Accelerometry: the key to measuring size-at-age and activity in fish

Abstract

Recent advancements in tracking technology have increased the ability to unravel key parameters affecting behaviour patterns among marine animals where direct observations are scarce. Within the suite of biologging techniques, tri-axial accelerometers are particularly promising for providing data that can link physiological and ecological processes in the context of movement. The objective of my thesis research was to determine how the analysis of accelerometer data can provide reliable and complex information on fish locomotion and behaviour that are relevant for advancing the informed management of commercially and recreationally valued fish. To reach this objective, a high-frequency accelerometer data logger was developed. Based on a series of controlled-environment and field experiments using this technology, a library of automated signal-processing algorithms was developed that relate acceleration signals to rates of activity, swimming speed, size-at-time and behavioural states in a variety of fish species. The algorithms are efficient in extracting behavioural states (feeding, escape, swimming) relevant to energy budgets as well as behaviour associated with spawning and courtship and parasite dislodging while being independent of animal size or tag placement. The most novel contribution is the development of a scaling relationship between tail beat frequency, speed and length in free-swimming fish that is based on accelerometer signal-processing techniques and early theoretical predictions. In the future, the technology and the models may provide valuable input for fish stock modelling by the in situ delivery of more reliable time series of length-at-age, and thus growth rate, in wild fish than that achieved using conventional techniques. Throughout this thesis, accelerometer data analyses challenge the assumption that movement data collected by accelerometer tags represent the normal behavioural repertoire of the tagged animal given low rates of tag sampling frequency currently employed as well as significant behavioural changes caused by tagging and handling stress as demonstrated by post-release fish behaviour modification observed in a field study. This thesis presents a significant contribution to the field through the development of an advanced accelerometer tag and processing algorithms that can be applied to many animal species to advance ecological and physiological theory.