Author(s): Sophie Cauvy-Fraunie; Verena Trenkel; Herve Capra; Jean-Michel Olivier; Anthony Maire; Martin Daufresne; Jeremy Lobry; Bernard Cazelles; Nicolas Lamouroux
Linked Author(s):
Keywords: No Keywords
Abstract: Assessments of fish population in large environment are highly challenging and extremely expensive. Indeed, fish populations are estimated using repeated counts (electrofishing, net catches) characterized by a strong over dispersion that provide uncertain estimates of actual densities. Previous studies showed that fish-population time series allow detecting significant trends and shifts linked to continuous and sudden environmental changes (caused by global warming, anthropogenic alterations and restoration actions). However, those ecological time series exhibited also a high interannual variability still poorly understood. In this study, we proposed to use symbolic dynamics approach combined with techniques from Information Theory to identify the processes that drive the strong interannual variability in fish-population time series. We analysed five extensive data sets (aquatic populations in small streams, artificial channels and bypassed reaches of the Rh^one River, the Gironde estuary, and the Bay of Biscay) characterized by long-term ecological time series (～30 years) and including a detailed description of the environmental conditions. We transformed continuous environmental and ecological time series into symbolic sequences based on the time series trajectory at each timepoint and calculated the mutual information between symbolic sequences to evaluate the degree of synchrony between time series. We expected synchronisms between 1) fish populations of the closest sites, 2) juveniles and adults, 3) environmental drivers and fish densities. A lack of significant results for these three evident hypotheses might suggest a too important sampling noise to be able to examine the variability in fish populations at the interannual time scale and a reconsideration of these costly extensive fish samplings.
Year: 2018