Discrimination of phytoplankton classes using characteristic spectra of 3D fluorescence spectra

The discrimination of phytoplankton classes using the characteristic fluorescence spectra extracted from three-dimensional fluorescence spectra was investigated. Single species cultures of 11 phytoplankton species, representing 5 major phytoplankton divisions, were used. The 3D fluorescence spectra of the cultures grown at different temperatures (20 and 15 degrees C) and illumination intensities (140, 80 and 30 microM m(-2) s(-1)) were measured and their feature extraction methods were explored. Ordering Rayleigh and Raman scattering data as zero, the obtained excitation-emission matrices were processed by both singular value decomposition (SVD) and trilinear decomposition methods. The resulting first principal component can be regarded as the characteristic spectrum of the original 3D fluorescence spectrum. The analysis shows that such characteristic spectra have a discriminatory capability. At different temperatures, the characteristic spectra of Isochrysis galbana, Platymonas helgolanidica and Skeletonema costatuma have high degrees of similarity to their own species samples, while the spectra similarities of Alexandrium tamarense, Prorocentrum dentatum, Pseudo-nitzschia pungens, Chaetoceros curvisetus, Ch. Debilis, Ch. Didymus and Synechococcus sp. are not as significant as the other three species. C. curvisetus, Ch. Debilis and Ch. Didymus, belonging to genus Chaetoceros, have identical spectra and cannot be discriminated at all. Regarding all six diatom species as one class, the average discriminant error rate is below 9%. It is worth mentioning that the diatom class can be distinguished from A. tamarense and P. dentatum, which belong to Dinophyta.