aimed to develop strategies for applying Near Infrared (NIR) spectroscopy and chemometrics for predicting acidity in the liquid fraction and classifying Agave spp. genotypes. To achieve this, 40 samples were collected from six genotypes, with seven plants per genotype. The plant leaves were processed to obtain the liquid extract. Physical and chemical determinations were performed on the samples in their natural state and after treating the extract with a chemical preservative. Electric conductivity, pH, and mass loss were measured daily for 51 days. Subsequently, total acidity was determined through conventional methods, and NIR spectra were acquired. Based on Principal Component Analysis (PCA), the trend of distinction among the studied genotypes was evaluated. PCA results indicated that genotypes A, C, and F (1, 7, and 15) have distinct compositions. Supervised pattern recognition models using Linear Discriminant Analysis (LDA) and the Ant Colony Optimization (ACO) algorithm for variable selection achieved a classification rate exceeding 90%. For the models constructed to determine acidity using VIS-NIR spectra and the iPLS and iSPA-PLS algorithms, a satisfactory predictive capacity was attained, with Root Mean Square Error of Prediction (RMSEP) values of 0.870 and 0.771 mg NaOH/mL, prediction correlation coefficients (rpred) of 0.873 and 0.771, and Relative Error in Prediction (REP) of 8.408 and 10.765% for iPLS and iSPA-PLS, respectively. In light of the above, it can be affirmed that the application of NIR technology presents a fast, efficient, and accurate technique for acidity determination and classification of Agave spp. genotypes.