多尺度模型在动物生境选择中的应用——以驼鹿为例  

Multi-scale Model in Animal Habitat Selection: A Case Study of Moose

作　　者：于沿泽[1] 李倩[1] 张明海[1] 李宝国 Yu Yanze;Li Qian;Zhang Minghai;Li Baoguo(Northeast Forestry University,Harbin 150040,P.R.China;Inner Mongolia Great Khingan Forestry Investigation and Design Institute)

机构地区：[1]东北林业大学,哈尔滨150060 [2]内蒙古大兴安岭森林调查规划院

出　　处：《东北林业大学学报》2019年第8期58-61,共4页Journal of Northeast Forestry University

基　　金：黑龙江省青年基金项目(QC2014C027)

摘　　要：以内蒙古大兴安岭根河地区冬季驼鹿为研究对象,采用多尺度模型框架研究驼鹿的生境选择,并比较了单尺度模型与多尺度模型在偏差解释率上的表现。结果表明:①单变量逻辑斯蒂回归分析显示,驼鹿对环境因子反应尺度在不同因子间差别很大,海拔、坡向、灌丛比例、白桦林比例、混交林比例与驼鹿的出现在相对小尺度上具有强相关性(≤1000m),而其他变量如坡度、归一化植被指数在相对较大尺度上(≥2000m)对驼鹿生境选择产生强影响。②经多模型平均后,对冬季驼鹿生境选择具有重要影响的环境因子为坡度、道路密度、白桦林比例与混交林比例。③模型平均偏差解释率比较发现,多尺度模型的偏差解释率为0.72,高于任何一个单尺度模型,说明多尺度模型更适于研究动物生境选择。With moose(Alces alces cameloides)of winter in Genhe Forestry Bureau,Inner Mongolia as target species,we used multi-scale model framework to study moose habitat selection,and compared model performance of single and multi-scale model on proportion model-averaged deviance explained.Results shows:(1)analysis of univariate logistic regression recognize moose response to different habitat factor at very different spatial scale,elevation,aspect,bush ratio,brich ratio and mixed forest ratio were all most strongly related to moose winter home range selection at relatively fine scale(≤1 000 m),whereas,the remaining factors such as slope,normalized different vegetation index were strongly related at much coarser scales(≥2 000 m).(2)After multi-model averaging,four habitat factors have strong influence on habitat selection of moose in winter which are slope,road density,birch ratio and mixed forest ratio.(3)Model-averaged proportion deviance explained of multi-scale model reaches 0.72,and it is greater than all single scale model,demonstrating that multi-scale model is more suitable for animal habitat selection study.

关 键 词：驼鹿 生境选择 多尺度模型 

分 类 号：S718.6[农业科学—林学] S864.1