采用框架法对不同栖息环境中华绒螯蟹形态差异的比较研究

Comparative study on morphological differentiation of Eriocheir sinensis from different habitats by truss network analysis

为了把握不同栖息环境中华绒螯蟹(Eriocheir sinensis)的形态差异及其在产地鉴别方面的应用潜力,以长江水系(滆湖、骆马湖、固城湖、军山湖、梁子湖围网养殖环境)、黄河水系(东营黄河口自然环境)以及辽河水系(营口、盘锦等稻蟹共作养殖环境)8个不同产地的中华绒螯蟹为研究对象,应用框架法对其背甲、腹甲、侧面以及步足的32个形态学指标进行了调查分析和多元统计学研究。结果显示,4个背甲参数、1个腹甲参数、2个第四步足参数、2个第五步足参数及H值均存在显著的差异。基于32个特征的判别分析时初始产地鉴别正确率可达83.1%,但交叉验证正确率却仅52.5%。支持向量机分析时的3种分类结果显示,当测试样本为20%时,基于LDA-SVM的模型分类具有最好的产地鉴别效果,训练集正确率为88.1%,测试集正确率也达83.1%。结果表明,传统的形态学框架法具有对不同栖息环境的中华绒螯蟹产地鉴别潜力。

Truss network analysis of 32 measurement parameters was utilized with the Eriocheir sinensis from 8 geographical origins in the Yangtze River system(the seine culture habitats of the Gehu Lake,the Luoma Lake,the Gucheng Lake,and the Junshan Lake,Liangzi lake),the Yellow River system(the natural habitat in the Yellow River estuary of Dongyin city)and the Liaohe River system(the rice-crab coculture habitats in Yinkou and Panjin cities);in order to understand the morphometric differentiation of crabs from different habitats and its application potential in geographic origin determination.The significant differences were found in the morphometric measurements of 4 carapace parameters,1 abdomen parameter,2 fourth pereiopod parameters,2 fifth pereiopod parameters and H value.Based on 32 morphometric features,the linear discriminant analysis obtained 83.1%discriminant correct rate for the carbs from the geographical origins.However,the corresponding cross-validated rate was only 52.5%.In contrast,the approach of support vector machine could get the best correct classification rates for train set and test set based on LDA-SVM model using 20%level of the test sample as high as 88.1%and 83.1%,respectively.The present study suggested that morphometric truss network analysis has the potential to identify the crabs from the different geographical origins.