LAUSR

Abstract A fluorescence fingerprint (FF), also known as an excitation–emission matrix, was used to develop a new method of classifying nutmeg contaminated with aflatoxins. The experimental samples were collected from nutmeg with a wide range of fungal contamination levels. After grinding the samples, FF measurement and high–performance liquid chromatography (HPLC) analysis were carried out. The total concentration of aflatoxins (AFB1 + AFB2 + AFG1 + AFG2; total AF concentration) in the samples was determined using HPLC, which varied from 0.0 to 1781.8 ppb. The FFs of the same samples were measured with a fluorescence spectrophotometer. Although the FF patterns changed with increasing total AF concentration, the trend was unclear at lower concentrations. Threfore, models for predicting or classifying AF contamination in nutmeg from FF patterns were developed using partial least–squares regression (PLSR). The total AF concentrations predicted by the PLSR model showed a positive correlation to the actual concentration with a coefficient of determination of 0.69. Moreover, the variable importance in projection plot indicated that fluorescence from AFs as well as from kojic acid derivatives was important for the prediction of total AF concentration. Finally, it was indicated that samples with a total AF concentration of 10 ppb or higher could be reliably discriminated by setting the threshold to 2.2 ppb in the partial least–squares discriminant analysis (PLSDA). While no false negative was observed in the discrimination, the false positive rate was 13.3%. Future studies on nutmeg samples with different origins are necessary to confirm feasibility of FF as a rapid and simple method of predicting aflatoxin contamination of nutmeg.