Cashew apple quality by near infrared spectroscopy technique

• Ano de defesa: 2016
• Autor(a) principal: Samamad, Nancy Taera Ibraimo
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Viçosa
• Palavras-chave em Português: Annacardium occidentale; Cajú - Tecnologia pós-colheita; Cajú - Qualidade; Cajú - Propriedades físico-químicas; Espectroscopia de infravermelho; Ciências Agrárias; Fisiologia de Plantas Cultivadas
• Link de acesso: http://www.locus.ufv.br/handle/123456789/8350

The cashew tree (Anacardium occidentale L.) is a plant with great economic importance for the Brazilian Northeast, due to diversity of products and the amount of jobs generated. Growing demand for healthy products associated an increase in table cashew consumption encouraged the development of technologies to monitor quality criteria. These criteria are determined by destructive analyses, which are usually time-consuming, with high costs and do not take into account the individual cashew variability. Aiming to replace these analyses, the near-infrared spectroscopy (NIRS) allows simultaneous and nondestructive determination of multiple quality attributes. NIRS is a rapid technique that correlates the energy absorption properties in regions of the electromagnetic spectrum with the composition and concentration of molecules through regression models developed by chemometrics. The aim of this study was to develop predictive models in NIRS for bench top and portable device to estimate physical-chemical properties such as firmness, pH, total soluble solids (TSS), soluble sugars (SSC), titratable acidity (TA), flavor, ascorbic acid (vitamin C), carotenoids, total flavonoids, total extractable polyphenols (TEP) and antioxidant activity for monitoring cashew apple quality. For the bench device, the models were constructed with 34 genotypes of 17 samples collected from the reflectance spectra mode. The predictive models obtained for firmness and pH showed determination coefficient values for cross-validation (R2CV) of 0.92 and 0.84, respectively, while for external validation or prediction, coefficients of determination (R2P) were 0.87 for firmness and 0.78 for pH. The residual prediction deviation of cross-validation (RPDCV) have presented values of 3.0 e 2.4 and for external validation values of 2.4 and 2.2 were obtained for firmness and pH, respectively, indicating a good predictive ability. For variables of primary metabolism, the obtained values for R2CV were 0.86 for SSC, 0.83 for TSS, 0.90 for TA and 0.80 for flavor and the R2P values were respectively of 0.78, 0.75, 0.85 and 0.73. The presented RPDCV values were 2.6, 2.4, 3.1 and 2.1 for SSC, TSS, TA and flavor, while RPDP obtained values were 2.0 for SSC and TSS, 3.0 for TA and 1.8 for flavor. In secondary metabolism, models with 0.87 values for R2CV were obtained with R2P of 0.85 for vitamin C. These models presented good ability to predict both cross-validation and external validation with RPDCV and RPDP values of 2.6 and 2.8, respectively. Carotenoids models presented R2CV and R2P values of 0.89 and 0.79, with RPDCV and RPDP of 2.9 and 2.0, respectively, while for total flavonóides, models were obtained with values of 0.86 for both R2CV and R2P as well as RPDCV values of 2.6 and 2.0 to RPDP. Models obtained for TEP has presented values of 0.90 for R2CV and 0.89 to R2P and RPDCV values of 3.2 as well as 2.5 for RPDP. Antioxidant activity models were obtained with R2CV and R2P values of 0.87 and 0.81, respectively, and RPDCV values of 2.7 and 2.2 for RPDP. For portable device predictive models, 75 samples from 21 different genotypes were collected and evaluated of which firmness, pH, TSS, TA, flavor and vitamin C presented R2CV values of 0.77, 0.75, 0.90, 0.85, 0.80 and 0.89, respectively, with the average relative error of -1.1%, 0.2%, 0.5%, -1.3%, 2.6% and 4.9%. For these variables were obtained coefficients of determination values for prediction (R2P) of 0.76, 0.72, 0.88, 0.85, 0.82 and 0.83 with standard error of prediction (SEP) coefficient of variability of 18.2%, 3.0%, 5.6%, 19.6%, 15.4% and 12.1%. Besides these, a quality monitoring experiment in cold storage was evaluated by NIRS over nine days. Four genotypes were used with tree repetitions for TSS, vitamin C and pH assessment evaluated in split plot in time.