Dynamics of natural rotifer populations

Abstract

New tools for analyzing ecological time series have permitted the construction of rigorous models from relatively short series. We have applied these techniques to abundance data for nine natural rotifer populations to construct realistic models of their dynamics. Species included are Asplanchna girodi, Filinia pejleri, Keratella tropica, Monostyla bulla, Brachionus rotundiformis, and four other Brachionus species. The overall shapes of the time series were similar with an initial peak followed by oscillations of varying amplitude around a mean of lower population density. Auto correlation functions (ACF) for all populations were positive at small time lags and decayed rapidly to zero. This suggest that these are stationary, exponentially damped time series, fluctuating arround a constant mean with constant variance. The rapid decay of the ACFs indicates that the effect of a perturbation on these populations is quickly removed in one or two days. Phase portrait plots of log current population density vs log lagged density indicate that the time series are stable and non-chaotic. One type of model yielded the highest R2 for four of the nine species and was designated the consensus model. The mean R2 of this model for all nine species was 0.53 with a coefficient of variation of 38%. Lyapanov exponents were strongly negative, indicating that these populations rapidly return to equilibrium after an exogenous perturbation. Rotifer populations appear to be tracking very recent perturbations and their dynamics cannot be predicted from perturbations in the more distant past. We investigated the effects of increasing the level of stochasticity in the consensus model on the length of the growing season and resting egg production. Increasing stochastic variance increased the probability of extremely low population densities, shortening the growing season. In shorter growing seasons, fewer resting eggs were produced, other factors being equal. Counteracting this negative effect, was an increased probability of extremely high populations densities which increased mixis and resting egg production. Constructing models accurately depicting the dynamics of natural zooplankton populations should improve aquatic ecosystem models.