祁连山青海云杉林物种分布模型与变量相异指数

Species potential distribution models and evaluation based on dissimilarity index of variables of Qinghai spruce (Picea crassifolia) in Qilian mountains

选择了边界函数法、最大熵算法以及基于规则集的遗传算法,选取坡向、海拔、多年平均温度和多年平均降水量4个环境变量,对祁连山国家自然保护区的青海云杉进行分布预测.提出一种新的物种分布模型性能评价方法,通过提取分布区上各变量的特征,比较现实分布区和潜在分布区上环境变量的差别,来评价模型在性能上的表现.该方法以计算环境变量的边界相异指数和总体相异指数实现.应用两种指数对各模型的性能评价表明:GARP模型获得4组MDIV的最小值,BF和Maxen分别获得3,2组最小MDIV,表明GARP模型性能的优越性;GARP,Maxent和BF模型的DIV分别为10.9965,15.1455和18.8747,同样表现出GARP模型优于其他两种模型的物种分布预测性能.青海云杉主要分布在研究区中部,包括肃南、民乐、山丹县存在大面积的适宜分布区.研究区南部天祝县的潜在分布区也大于现实分布范围.

Three species potential distribution models（border function（BF）,maximum entropy（Maxent） and genetic algorithm for rule set production（GARP）） and four environmental variables（aspect,elevation,annual mean temperature and precipitation） were selected to predict the potential distribution of Qinghai spruce（Picea crassifolia） in Qilian mountains national reserve.A new approach was introduced which could extract environmental variables above the actual and potential distribution areas;then differences between two sets of environment variables were compared and the differences were demonstrated by computation of marginal dissimilarity index of variables（MDIV） and dissimilarity index of variables（DIV）.Our results indicate that GARP obtained four minimum MDIVs in all the nine variables,Maxent and BF obtained three and two minimum MDIVs,showing the best performance of GARP among the three models.Total dissimilarity indexes of variables（TDIVs） for Maxent,GARP and BF were 15.145 5,10.996 5 and 18.874 7 respectively,also indicating that GARP is better than BF and Maxent in prediction of distribution of species.The middle part of the study area,administerially including Sunan,Minle and Qilian counties,are the most suitable growth area for Qinghai spruce（Picea crassifolia）.The potential distribution is larger than the real distribution at southern part of thestudy area（Tianzhu county）,which indicates that there has been an improvement of potential ecology at this part of the study area under current environmental conditions.