Can We Develop Generic BIA Models of Fish Composition?

Bioelectrical Impedance Analysis (BIA) has emerged as a useful tool permitting the estimation of fat, water, and protein levels in fish without the need to sacrifice and conduct proximate composition analysis.  BIA models have been developed for freshwater, marine, and anadramous species to date including several species of salmonids.  However, to date models have been developed only for individual species.  Given the similarity in morphology of species and the scarcity of many salmonid species this study sought to evaluate whether a more robust BIA model developed from multiple species could accurately predict body composition of species included in the model as well as novel species.  To develop the model we included BIA and composition data from models for Brook Trout Salvelinus fontinalis, Rainbow Trout Oncorhynchus mykiss, Brown Trout Salmo trutta, and Chum Salmon Oncorhynchus keta.  Models were developed for all species (Salmonid model) and for the trout species (Trout model).  The resulting models were significant (P< 0.001) and explained much of the variability in the data.  Relative success of the Salmonid and Trout models versus individual species models will be discussed.  We also make recommendations for applying BIA models to rare species for which development of species-specific models is impossible.