Multidimensional fluorescence fingerprinting for classification of shrimp by location and species

Parallel factor analysis with soft independent modeling by class analogy (PARAFAC-SIMCA) was used to analyze fluorescence data from shrimp extracts (organic and aqueous phases) to create classification schemes for two species of shrimp from four different countries. Twenty-four shrimp (six from each location: Ecuador, Philippines, Thailand, and United States) were studied; two were classified as statistical outliers. Using PARAFAC scores from the two aqueous fluorescent components and the strongest four components from the organic phase, country of origin was correctly identified at the 95% confidence level for all 22 remaining specimens; three false positives, at lower confidence levels than the true positives, were also indicated. A classification scheme which used all eight fluorescent components reproduced the 22 correct classifications and reduced the number of false positives to one. Finally, a scheme using PARAFAC scores from the two aqueous fluorescent components and the strongest four components from the organic phase, designed to classify according to species, produced 22 correct matches with no false positives. Spectral similarities between known chemical species and the components identified by PARAFAC are suggested for most cases. The results indicate that environmental effects appear in the fluorescence fingerprints of shrimp collected in different locations; therefore, fluorescence measurements on shrimp have the potential to permit geographical classification of shrimp or, conversely, to permit inferences to be made about the animal's environment.