欢迎访问林业科学,今天是

林业科学 ›› 2019, Vol. 55 ›› Issue (1): 38-46.doi: 10.11707/j.1001-7488.20190105

• 论文与研究报告 • 上一篇    下一篇

基于几何形态测量学的天蛾科成虫数字化分类

蔡小娜1, 苏筱雨2, 黄大庄2, 沈佐锐3   

  1. 1. 河北金融学院基础课教学部 保定 071051;
    2. 河北农业大学林学院 保定 071000;
    3. 中国农业大学IPMist实验室 北京 100193
  • 收稿日期:2017-06-19 修回日期:2018-04-07 出版日期:2019-01-25 发布日期:2019-01-06
  • 基金资助:
    河北省高等学校科学技术研究项目(QN2017077);河北省林业科学技术研究项目(1507473);国家林业局948项目(2013-4-75)。

Digital Classification of Sphingid Moths Adults (Lepidoptera: Sphingidae) Based on Geometric Morphometry

Cai Xiaona1, Su Xiaoyu2, Huang Dazhuang2, Shen Zuorui3   

  1. 1. Department of Basic Courses, Hebei Finance University Baoding 071051;
    2. College of Forestry, Hebei Agricultural University Baoding 071000;
    3. IPMist Laboratory, China Agricultural University Beijing 100193
  • Received:2017-06-19 Revised:2018-04-07 Online:2019-01-25 Published:2019-01-06

摘要: [目的]对天蛾科10种成虫的前翅进行几何形态测量学分析,探讨利用几何形态测量学实现天蛾科成虫数字化分类的可行性,为逐步实现蛾类昆虫的数字化分类提供新的形态学依据。[方法]首先,进行几何形态测量学分析:以天蛾右前翅为研究对象,按特定次序选取17个翅脉交点作为标记点,并获得坐标数据;对原始标记点坐标数据进行普氏叠加分析消除标本摆放位置、方向和大小等非形状因素等信息;对普氏叠加后的标记点数据进行相对扭曲分析,得到17个标记点对10种天蛾分类作用的大小。其次,对几何形态测量学分析所得数据进行多元统计分析:利用单因素方差分析对17个标记点的差异显著性进行检验,再对普氏叠加后的标记点数据进行主成分分析,然后利用前3个主成分进行判别分析。[结果]相对扭曲分析表明标记点2、4和5对于10种天蛾成虫的分类作用相对较大;单因素方差分析显示17个标记点均具有显著的差异,即标记点在种间是有显著差异的,可以用于本文10种天蛾成虫的分类鉴定;主成分分析的前3个主成分的累计贡献率为97.7%,可作为对10种天蛾成虫进行分类鉴定的变量;判别分析结果显示回归判别和交叉判别的准确率均为100%,实现对天蛾科10种成虫的分类鉴定。[结论]研究表明几何形态测量学可应用于天蛾成虫的数字化分类鉴定,可为未来进一步实现蛾类成虫的自动识别奠定基础。

关键词: 天蛾, 前翅翅脉, 几何形态测量学, 分类鉴定, 森林昆虫

Abstract: [Objective] The present paper analyzed geometric morphology of 10 species of sphingid adults using geometric morphometric method to explore the feasibility of geometric morphometry on digital classification of sphingid moths for providing new morphological foundation on realizing digital classification of moths progressively.[Method] In the first step, geometric morphometric analysis was carried out. Right forewings of 10 sphingid species were used as research objects and 17 intersections of wing veins were selected as landmarks according to specific order, and coordinate data were obtained. Procrustes superimposition was applied to data of the 17 landmarks to remove non-shape variation from the landmark coordinates such as placing location, direction and sizes; and further, relative warp analysis was used to the new landmark coordinates to obtain effects of 17 landmarks to the 10 sphingid species. In the second step, multivariate statistical analysis was applied to the data collected by geometric morphometric analysis. One-way variance analysis was used for significant test of difference of landmark coordinates firstly; then, principal component analysis of the Procrustes transformed data-set was implemented; at last, discriminant analysis was carried out on the first three components.[Result] The result showed that Landmarks 2, 4 and 5 had relative large contributions on the classification of the 10 sphingid adult species according to relative warp analysis. All of the 17 landmarks had significant difference and could be used for the classification of those 10 Sphingidae adult species by using one-way variance analysis. The accumulated rate of contribution of three principal components of the principal component analysis reached 97.7%, which could be used as variables of the classification and identification of the 10 sphingid adult species. The accuracy of original and cross validation tests reached to 100% and 99.7%, respectively. Classification and identification of the 10 sphingid moths were realized.[Conclusion] It was indicated that geometric morphometric analysis could be used to identify the sphingid adults accurately and would be useful in gradually realizing the automatic recognition of moths in the future.

Key words: Sphingidae species, forewing veins, geometric morphometry, identification, forest insect

中图分类号: