Abstract
We used a new approach based on measurements of four scale characters, for identifying Atlantic salmon (Salmo salar) caught in the commercial fishery at St. Ann's Harbour, Nova Scotia, and that were returning to North River. Utilizing Hotelling's T2, we compared the vector of observations of each member of the commercial sample (size = 29) with the vector of means of the sample from the angling fishery on North River (size = 48). Misclassification rate was estimated by the "leave-one-out" technique. The variables chosen for analysis were the measurements of the longest oral radii between each of the first three river and first sea annuli. The analysis indicated that 79–88% of the individuals in the commercial sample could have been migrating to North River. This technique is proposed for cases in which the requirement to have samples from all contributing stocks for discriminant analysis is not possible.