Improved estimates of phytoplankton community composition based on in situ spectral fluorescence: use of ordination and field-derived norm spectra for the bbe FluoroProbe

Abstract

The use of spectral fluorometers for assessing phytoplankton concentrations and taxonomic composition in aquatic environments is increasingly common. However, the accuracy of such assessments suffers because the necessary norm spectra (spectral fingerprints) are derived using selected taxa and laboratory conditions that may not adequately represent the taxa and environmental conditions in the study area. Ordination analysis of raw fluorescence data has been proposed as a better means of interpreting spectral fluorescence data. We applied nonmetric multidimensional scaling and cluster analysis to raw in situ fluorescence data from Sturgeon Bay, a small, mesotrophic embayment of Georgian Bay (Lake Huron) to obtain system-specific norm spectra for the bbe FluoroProbe. The revised spectra gave improved estimates of phytoplankton taxonomy (root mean square error of 10% versus 14%) and of dissolved organic carbon and chlorophyll a concentrations. While promising, this method should be further explored in other systems with different and (or) weaker gradients in phytoplankton biomass and taxonomic composition.