太行山猕猴跖骨性差的逻辑回归分析

用猕猴跖骨变量建立性别判别逻辑回归模型,用ROC分析来比较跖骨不同变量、不同侧别和每根跖骨的性差大小.成年跖骨标本45例(31雌,14雄).选择跖骨的7个线性变量.数据分析采用SPSS20.0软件.结果表明:跖骨长度变量是最好的性别标识变量,然后是跖骨头和跖骨底变量,最小是跖骨骨干变量.总体上性别正确判别率较高(84.4%~95.6%).ROC分析显示左右两侧之间的差异很小(AUC分别是0.922和0.956).5根跖骨逐步逻辑回归结果的是MT3,MT4,MT5判别率较高(88.9%~95.6%).MT1和MT2判别率较低(84.4%~86.7%).结果提示:猕猴跖骨性差显著,跖骨的长度变化主要受基因控制,骨干宽度主要受到环境因素的影响.跖骨的性差模式主要反映了遗传、环境与行为因素的交互作用.

Sex Determination of Logistic Regression from the Metatarsal Measurements in Macaac Mulatta

Because of the dichotomous nature of the sex type,logistic regression was used to be able to predict sex based on values of predictor variables of the metatarsals and ROC analysis was used to assess sexing performance at different variables,different sides of the feet and for each of the five metatarsals.The samples selected for this study consist of 45 adult skeletons of Macaac mulatta(31females,14males)from the Taihang Mountains.Seven variables were recorded on each bone of five metatarsals.Data were analyzed using SPSS 20.0version.For the models,the best variable for the metatarsals was LG followed by HH,HW,BH and BW.A logistic regression model was formed to estimate sex,ranging from 84.4%-95.6%for the pooled individuals being correctly classified.There were very small bilateral differences by using ROC analysis,and the AUC values were similar between the two sides,with 0.922 and 0.956 for left and right bones,respectively.For five metatarsals of both feet,the percentage of correct classification was higher for MT3,MT4 and MT5(88.9%-95.6%)and lower MT1 and MT2(84.4%-86.7%).In general,results from this study suggest that the measurements of metatarsals appear to be good discriminators of sex.The lengths of metatarsals are largely genetically determined;and the measures of robusticity of feet are largely subject to environmental modification.These findings suggest that the pattern of the sex differences for metatarsals may well reflect the interplay from both genetic,environmental and behavioral factors.