Potato, an important cash crop, is critically affected by supply chain losses that require attention of both research and industry to ensure sustainable production (SDG 12.3). Within potato supply chain management, dry matter (DM) determination is a key factor affecting the harvest quality, storability, food losses, designated user and tuber prices. Even though rapid methods like near-infrared spectroscopy (NIR) enable operators to carry out fast, reliable in situ measurements, the prediction results are often unsatisfactory due to inherent variability of tuber's composition. Thus, this study aims to evaluate Fourier Transformed-NIR based DM prediction in different potato tissues: i) periderm, ii) cortex, iii) outer medulla and iv) inner medulla of 12 different cultivars. The DM prediction models developed using standard chemometrics approaches (i.e., partial least squares, PLS; and interval-PLS, iPLS) showed that features selection by iPLS allowed for better performances. Among the evaluated tissues, the inner medulla had more precise and reliable measurements with RMSE and R2 prediction values of 0.93% and 0.90, respectively. This information can be used to develop more advanced systems inclusive of portable tools and advanced technologies for in situ evaluation to support operators and ensure better pre- and post-harvest management of potatoes.
Read full abstract