We tested the ability of a stereo-video method to provide estimates of the metabolic costs of spontaneous swimming by simultaneously video taping free-swimming fish and measuring their oxygen consumption. Three swimming characteristics (speed, turning rate, and acceleration) obtained from image analysis of video recordings explained, on average, 80% of the variability in spontaneous swimming costs. Colinearity among the swimming characteristics prevented us from estimating their relative importance. We converted the swimming speeds recorded by the stereo-video system to metabolic costs using relationships from the literature between oxygen consumption and speed during forced swimming (swimming with constant speed and direction). Spontaneous swimming costs, measured directly by respirometry, were, on average, six times higher than the costs predicted by the forced swim relationships for the same speeds. This difference suggests that the metabolic costs of turning and acceleration can be substantial. Relationships used to convert movements obtained with the stereo-video method into energy expenditure should therefore be developed using spontaneously swimming fish and should include turning rate and acceleration as well as speed.

No abstract