USE OF OTOLITH MORPHOLOGY IN STOCK DISCRIMINATIONS OF ATLANTIC SALMON (SALMO-SALAR)

Shape of sagittal otoliths was used to develop discriminant function models to classify Atlantic salmon (Salmo salar) by continent and country of origin. The outline shape was digitized with an image processor and used to calculate a complex Fourier transform of the closed contour and two shape indices, rectangularity and circularity. The shape indices and the magnitudes of the first 20 nonzero or nonidentity harmonics were used as input variables. Samples were obtained from known-origin one-sea-winter salmon captured in either Canada or West Greenland during 1 986-88. Classification by continent of origin was designed to discriminate between North American and European salmon. Jackknifed classification efficiency was 88% for the continent model. Classification by country of origin was designed to discriminate between salmon of United States and Canada origin and between salmon of Ireland and United Kingdom origin; these models performed with classification efficiencies of 64 and 69%, respectively. These results suggest that otolith morphology may be an effective tool to identify continent of origin for salmon fisheries management. However, before the model is implemented, the discriminant function training set should be broadened to include a wider cross section of stocks likely to occur in mixed-stock fisheries.