Explore the Inter- and Intra-Stock Dynamics of Atlantic Salmon Using Nonlinear Forecasting

Broadly distributed in the northeastern America and northwestern Europe, Atlantic salmon Salmo salar experience differential freshwater and oceanic conditions as well as harvesting intensities. These salmon stocks are iteroparous, and they display variable within- and among-stock life history tactics (e.g., variable schedules for smolting and maturation). We hypothesize that the differential environmental and anthropogenic forcing and complex life histories give rise to deterministically nonlinear patterns in salmon population dynamics. To test this hypothesis, we compiled time series (min=20 years) of adult salmon abundance indices, including those of different sea ages or size classes when available, for stocks from Spain (N=5), England (N=9), and eastern coastal Canada (N=46). We examined degrees of nonlinearity using Simplex and S-map. We will identify the embedding dimensions and infer significance of the nonlinear signals within each time series. Furthermore, we will evaluate co-predictability (i.e., a time series’ forecasting power for another; measured by the correlation coefficient between the predicted and observed values of a time series) among different sea ages or size classes within a stock, as well as among stocks within a geographic region.