DISCRIMINATION OF MARINE-PHYTOPLANKTON SPECIES THROUGH THE STATISTICAL-ANALYSIS OF THEIR FLOW CYTOMETRIC SIGNATURES

Flow cytometry is a research technique for the rapid analysis of phyto plankton abundance and distribution in marine waters. Although the tec hnique is inherently much faster than optical microscopy for counting phytoplankton, its capability for analysing individual taxa is restric ted, due largely to the lack of suitable data analysis protocols. Thes e protocols. which use univariate and bivariate plots, can typically d ifferentiate a maximum of three or four phytoplankton taxa under labor atory conditions. We present here two multivariate statistical techniq ues, quadratic discriminant analysis (QDA) and canonical variate analy sis (CVA), used to identify 32 species of phytoplankton from how cytom etric data. CVA was shown to be a useful graphical technique for analy sing and displaying data, while QDA was successful at discriminating o ver two-thirds of the phytoplankton species, with classification rates >70%. QDA was also shown to be more than Two orders of magnitude fast er than conventional flow cytometric analyses for discriminating and e numerating phytoplankton species. We discuss the potential of multivar iate analysis, and ways of developing these techniques for the detaile d analysis of complex natural phytoplankton populations in the marine environment.