中文摘要：

利用表征性耳石形态变量（大小变量和轮廓变量）分析方法研究了渤海近岸与长江口水域四个刀鲚（Coiliahas,is）群体的识别问题，对比分析了协方差（ANCOVA）校正法和异速生长（allometricgrowth）校正法在耳石形态分析中去除体长效应的应用效果。研究结果表明，两种体长校正法对四个群体的判别分析结果未产生显著影响，总体识别成功率分别为50．3％和56．5％，群体内个体识别成功率分别为39．5％-90．0％和34．0％-65．1％；长江口水域不同体长组刀鲚个体的耳石形态差异显著，表明其耳石形态特质随鱼类个体生长而变化，协方差校正法和异速生长校正方法的体长组别间的总体识别成功率分别为94．8％和98．8％。整体上，耳石轮廓变量（如矩形趋近率、圆度、分形维数、傅立叶系数等）对刀鲚群体间或体长组别间识别的贡献值高于耳石大小变量（如周长、平均直径、重量、面积等），而二者结合能更有效地识别各地理群体或体长组别个体。

英文摘要：

This study was to discriminate four tapertail anchovy Coilia nasus stocks in the coastal waters of the Bohai Sea and the Yangtze River Estuary of China by means of otolith shape analysis including size variables and shape variables Removing fish length effects on otolith shape analysis and stock identification by both ANCOVA adjustment and allometric growth adjustment were also critically assessed. Results showed that these two adjustment methods achieved an overall classification success （discriminant function analysis） of 50.3% （39.5%--90.0%） and 56.5% （34.0%--65.1%） for the four geographical stocks, respectively, where no significant difference is observed. Otolith morphometrics differed significantly among the three fish length groups of the tapertail anchovy in the Yangtze River Estuary, suggesting that otolith shapes could change markedly as fish grows. ANCOVA adjustment and allometric growth adjustment on fish length produced an overall classification success of 94.8% and 98.8%, respectively, for the three fish length groups in this region. In general, otolith shape variables （e.g., rectangularity, roundness, fractal dimension and the Fourier coefficients） could contribute more to the success of stock identification than size variables （e.g., perimeter, mean diameter, weight and area）, whereas a combination of both kinds of variables could obviously improve the ability of otolith shape analysis to effectively dis- criminate stocks.