New perspectives on fish movement: kernel and GAM smoothers applied to a century of tagging data on coastal Atlantic cod

Abstract

Tagging–recovery studies have been used successfully on many occasions, and the data analysed in a variety of ways with respect to fish movement. Here we provide a synoptic account of useful statistical techniques for delimiting area utilization and movement patterns of a population of individuals. We focus specifically on kernel smoothing for population utilization distribution and generalized additive models for temporal movement patterns. Furthermore, we use these 2 techniques to re-analyse data from tagging studies on coastal Atlantic cod Gadus morhua, shedding new light on one of the underlying mechanisms determining the population structure of the coastal cod along the Norwegian Skagerrak coast. Cod were found to use small areas. On average 25% of a population used areas smaller than 10 km2, while 95% of a population used areas smaller than 160 km2. The movements displayed were fjord-limited and predominantly within the bounds of the population’s home range. By re-analysing existing data on neutral genetic markers, we show that the cross-distance of the area utilized was smaller than the distance over which pair-wise genetic differences appear.