Continuous monitoring of phytoplankton dynamics in Lake Balaton (Hungary) using on-line delayed fluorescence excitation spectroscopy

1. This study introduces delayed fluorescence (DF) excitation spectroscopy as an on‐line tool for in situ monitoring of the composition and biomass of various colour classes of phytoplankton when they are photosynthetically active (cyanobacteria, chlorophytes, chromophytes and cryptophytes). The DF data are validated by comparison with those from conventional methods (weekly microscopic counts and the measurement of chlorophyll concentration). 2. The composition of phytoplankton as assessed by DF agreed reasonably well with the results from microscopic counts, particularly when differences in chlorophyll‐specific DF integrals of the various colour classes were taken into account. 3. Integrals of DF spectra were converted into concentration of chlorophyll a using empirical factors derived from field data. The value of the conversion factor was nearly twice as high when the relative abundance of cyanobacteria was low (<15%) than when it was high. The converted DF‐chl time series agreed well with chlorophyll measurements particularly when blooms were developing. As the DF method is inherently free of the interference caused by pigment degradation products, the discrepancy between the two data sets increased during the collapse of blooms and when sediment resuspension was intense. 4. Fourier spectrum analysis of the time series of DF‐chl indicated that samples must be taken, at a minimum, every 2–3 days to capture the dynamics of phytoplankton. As a consequence, the dynamics of various algal blooms, including their timing, duration and net growth rate, could be estimated with greater confidence than by using conventional methods alone. 5. On‐line DF spectroscopy is an advanced technique for monitoring daily the biomass and composition of the photosynthetically active phytoplankton in aquatic environments, including turbid shallow lakes. At present, the detection limit is around 1 mg DF‐chl a m^?3 in terms of total biomass but confidence in estimates of phytoplankton composition declines sharply below about 5 mg chl a m^?3. 6. On‐line DF spectroscopy represents a promising approach for monitoring phytoplankton. It will be useful in water management where it can act as an early‐warning system of declines in water quality. In basic ecological research it can supplement manual methods. While default calibration spectra may be acceptable for routine monitoring, we suggest a careful individual calibration of the DF spectrometer for basic research. The statistical methods developed here help to assess the adequacy of various calibration sets.