Optimal monitoring frequency on detection of natural variability in ciliate plankton

Abstract

TheTo determine a desirable sampling interval for analyzing the temporal dynamics of ciliate plankton, a major group of planktonic ciliates, tintinnids were collected almost every day from Aug. 2008 to Nov. 2009 at a fixed sampling station in Korean coastal water. Species composition and abundances of tintinnids were analyzed. The temporal pattern in daily interval was displayed detail changes of species-specific sensitivities to preferred hydrographic conditions. Optimal sampling interval can be considered as an optimal span which controls the degree of the data smoothing. There are two extreme cases. One case is that the smoothing curve becomes the monitoring data curve having every fluctuations as the span becomes the smallest intervals (= 1.5 days). The other case is that the smoothing curve becomes the mean curve of the data as the span becomes the total recording period (442 days). The optimal span, smoothing parameter, was estimated between these two extreme cases by solving the optimization problem which minimizes the mean squared estimation error of the log-transformed data of ciliate abundances. 6.6 days of the optimal span was yielded using the ‘dpill’ function supported by the R “KernSmooth” package. Ten peaks of the ciliate abundance change on the optimal span reflected well dominant species succession. It is therefore suggested that data harvesting in weekly interval is desirable to trace temporal