Relative weight (W,) was originally developed to assess the physiological status of largemouth bass. Later it was expanded to other centrarchids and, more recently, to noncentrarchids without consideration of how well the concept may fit other taxa. Our goal was to compare the ability of W, to convey physiological information among species. We considered the following alternative hypotheses: that the relationship of W, to physiological status would vary according to phylogeny, according to ecological specialization, or according to morphology. We collected data (water, protein, lipid, liver, gonad and viscera percentages) from 2 16 fish of four species (bluegill, largemouth bass, black crappie, chain pickerel) from two Virginia impoundments. All but one populationspecific multiple regression model explained a significant amount of variation (F > 2.3 1, P < 0.05). Both bluegill models explained the least amount of variation (adjusted R' < 0.15). Piscivore models fit the data best (adjusted R~ > 0.45). Correlations of lipid to Wr were lowest for bluegills (3 = 0.1 1 and 0.15). Variables describing energy storage and short-term growth were positively related to Wr in all species; however, the strength of this correlation coincided with trophic role. We concluded that differences in physiology and life history affect interpretation of Wr (and other whole-body condition indices) among species. Further, W, may convey less physiological information for ecologically plastic species (e.g., bluegill) if they are treated as a honlogeneous population.
INTRODUCTIONIn fisheries science, the use and application of the condition index of relative