2017 ASHS Annual Conference

Plant breeding programs and industry require fast and low-cost screening techniques for determination of nutritional and bioactive compounds in food matrices. In this study, the possibility of near-infrared spectroscopy (NIRS) for predicting alpha-carotene, beta-carotene and total carotenoid from NIRS spectra of winter squash puree, freeze-dried powder, puree methanol extract were evaluated. The carotenoid content was analyzed by Waters UPLC and ranged from 0.3-2.9, 0.1-17.3, and 0.9-18.2 mg/100 g dwt for alpha, beta, and total carotenoids, respectively. The NIR spectra of the squash samples were collected by a Bruker FT-NIR spectrometer, Multi Purpose Analyzer (MPA). The regression models were obtained by partial least square regression. Four combinations of NIRS spectra and UPLC data were tested 1) puree spectra vs UPLC puree data; 2) freeze-dried (FD) powder spectra vs UPLC FD data; 3) FD powder spectra vs UPLC puree data. 4) puree methanol spectra vs UPLC puree data. Best models were obtained using spectra and UPLC data from freeze dried powder. The coefficients of determination (R²) in the cross validation in combination 2 for alpha-carotene, beta-carotene and total carotene were 0.46, 0.83 and 0.83, respectively. The R² for total carotenoids and beta-carotene was 0.75 in models from combinations 1 and 3, but the R² of total carotenoids from the combination 3 model was only 0.2 to 0.33. This study indicates that NIRs is highly effective for predicting beta carotene and total carotenoids in butternut squash, but was not good at predicting alpha carotene. Freeze dried powder was the best material for butternut squash NIRS prediction and methanol extracts were not acceptable. NIRS offers a means of quickly screening butternut squash selections for high pro-retinol A carotenoids.