基于GLM模型Logistic回归的菲律宾蛤仔地理群体判别

菲律宾蛤仔(Ruditapesphilippinarum)是重要的海洋经济贝类和海水养殖品种,具有较强的表型可塑性特点,不同地理群体之间存在表型差异。本研究根据实测数据建立了胶州湾与天鹅湖菲律宾蛤仔壳长(SL)与壳高(SH)、壳宽(SW)和活体重(W)的回归方程,定量分析了两个地理群体的形态学差异。由于其近似圆形的形态使得所能测量的外部指标较少,限制了传统形态测量学在其地理群体判别中的应用。本研究提出了将传统形态测量学和广义线性模型(GLM)相结合的方法,建立了判别菲律宾蛤仔地理群体的逻辑斯蒂(Logistic)回归方程。结果表明:菲律宾蛤仔的天鹅湖群体比胶州湾群体具有更高的SH/SL和SW/SL比值,壳形椭圆较圆且隆起程度高,"凸"形明显;基于GLM模型的Logistic回归能够快速准确地判别菲律宾蛤仔的地理群体归属,判别正确率高达94.23%。本研究为双壳类物种地理群体判别提供了新的分析方法和技术手段。

DISCRIMINATION BETWEEN GEOGRAPHICAL GROUPS OF RUDITAPES PHILIPPINARUM BY LOGISTIC REGRESSION BASED ON GLM MODEL

Manila clam Ruditapes philippinarum is one of the most economically important marine clams and marine aquaculture species in China. It has a strong phenotypic plasticity and there are some morphological variations among different geographical groups. Based on measured data, regression equations of the shell length (SL) and shell height (SH), shell width (SW) and live weight (W) in Manila clam in Jiaozhou Bay (Qingdao, Shandong, China) and the Moon Lake (or the Swan Lake, a lagoon in Rongcheng, Shandong, China) were established, and the morphological differences between the two geographical groups were quantitatively analyzed. Its approximate circular shape makes it possible to measure less external indicators, thus limiting the application in geographical group discrimination based on traditional morphometry. In this paper, the method combing traditional morphometry and generalized linear model (GLM) was proposed and a logistic regression equation was established to discriminate the geographical population of the clam. The results show that the Moon Lake group had larger SH/SL and SW/SL than those of the Jiaozhou Bay, which indicates that the clam of the Moon Lake group had rounder and more globular shells. Therefore, logistic regression based on GLM model could discriminate the geographical group of Manila clam quickly and accurately, and the discrimination accuracy was as high as 94.23%. This study would provide a new analytical method and technical means for discriminating geographical groups of bivalve mollusks.