Effect of sample presentation on the near infrared spectra of wild harvest Kakadu plum fruits (Terminalia ferdinandiana)

Abstract

Consumer awareness of the health value of native and tropical fruits and its products, is increasing, where demands for new applications are driving food industry to find novel applications of these fruits as functional ingredients. Kakadu plum fruit (Terminalia ferdinandiana) is endemic to Australia and has been used as a functional food or ingredient. This fruit is sold as whole fruit, puree or powder. Critical to the commercialization of this fruit is the development and use of methods and tools to better monitor and understand the quality of these products throughout the value chain. The aim of this study was to evaluate four scanning positions used to collect near infrared (NIR) spectra of Kakadu plum fruits wild harvested, in order to develop a rapid and high throughput method to monitor fruit chemical composition. A portable NIR instrument (Micro-NIR, Viavi, Milpitas, CA, USA) was used to collect the spectra of whole fruit (four positions) samples from the Kimberley region in Western Australia. The NIR data was analysed using principal component analysis (PCA) and partial least squares (PLS) regression models for moisture and total soluble solids. The results showed that the scanning position affected the performance of the models (standard error of cross validation). It was concluded that NIR spectroscopy is a promising method to assure the integrity of the value chain of these products. However, a more robust protocol must be defined when whole fruit are analysed.