Revealing dynamics of ecological systems from natural recordings

Two series obtained from natural recordings were analysed by non-linear methods, using a previously reported strategy to deal with short and very noisy time-series. Our aim was to show the possibility of characterising the dynamics of ecological systems that retain some information from the past. The first series, on tree-ring thickness variations in birch trees in Finnish Lapland, which were periodically defoliated by parasites, showed evidence of a 4-dimensional strange attractor. Analysis of the second series, on Pacific sardine scale-deposit in sediments, did not lead to a reliable description of the attractor, probably because the series was too short compared to its large topological dimension. Nevertheless, for the two series, the occurrence of peaks in outbreaks was predicted well by the models.