Recurrence quantification analysis as a tool for characterization of non-linear mesocosm dynamics

A mesocosm experiment was performed in order to observe the behaviour of an aquatic community under stable external conditions. The hypothesis was that the community would show temporal fluctuations that correspond to the principles of deterministic chaos. Water from a brackish (9 psu) location was kept from 31 March 1989 to 11 July 1990 in a 100 l mesocosm under artificial laboratory conditions to develop a species-depleted climax community of plankton. The mesocosm was reared for a further period of 1764 d (12 July 1990 to 11 May 1995) under controlled light and temperature conditions and sampled in short intervals of 3 to 4 d. During the period of observations there were no changes in species composition, but strong fluctuation in biomass occurred. The biomass data for picophytoplankton, nanophytoplankton, copepods and rotifers were used to check the systems dynamics for chaotic behaviour by recurrence quantification analysis (RQA), a new method which is applied here to analyse non-linear dynamics in marine ecosystems for the first time. Recurrence histograms show a single period of 7 wk of picophytoplankton, a broad range of periods for nanophytoplankton between 4 and 7 wk with isolated peaks at 5 and 7 wk, a broad range of periods for copepods with a peak at 5 wk and 2 minor peaks at 3 and 12.5 wk, and transient behaviour in rotifers with peaks at 5.5 and 11.5 wk, with non-linear interaction between selected bands of periods. A comparison with shuffled time series shows that the system dynamic is deterministic with transients and that a non-stationary trend exists in the phytoplankton time series. No chaotic behaviour could be detected.