Assessment of phytoplankton class abundance using absorption spectra and chemometrics

In this study, the feasibility of using the in vitro absorption spectra of phytoplankton samples, for determining the relative abundance of specific classes of the phytoplankton, was investigated. Single species cultures of nine algae, representing six phytoplankton classes, were used. Mixtures of algal cultures were also prepared for analysis. In total, 25 samples were examined. Spectra were recorded in the visible wavelength range, 350-770 nm, and evaluated using chemometric methods. First, a principal component analysis (PCA) model was calculated. The score-plot of the first two components showed separation of the different classes, with mixtures of two species appearing between the clusters of each pure culture. Second, a partial least-squares regression, with several independent variables (PLS2), was calculated. The relative phytoplankton class abundances (as a percentage) were used as the dependent variable (y-axis). The average root mean square error of cross-validation (RMSECV) was 8.6%. Hence, for this particular data set, it was possible to estimate the relative abundance of different classes within the samples. The proposed method requires further development, including evaluation of more species representing more phytoplankton classes, and including non-taxonomic features, such as photoadaptation.