Integrated Chemometric and Spectroscopic Platform for Real-Time Authentication of Nigerian Export Agricultural Commodities at Port-of-Entry Inspection Stations

Nigerian export agricultural commodities including sesame, cocoa, cashew, and groundnut face persistent rejection at international port-of-entry inspection stations due to mycotoxin contamination, adulterant content, and misrepresentation of geographic origin, costing the country hundreds of millions of dollars in annual export value and reputational damage that can only be addressed by scientifically rigorous, rapid authentication methods applicable in the challenging analytical conditions of port inspection environments. This study designs, develops, and validates an integrated chemometric and spectroscopic platform for real-time authentication of four priority Nigerian export commodities, enabling simultaneous assessment of geographic origin, varietal identity, mycotoxin contamination status, and adulteration detection at port-of-entry inspection stations. The platform integrates four complementary rapid spectroscopic techniques: near-infrared reflectance spectroscopy, Raman spectroscopy, laser-induced breakdown spectroscopy for elemental fingerprinting, and front-face fluorescence spectroscopy, deployed in a rugged portable instrument array. Reference datasets are constructed from 480 authenticated commodity samples sourced from all six geopolitical zones over two growing seasons, with laboratory characterisation of fatty acid profiles, stable isotope ratios (delta-13C, delta-15N, delta-34S), mineral element fingerprints by inductively coupled plasma mass spectrometry, and aflatoxin content by immunoaffinity high-performance liquid chromatography as ground truth data. Chemometric models employing random forest ensemble learning, support vector machine classification, and convolutional neural network algorithms trained on spectroscopic data are validated against the reference chemical dataset using external test sets reserved from the calibration phase. Model interpretability is enhanced using Shapley additive explanations to identify wavelength regions and spectral features of greatest predictive importance. An original multi-commodity authentication framework integrating probabilistic geographic origin and adulteration probability indices is developed, along with field validation data from three Nigerian seaport inspection stations. Keywords: chemometrics, commodity authentication, near-infrared spectroscopy, agricultural exports, mycotoxin detection Nigeria