Combining Field Based Data Collection and Process Based Modelling to Improve Understanding of Atlantic Salmon Population Dynamics

The results of a survival analysis of the complete life-cycle of a population of Atlantic salmon at the Girnock Burn, eastern Scotland, are presented for a 40 year period. Survivals are shown to be distinctly non-linear between the stages ova (or spawners) to fry and again from fry to smolt, but very closely linear for the marine phase of the life-cycle.  Stochastic annual variations about these density-dependent responses are large compared to the (de-trended) signals from environmental parameters (temperature, stream-discharge, climate change). Process-based dynamic life-cycle models show that the extent of these density-dependent survivals can easily be mis-estimated if data from only two adjacent sub-populations (c 15 km river),  governed by the same underlying functional relationships but out of phase, are incautiously combined. The consequences are worse if the adjacent populations respond to different process functions. Possible effects of incautiously ‘borrowing’ key parameters from convenient literature sources are briefly illustrated by comparing parameters between the top and bottom of single c 120 km catchments in Scotland. Analogies are drawn with more extreme life-cycle variations between different salmonid species. A judicious combination of theoretical modelling to guide and enhance empirical studies seems a powerful way forward.