Application of a newly developed statistical procedure to morphometric data from American plaice (Hippoglossoides platessoides) in the Canadian northwest Atlantic

The population structure of American plaice (Hippoglossoides platessoides) in the Canadian Northwest Atlantic was examined by analyzing 11 morphometric features from 18 sampling sites. The sampling sites were determined based on previous knowledge of distribution, spawning, migration, and commercial fishing patterns of American plaice. The analysis was conducted by applying a recently developed multivariate statistical procedure. The procedure is not constrained by the requirement of homogeneity in the residual covariance matrices as in the MANCOVA techniques normally used. Since this requirement is generally violated by morphometric data, this new method for analyzing these types of data is much more robust and provides statistical validity to the results. The analysis indicated that all samples were statistically different from each other, suggesting they were all from separate stocks. While it is believed that then are several self sustaining American plaice stocks throughout the area, the analysis presented here suggesting so many separate stocks must be rejected based on a large body of biological evidence to the contrary. It is concluded, therefore, that morphometric data are very limited in their usefulness for stock identification of American plaice and researchers should be cautious in their approach when using similar data for stock identification purposes. (C) 1998 Elsevier Science B.V.