海洋生态与环境科学重点实验室知识库

       耳石是硬骨鱼类的听平衡器官，存在于内耳中，为碳酸钙质结晶。由于耳石的沉积受到遗传背景与鱼体生活环境的共同影响，所以其形态特征具有种属特异性，同时同种鱼类的不同群体间也会形成耳石形态的分化。因此，鱼类耳石形态研究被广泛应用于鱼种及群体判别，并且在多鱼种应用实例中取得了良好的判别效果。然而任何研究手段都存在其适用范围与应用局限性，耳石形态分析也不例外，本文利用已知群体结构的小黄鱼Larimichthys polyactis样本为范例，对耳石形态分析过程中的一系列潜在影响群体判别结果准确性的因素进行了细致探讨，总结了一套严谨有效的分析流程，并应用此流程对中国近海5种常见石首科鱼类：棘头梅童Collichthys lucidus、黑鳃梅童Collichthys niveatus、鮸鱼 Miichthys miiuy、白姑鱼Argyrosomus argentatus、黄姑鱼Nibea albiflora进行了鱼种与群体判别的实证研究。主要结论如下：

1）耳石形态会随着鱼体的生长发生变异，因此为了降低个体大小对形态参数的影响，研究人员通常会使用一些常规的体长校正方法对原始形态参数进行校正。本文利用中国近海渤海与长江口小黄鱼群体样本对基于线性生长和基于异速生长模型的两种体长校正手段进行了实证研究。结果表明此两种常规体长校正手段都不能有效地消除体长对形态参数造成的影响，小黄鱼样品组间 20mm的平均体长差异对耳石形态参数造成的影响即可混淆甚至掩盖群体间真实存在的形态变异从而导致基于耳石形态分析得到的群体判别结论无法客观反映鱼类真实群体结构。因此应用鱼类耳石形态分析技术进行群体判别时，要格外注重对样品组间体长分布差异的控制，以此来减小个体大小对形态变异的影响，从而提高判别的准确性。

2）形状指数（shape indices，如：圆度、椭圆率、矩形趋近率等）是由传统形态参数（如：长、宽、面积、周长等）根据一系列公式计算获得，这类参数涵盖有特定的形态学意义并且与个体大小关系不大。然而本文通过对长江口全长在105-176mm范围内小黄鱼的研究发现形状指数并不能达成消除体长影响的初衷，并且作为多元变量进行统计分析时会对原始参数分布造成潜在影响从而造成多元统计结论的偏差，因此不应将形状指数作为多元变量纳入判别分析中。形状指数相比轮廓描述参数可以提供具体、形象的形态特征，因此可对形状指数做群体间单因素方差分析，用以形象描述群体间耳石形态差异。

3）标准化椭圆傅里叶谐值（NEFDs）可以高效描述图形的轮廓信息，由其所描述的轮廓信息不受图像大小及摆放方向的影响。本文通过对真实小黄鱼样本的耳石图像进行等比例缩放达成了模拟等比（线性）生长的实验设计，对模拟数据的分析印证了标准化椭圆傅里叶谐值对等比生长的校正能力。然而自然界生物体普遍存在的并非等比生长而是异速生长现象，因此本文又利用全长在105-176mm范围内的长江口小黄鱼进行了真实异速生长条件下的NEFDs校正效果评价。结果显示NEFDs难以对自然界普遍存在的异速生长现象进行校正。因此使用标准化椭圆傅里叶谐值作为多元变量进行鱼类群体判别分析时依旧需要控制组间样本的生长差异。

4）应用耳石形态分析对棘头梅童、黑鳃梅童的种间判别得到了97.8%的总体判别成功率，对取自辽东湾、黄河口、胶州湾的棘头梅童和黑鳃梅童进行的群体判别分别取得了67.7%和65.2%的总体判别成功率；对取自黄河口、胶州湾、长江口的鮸鱼、白姑鱼和黄姑鱼进行的群体判别分别取得了71.5%、72.2%、71.1%的总体判别成功率，表明了形态分析作为鱼种及群体判别工具的有效性。本文使用耳石形态分析法将在遗传结构上同属北方区系的辽东湾、黄河口、胶州湾海域的棘头梅童判别为不同群体；将在黄、东海遗传结构无显著差异的鮸鱼以76.9%和68.1%的判别成功率区分为胶州湾和长江口群体，可见耳石形态分析可取得不同于基于遗传学手段所获得的群体结构信息，因此耳石形态分析可以作为遗传结构研究的一个重要补充，为科学高效的渔业资源管理提供依据。

5）基于耳石形态分析的鱼类群体判定容易受到生长差异的影响，这种影响难以通过常规体长校正方法来修正，并且目前不同研究人员并未对样品控制、参数筛选、数据校正等诸多关键分析过程标准化，进而影响群体判别结果的准确性和一致性。因此，在利用耳石形态分析手段进行群体判定时需慎重考虑这一问题，需要从源头上对样品体长分布加以控制此外应当结合耳石轮纹结构、耳石微化学分析、分子标记技术等其它不易受生长差异影响的分析手段进行相互补充。

 

    Otoliths are calcium carbonate crystals functioning hearing and balance of teleosts, which are located in their inner ears. Since the deposition of otolith is affected by the genetics and the living environment of the fish, the morphological characteristics of the otoliths are species- and stock-specific. Therefore, the shape of fish otolith is widely used in fish species identification and stock discrimination. So far, a variety of studies have proved the validity and efficiency of otolith shape analysis for fish stock identification. However, the techniques have their own suitable applying conditions, advantages and disadvantages. And the otolith morphology analysis is no exception. In this paper, a series of factors influencing the accuracy of the stock discrimination in the process of otolith morphological analysis are investigated and discussed. To achieve this goal, length groups from two known geographical stocks of yellow croaker, Larimichthys polyactis, along the Chinese coast were subjected to otolith shape analysis. And 5 Sciaenidae fishes (Collichthys lucidus, Collichthys niveatus, Miichthys miiuy, Argyrosomus argentatus, Nibea albiflora) were used for empirical study. The main conclusions are as follows:

1) In the analysis of otolith shape, fish size is a potential source of variability in morphometric measures because size is associated with individual ontogeny. Thus, removal of the length effect in otolith shape analysis for stock identification using length scaling is an important issue and some statistical procedures have been used to eliminate the size effect of fish individuals. In the present study, two known stocks of L. polyactis were subjected to otolith shape analysis to evaluate whether commonly used size scaling methods could effectively remove the size effect of fish in stock discrimination. The results indicated that the variation of otolith shape caused by intra-stock fish length might exceed that due to inter-stock geographical separation, even when otolith shape variables are standardized with length scaling methods. This variation could easily result in misleading stock discrimination through otolith shape analysis.

2) In order to remove the size-related effects, shape indices (e.g. roundness, circularity, rectangularity et al.) were used in numerous studies. The shape indices were obtained from traditional morphological variables (e.g. length, width, area, perimeter et al.) through specified mathematical equations. However, our present study showed that shape indices cannot achieve the original intention of eliminating the influence of body size only if the relationship between numerator and denominator is a straight line and the y-intercept equals zero and the indices changed raw dada and data error distribution thus misrepresent the true relationship between numerator and denominator. Shape indices differentially transform the metric relationship among samples inter- or intra-size classes make it difficult to evaluate the statistical and biological meaning of quantitative differences. Therefore shape indices should not be included as a multivariate variable in the discriminant analysis. Fourier analysis can be an effective method for describing outline shapes, but does not encourage intuitive understanding of the reason for subtle shape differences. Therefore, shape indices could be analysed with one-way ANOVA to provide an intuitive understanding of the morphological differences among sample sites.

3) Normalized elliptical Fourier harmonics (NEFDs) can efficiently describe the contour information of the shape, which can correct the morphological variation caused by linear growth, but cannot correct the allometry growth which is common in nature. Therefore, it is necessary to minimize the difference of individual growth among sample groups when the normalized elliptical Fourier harmonic is used as a multivariate variable for discriminant analysis.

4) As case studies, morphology analysis was applied to the interspecific discrimination between the C.lucidus and C.niveatus, and a classification success rate of 97.8% was obtained and modest classification success rates (67.7% and 65.2% for C. lucidus and C. niveatus, respectively) for discriminating among the Liaodong Bay, the Yellow River Estuary and the Jiaozhou Bay of the two species were obtained. Stock identification conducted on another three Sciaenidae species among sample sites of the Yellow River Estuary, the Jiaozhou Bay and the Changjiang River Estuary gained modest overall classification success rates of 71.5%, 72.2% and 71.1% for M. miiuy, A. argentatus and N. albiflora. The finding that significant variations of otolith shapes occurred among samples with high degrees of genetic homogeneity (C. lucidus in the Bohai Sea and the Yellow Sea; M. miiuy in the Yellow Sea and the East China Sea) indicated that an otolith shape analysis could yield information complementary to that derived from genetic studies. Therefore, otolith morphology analysis could be used as an important supplementary tool to the study of genetic structure, which was to provide information for efficient management of fishery stocks. 

5) The stock discrimination based on otolith morphology analysis is easily susceptible to growth variation. At present, researchers have different criterions on sample selection and data scaling, which in turn affects the accuracy and consistency of the stock discrimination results. Therefore, conclusions about fish stock structure should be carefully drawn from otolith shape analysis because the observed discrimination may primarily be due to length effects, rather than differences among stocks. A holistic approach involving a broad spectrum of complementary techniques (e.g., otolith shape analysis, otolith microchemistry and genetics) should be applied for accurate stock discrimination and structure analysis, which is essential to effective fishery management.