3月龄牙鲆形态性状对体质量的通径分析

Path analysis of the effects of morphometric traits on body weight for 3-month aged Paralichthys olivaceus

中文关键词: 牙鲆  形态性状  相关分析  通径分析  多元回归方程

英文关键词: Paralichthys olivaceus  Morphometric traits  Correlation analysis  Path analysis  Multiple regression equation

中文摘要:

      随机选取3月龄牙鲆Paralichthys olivaceus 100尾，分别测量全长(X1)、体长(X2)、头长(X3)、体厚(X4)和体高(X5)共5个形态学指标和体质量(Y)。通过SAS软件进行相关性分析和回归分析，分别计算以形态性状为自变量对体质量的通径系数和决定系数，进而对各性状对体质量的影响进行剖分。明确影响3月龄牙鲆体质量的主要形态性状, 为牙鲆选育提供数据支持和理想的测度指标。结果表明，各形态性状间及与体质量的相关系数均达到极显著(P<0.01)。体长、体厚和体高对体质量的通径系数均达到极显著的水平(P<0.01)；所选形态性状对体质量的决定系数R²=0.945，表明所选形态性状是影响3月龄牙鲆体质量的主要性状。通过对各形态性状偏回归系数的显著性检验，剔除不显著的性状自变量后，建立最优的多元回归方程：Y=-4.988+0.763X2 2.997X4 1.089X5，各偏回归系数均达到极显著水平(P<0.01)。

英文摘要:

      One hundred 3-month aged Paralichthys olivaceus were randomly sampled for measuring five morphometric traits including full length (X1), body length (X2), head length (X3), body thickness (X4), body height (X5) and body weight (Y). Through path analysis and multiple linear regression with SAS 8.2 software，this study estimated the path coefficients and determination coefficients of each morphometric trait to body weight and dissected effect of each morphometric trait on body weight as direct effect and indirect effect. This study also confirmed the main morphometric traits influencing the body weight of 3 month aged P. olivaceus and provided data support and optimal measuring indexes for selective breeding of P. olivaceus. The results showed that the correlation coefficients between all morphometric traits and body weight were extremely significant (P<0.01). The path coefficients of body length, body thickness and body width of 3-month aged P. olivaceus to body weight were all extremely significant (P<0.01). The total determination coefficient (R²) of all selected morphometric traits to body weight was 0.945, Indicating that the morphometric traits included in this study were major factors influencing body weight. The insignificant morphometric traits were excluded through coefficient test of partial regression, and then the optimal multiple linear regression equation was established as Y=-4.988+0.763X2 2.997X4 1.089X5, for which all included partial regression coefficients were extremely significant (P<0.01).