兽类学报

• 论文 • 上一篇    下一篇

对应分析和去趋势分析在有蹄类取食模式研究中的应用

张玮琪 张宁 何欢 马建章 钟林强 孙悦 张明海   

  1. (1东北林业大学,野生动物资源学院,哈尔滨 150040)
    (2 国家林业和草业局林产工业规划设计院,北京 100010)
  • 出版日期:2019-03-30 发布日期:2019-03-26
  • 通讯作者: 张明海 E-mail: zhangminghai2004@126.com
  • 基金资助:
    国家自然科学基金项目(31500328);中国博士后科学基金项目(2016M601400);黑龙江省科学基金项目(QC2017011)

Application of correspondence analysis and detrended correspondence analysis in foraging pattern of ungulates

ZHANG Weiqi, ZHANG Ning, HE Huan, MA Jianzhang, ZHONG Linqiang, SUN Yue, ZHANG Minghai   

  1. (1 College of Wildlife Resources, Northeast Forestry University, Harbin 150040, China)
    (2 Forestry Planning and Design Institute of Forest Products Industry, National Forestry and Grassland Administration, Beijing 100010, China)
  • Online:2019-03-30 Published:2019-03-26

摘要:

本研究将2009年1月和2010年1月小兴安岭大沾河湿地自然保护区二可河林场内驼鹿冬季食性作为原始数据,分别以对应分析(CA)、去趋势分析(DCA),并将数据以样本为单位进行标准化后,再进行去趋势分析(DCA_std)3种排序方法,对驼鹿冬季取食模式进行了研究,后通过普鲁克分析,比较了不同排序方法对大型有蹄类取食模式研究的效果。结果表明,3种排序法的1轴和2轴均能涵盖绝大多数信息量,CA涵盖79.27%,DCA涵盖66.65%,DCA_std涵盖68.22%;3种方法均能够在1轴上区分针叶树和落叶乔木类食物,在2轴上,3种方法主要能够达到针叶树种与除落叶乔木外的其他植物类别的区分。虽三者均能够展现有蹄类取食模式,但在图形可视化后,仅DCA_std无明显的弓形效应。普鲁克分析结果表明,DCA_std样本位移平方和与CA和DCA均有很大差异,即将数据先进行标准化再进行DCA分析能够有效去除弓形效应。因此,在由多度组成的食性数据在进行标准DCA分析前,应对数据进行前期处理会得到更好的效果。同时,以样本为单位的标准化将使排序分析结果生态学意义更明确。

关键词: 排序法, 去趋势分析, 对应分析, 大型食草动物, 取食模式

Abstract:

Winter foraging patterns of moose were analyzed with 3 ordination methods, including correspondence analysis (CA), detrended correspondence analysis (DCA) and DCA with standardization in sample (DCA_std), using diet composition data of moose, in January of 2009 and January of 2010, at Erkehe forest farm of Dazhanhe wetland nature reserve in the Lesser Khingan Mountains. Procrustes analysis was used to compare effects of the 3 ordination methods. The results showed that axes 1 and 2 of all the three ordination methods could explain most variance of all variables with CA, DCA  and DCA_std  indicating 79.27%, 66.65% and 68.22% respectively. On axes 1, the methods could distinguish coniferous and deciduous trees while on axes 2, coniferous and other plant taxa except deciduous trees could be distinguished. Though all the three methods could reveal feeding pattern of ungulates, only DCA_std lacked apparent arc effects after graph visualizations. Procrustes analysis demonstrated that sum of displacement squares is the largest between DCA_std and CA, and DCA_std has the most unapparent arch effect, which in this case means performing data standardization before typical DCA could effectively eliminate the “arch effect”. Therefore, for diet composition data composed of food abundance, data standardization before typical DCA is necessary and might eliminate arch effect better. Meanwhile, standardization in samples would make ordination results more ecologically explicit.

Key words: Ordination, Detrended correspondence analysis, Correspondence analysis, Large herbivores, Feeding pattern