A METHOD FOR SEPARATING BIMODAL LENGTH FREQUENCY DISTRIBUTIONS IN JUVENILE ATLANTIC SALMON POPULATIONS

Growth studies on fish populations are often complicated by the presence of bimodality in the data. Traditional parametric tests are not applicable in these circumstances and familiar statistics such as mean and variance become difficult to interpret. Recent developments in catastrophe theory allow a new approach to working with multimodal distributions which provides specific parameters that have particular biological meaning. This paper presents an application of the nonmixture multimodal probability density function to length frequency distribution data collected on juvenile Atlantic salmon at the Atlantic Salmon Federation hatchery in St. Andrews, New Brunswick during the fall of 1982. The model was found to be a useful tool for describing the bimodal distributions and provided a set of parameters that can serve as a basis by which different population distributions can be compared. Key words: Fish breeding, bimodal distributions