中文摘要：

采用2010年1-9月采自北部湾及邻近海域流刺网渔获的54尾金线鱼（Nemipterus virgatus（Houttuyn，1782））、71尾深水金线鱼（Nemipterus bathybius Snyder，1911）和65尾日本金线鱼（Nemipterus ja ponicus（Bloch，1791））耳石样本，进行了椭圆傅立叶分析。耳石轮廓经图像处理后进行灰阶转换和二值化处理，消除噪点，然后解析为椭圆傅立叶标码（EFDs）。采用77个EFD系数进行主成分分析和判别分析来区分3种金线鱼的耳石。结果表明，日本金线鱼幼鱼和成鱼耳石在性成熟前后存在显著的体长效应（P〈0．05），金线鱼和深水金线鱼则体长效应不显著（P〉0．05）。因此在耳石椭圆傅立叶分析中，日本金线鱼采用性成熟个体样本，金线鱼和深水金线鱼采用全部样本。主成分分析中，第1、第2和第3主成分，分别解释了总变异的65．1%，15．3％和5．3％，可以将3种金线鱼耳石区分开来。EFD系数的逐步判别分析中，3种鱼的平均正判率达98．6％；交互验证结果显示，3种鱼的平均正判率达97．9％，这表明耳石椭圆傅立叶分析可以作为鱼种识别的有效工具。

英文摘要：

The variability in sagittal otolith contour shape among three sympatric species of the genus Nemipterus： Golden threadfin bream N. virgatus （Houttuyn, 1782） （n = 54）, Yellowbelly threadfin bream N. bathybius （Snyder, 1911） （n=71） and Japanese threadfin bream N. japonicus （Bloch, 1791） （n= 65）from the Beibu Gulf were investigated, using the specimens collected from gillnet caught in Beibu Gulf central and open waters from January to September in 2010. Otolith contours were extracted and dig- itized by image processing, converted the gray scale image to a binary image, removed the noises, and described by elliptic Fourier descriptors（EFDs）. The 77 EFD coefficients were then identified using principal component analysis （PCA） and gradual discrimination analysis （GDA） to differentiate the otoliths of three Nemipterus species. The result indicated that, length effects were significant between immature and mature individuals for N. japonicus （P〈0. 05） on otolith shape, while not significant for N. virgatus and N. bathybius （P〉0. 05）. Thus only mature individuals of N. japonicus were used to analysis, while all individuals of N. virgatus and N. bathybius were used. The result showed that the 1^st, 2^nd and 3^nd components accounted for 65.1%, 15.3% and 5.3% of the total variation respectively. The GDA of EFD coefficients indicated that the average exactitude distinguish ratio of three species was 98. 6%, and the cross validation also got the high ratio 97.9%. It indicates that the otolith elliptic Fourier shape analysis can be provided as a tool in distinguishing Nemipterus species.