AN EMPIRICAL-COMPARISON OF STOCK IDENTIFICATION TECHNIQUES APPLIED TOSTRIPED BASS

Managers of migratory striped bass stocks that mix along the Atlantic coast of the USA require periodic estimates of the relative contributi ons of the individual stocks to coastal mixed-stock fisheries; however , to date, a standard approach has not been adopted. We compared the p erformances of alternative stock identification approaches, using samp les taken from the same sets of fish. Reference (known) samples were c ollected from three Atlantic coast spawning systems: the Hudson River, Chesapeake Bay, and the Roanoke River. Striped bass of mixed-stock or igin were collected from eastern Long Island, New York, and were used as test (unknown) samples. The approaches applied were discriminant an alysis of morphometric data and of meristic data, logistic regression analysis of combined meristic and morphometric data, discriminant anal ysis of scale-shape features, discriminant analysis of immunoassay dat a, and mixed-stock analysis of mitochondrial DNA (mtDNA) data. Overall correct classification rates of reference samples ranged from 94% to 66% when just the Hudson and Chesapeake stocks were considered and wer e comparable when the Chesapeake and Roanoke stocks were grouped as th e ''southern'' stock. When all three stocks were treated independently , correct classification rates ranged from 82% to 49%. Despite the mod erate range in correct classification rates, bias due to misallocation was relatively low for all methods, suggesting that resulting stock c omposition estimates should be fairly accurate. However, relative cont ribution estimates for the mixed-stock sample varied widely (e.g., fro m 81% to 47% for the Hudson River stock, when only the Hudson River an d Chesapeake Bay stocks were considered). Discrepancies may be related to the reliance by all of these approaches (except mtDNA) on phenotyp ic features. Our results support future use of either a morphometrics- based approach (among the phenotypic methods) or a genotypic approach based on mtDNA analysis. We further recommend a conservative strategy of reliance on a single approach in tracking changes in relative contr ibutions of striped bass stocks to coastal fisheries.