Search

Bioenergetic models frequently rely on published values or models for estimating the energy density of fish, principally because of the cost and effort of obtaining direct measurements. In this study, we developed empirical models of energy density for free-ranging juvenile coho salmon Oncorhynchus kisutch and Chinook salmon O. tshawytscha sampled at sea from the west coast of Oregon to Kodiak Island, Alaska, and we evaluated the accuracy of published energy density models commonly used for these species. Our analyses showed that the energy density of juvenile coho and Chinook salmon was strongly correlated to percent dry weight and proximate constituents (especially lipid and, to a lesser extent, protein concentrations) but poorly correlated to body size and condition factor. Percent dry weight of whole fish was the single best predictor of energy density for both species, accounting for more than 90% of the variance in energy density. We also found that percent dry weight in the muscle tissue accounted for 65% of the variance in energy density. Changes in energy density mainly reflected changes in lipid composition. These results indicate that accurate estimates of energy density could be obtained at low effort and cost for juvenile coho and Chinook salmon simply by determining the water contents in whole-fish or muscle samples. Published models overestimate the energy density of juvenile coho and Chinook salmon collected from the Pacific Ocean. This may result from the extrapolation of the models to different size-classes, life stages, or habitats. More caution is needed when models are extrapolated to conditions beyond those that were used for their development. [ABSTRACT FROM AUTHOR]