Abstract:

This study developed a rapid and accurate method for the determination of total sugar in Goji berries (Lycium
barbarum) by using Fourier transform near infrared (NIR) spectroscopy. A total of 114 samples collected from various
producing areas were detected by NIR spectroscopy for the acquisition of spectra in the integrating sphere diffuse reflectance
mode. The relationship between spectral information and total sugar content was established using a chemometric method.
The effective wavenumbers by X-loading weights and correlation coefficients were selected. A quantitative model for
predicting the total sugar content of Goji berries was developed and verified. The results showed that the model was built
with spectral pretreatment using Norris Gap Derivative, and effective wavenumbers (EWs) of 8 700–4 000 cm-1 chosen by
correlation coefficients. The coefficients of determination for calibration (Rc²) and for validation (Rv²) of the model were
both higher than 0.91 when the number of principal components (PCs) was 8. The above results demonstrate that rapid
determination of the total sugar content of Goji berries by NIR is workable.

Key words: Goji berries (Lycium barbarum), near infrared (NIR) spectroscopy, effective wavenumbers, total sugar, quantitative analysis