Effect of sampling frequency on detection of natural variability in phytoplankton: unattended high-frequency measurements on board ferries in the Baltic Sea

Abstract

This paper demonstrates the importance of adequate sampling frequency in both time and space when collecting data for the evaluation of regional differences and long-term trends in phytoplankton estimates. Variability in phytoplankton in the Baltic Sea has been recorded using automated flow-through analysers on board ferries with an average spatial and temporal resolution of 200 m and one day, respectively. Data collected in the Arkona Sea and in the Western Gulf of Finland during 1994 are used as examples. The analysis shows that changes in phytoplankton biomass may not be reliably detected when temporal and spatial sampling resolution is low. Because a fixed sampling station may not be representative of a sea area, a network of stations covering the entire basin with an interval of a few kilometres, or at least a transect over the area, has to be sampled. The appropriate resolution depends on the variability in Chl a concentrations. In the Arkona Sea, one weekly transect with a 1-km sampling interval is required during bloom periods when variability is highest, especially in spring, to reveal a 5% difference in the average Chl a concentrations between two different years at the 0.05 significance level. In summer, the resolution needed to detect a similar difference is one weekly transect with a 2.5 km sampling interval. Such high spatial and temporal sampling frequencies are difficult to obtain by traditional methods and the use of unattended recordings on board “ships-of-opportunity” may solve this problem.