作者：杨兴恺，张兆群 

　　摘要：三维激光扫描技术在骨骼形态学研究中得到越来越广泛的应用。运用激光扫描可得到骨骼三维几何形态学的全部信息，用于骨骼形态的定量分析，对于建立客观的分类指标以及形态生态学研究有重要意义。本研究以大熊猫、亚洲黑熊、棕熊和北极熊为研究对象，为亚洲黑熊、棕熊和北极熊的共计28件头骨和对应下颌建立三维模型，提取熊类下颌的9个形态测量的比例指标和头骨的15个形态测量的比例指标，对下颌和头骨进行数学建模。使用Weka的J48算法，构建决策树，成功实现了模式分类。又在上述三维模型的基础上，确定了23个下颌形态标志点和29个头骨形态标志点；以标志点的三维坐标为初始变量，进行主成分分析，比较4种熊类的下颌和头骨形态差异。结果显示大熊猫的下颌和头骨形态与其他3类熊差别显著；棕熊的头骨和下颌形态介于黑熊和北极熊之间。在此基础上，以同样方法将一例大熊猫小种化石头骨与4种现生熊类头骨形态进行比较，结果显示大熊猫小种头骨形态与现生大熊猫相似，而又不在其种内差异的形态区间内。除去大小因素，大熊猫小种在头骨形态上与现生大熊猫仍有一定差异，略倾向于杂食熊类。 

　　关键词：三维几何形态学，熊，大熊猫，大熊猫小种，决策树，主成分分析 

　　卷期：51卷 04期 

3D geometric morphometrics of some ursid skulls

    YANG Xing-Kai, ZHANG Zhao-Qun

Abstract  The 3D laser scanning technique is nowadays more and more applied in the study of skeleton morphology. Complete 3D model of external geometry can be easily obtained by laser scanning. It contains all the external morphologic information, both size and shape, which is the prerequisite for quantitative study of fossil morphology. The materials in this study include specimens of giant panda, Asian black bear, brown bear, and polar bear. Using 3D laser scanning, the complete 3D models of skulls and mandibles of 28 individuals are built. Nine mandible morphological indexes and 15 skull morphological indexes are selected to build mathematical models. We use J48 algorithm of WEKA to build decision trees for pattern classification. 23 landmarks of mandibles and 29 landmarks of skulls are sorted, of which 3D coordinates are used as initial variables to tell the morphological differences of mandibles and craniums of the 28 individuals in the principal component analysis. The result shows that the morphology of giant panda’s skull and mandible is significantly different from the other three ursid forms; the morphology of brown bear’s skull and mandible is in between the black bear and polar bear. The overall cranial morphology of Ailuropoda microta is similar to that of giant panda, but also out of the intraspecific variation, which is another evidence of the durophagous feeding behavior of this early Pleistocene giant panda. The methods of 3D modeling, decision tree and principal components analysis introduced herein promise future application in morphologic studies.

Key words  3D geometric morphometrics, bear, giant panda, Ailuropoda microta, decision tree, principal component analysis