生物多样性 ›› 2010, Vol. 18 ›› Issue (5): 461-472.DOI: 10.3724/SP.J.1003.2010.461

所属专题: 生物多样性信息学专题(I)

• 生物多样性信息学专题 • 上一篇    下一篇

PSDS 2.0: 一个基于GIS和多个模型的生物潜在分布地预测系统

林聪田1,2, 纪力强1,*()   

  1. 1 中国科学院动物研究所, 北京 100101
    2 中国科学院研究生院, 北京 100049
  • 收稿日期:2010-01-19 接受日期:2010-06-11 出版日期:2010-09-20 发布日期:2010-09-20
  • 通讯作者: 纪力强
  • 作者简介:* E-mail: ji@ioz.ac.cn
  • 基金资助:
    国家科技基础条件平台工作重点项目子项目资助(2005DKA21402)

PSDS (predictive species distribution system) 2.0: a system based on GIS and multiple models for predicting potential distribution of species

Congtian Lin1,2, Liqiang Ji1,*()   

  1. 1 Institute of Zoology, Chinese Academy of Sciences, Beijing 100101
    2 Graduate University of the Chinese Academy of Sciences, Beijing 100049
  • Received:2010-01-19 Accepted:2010-06-11 Online:2010-09-20 Published:2010-09-20
  • Contact: Liqiang Ji

摘要:

根据对生物分布地预测模型和软件发展现状的分析和总结, 本研究在PSDS 1.0的基础上提出并实现一个基于GIS且具有多个代表性模型的生物分布地预测系统(PSDS 2.0)。PSDS 2.0系统继承了1.0的环境包络和聚类包络模型, 进一步引入了限制因子包络、马氏距离、支持向量机等新模型, 并针对本领域中模型比较与选择的难点增加了迭代交叉验证的多模型选择功能。系统还实现了灵活定制和评估伪负样本的功能, 通过用只需要正样本的I类模型预测的结果对随机产生的伪负样本进行评估, 减小其落入适宜地区的概率, 进一步提高需要正负样本的II类模型的准确率。GIS功能在PSDS 2.0中也得到加强, 被应用于数据准备及结果分析等重要环节。文章最后以白冠长尾雉(Syrmaticus reevesii)为例, 运用PSDS 2.0系统预测其在中国范围内的潜在分布地, 并对各种模型的预测结果进行评估和比较。

关键词: 生态位模型, 限制因子, 生物潜在分布地, 伪负样本, 交叉验证

Abstract

Herein, we have proposed and implemented a predictive species distribution system (PSDS) based on GIS (Geographic Information Systems) and using multiple models, according to the situation in the field of species habitat modeling. The new system (PSDS 2.0) was developed from PSDS1.0, originally initiated by our research group. We introduced three models into PSDS 2.0, including a mahalanobis distance model (MD), an environmental envelope plus limiting factor model (EELF), and a support vector machine (SVM) model. In this paper, we describe the implementation of the system and introduce the main functions in detail. In order to compare and evaluate results from different algorithms, an iterative cross-validation technique has been implemented in PSDS 2.0, which also facilitates the selection of suitable algorithms for different sample data. A function for flexibly dealing with pseudo-absences has been incorporated into presence-absence models. A GIS interfaces with the software for data preparation and further analysis of the model results. We also present a case study using the Reeve’s pheasant, Syrmaticus reevesii, as a practical application to introduce the entire modeling process. The performance of all model types is compared within this unified system.

Key words: ecological niche model, limiting factors, species’ potential distribution, pseudo-absence, cross validation