Temporal predictability of soil microarthropod communities in temperate forests

The temporal predictability of soil microarthropod communities in stable forest ecosystems was analysed using data gathered from the literature and original data. Four studies were found with time series of at least five consecutive years; they all reported data on Collembola from temperate forests (as do the original data included). We applied the time lag analysis (regressions of Euclidian distance between communities on time lag) for detecting directional change of community composition, the partial rate autocorrelation function for detecting density dependence, and we regressed standard deviations of the logarithms of abundance on observation period for detecting the colour of the time series. Collembolan communities did not show directional change over time, and the composition of the community appeared to be predictable at a gross qualitative level: dominant species tend to stay dominant over time. A high degree of density dependence could be detected, which might be responsible for the observed temporal stability of community composition. Intra- and interspecific competition and/or predation may cause deterministic community structure and the results of this study support the increasing body of evidence that Collembola communities are structured by intensive biotic interactions. Collembolan populations did not show any fluctuations characterised by scaling laws. Either there actually exist point attractors and stable equilibria in Collembola (brought about by non-overcompensating density-dependent interactions), or Collembolan community trajectories are restarted every year by external forces. Due to the large time-scale of environmental variation in forests, low-frequency noise in Collembola may be recognisable only in time series spanning at least some decades.