Application of reflectance spectroscopy to identify maize genotypes and aflatoxin levels in single kernels

Spectroscopy is a rapid, non-destructive, and low-cost analytical technique that has the potential to complement more resource-intensive methods. We explored use of spectral methods differentiate maize genotypes assess aflatoxin (AF) contamination in kernels. compared performance two instruments: research-grade ultraviolet-visible-near infrared (UV-Vis-NIR) spectrometer measures reflectance from 304 -1,085 nm, miniaturised NIR 740-1,070 nm. Both systems were used predict AF levels kernels single genotype across 10 genotypes, for latter. A partial least square discriminant analysis model was trained on 70% tested remaining 30%. The classification accuracy 71-72% using UV-Vis-NIR instrument 1,170 kernels, 65-66% device 740 247 AF-contaminated 71, 82, 92% thresholds 20, 100, 1000 ?g/kg, respectively. Using same 872 67, 90, 95% validation sets had similar accuracies 100 whereas higher threshold 20 ?g/kg. Reflectance spectroscopy outperformed visual sorting bright greenish yellow fluorescence test identifying levels. Applying estimate mycotoxin identify could contribute regional toxin surveillance action efforts. Further, AF-associated features grain can reduce exposure.