基于矢耳石形态分析的凤鲚(Coilia mystus) 群体识别研究

以长江口、胶州湾、渤海湾和辽东湾4个地理群体的凤鲚(Coilia mystus)矢耳石样本为研究对象,采用形状指数和椭圆傅里叶系数分析相结合的方法研究了该4个凤鲚群体的矢耳石形态特征及差异性。方差分析结果表明,大部分形态变量存在显著的群体差异,引入协变量(耳石长)之后除形状因子外这种差异依然显著。基于协方差校正的判别分析只保留了耳石重量及9个傅里叶系数用于群体识别,而非参数检验的判别分析则保留了耳石重量、分形维数、环状度、矩形趋近率、圆度和19个傅里叶系数用于群体识别。相应地,基于非参检验的凤鲚群体的总体识别成功率(68.2%)明显高于协方差校正的判别分析结果(46.2%),表明前者比后者更能提高耳石形态分析的群体识别能力。

STOCK IDENTIFICATION OF COILIA MYSTUS USING OTOLITH SHAPE ANALYSIS

In this study, the morphology of the sagittal otoliths of Coilia mystus from four stocks along the Chinese coastal waters (Liaodong Bay, Bohai Bay, Jiaozhou Bay and Changjiang River Estuary) were investigated using otolith shape analysis. Shape indices and Fourier coefficients were used as morphological variables and were statistically analyzed by both ANOVA and ANCOVA (otolith length as a covariate) to examine their differences among stocks. The results revealed that most of the morphometric variables of the otoliths examined in the study differed significantly among the four stocks in both statistical analyses. ANCOVA adjustment and nonparametric test were thereafter run to identify the effective variables for stock identification using discriminant function analysis. ANCOVA adjustment identified only otolith weight and 9 Fourier coefficients, whereas nonparametric test could keep more variables (otolith weight, roundness, fractal di-mension, circularity, rectangularity and 19 Fourier coefficients) as the effective morphological variables of otoliths for identifying stocks. Accordingly, classification success (68.2%) of stock identification based on nonparametric test was higher than that (46.2%) based on ANCOVA adjustment, suggesting that the former statistical analysis could show stronger ability of otolith shape analysis to identify the stocks than the latter one.