7350

6856

كارشناسي ارشد

مرتعداري

اصفهان: دانشگاه صنعتي اصفهان، دانشكده منابع طبيعي

1391

پانزده، 119ص.: مصور، جدول، نمودار

ص.ع. به فارسي و انگليسي

مهدي بصيري، مصطفي تركش

محمدرضا وهابي

13/9/91

حميد متين خواه، حسين بشيري

1396/09/21

كتابنامه

منابع طبيعي

مهندسي منابع طبيعي

ID6856

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

Potential Habitat Modelling of Astragalus verus Olivier Using DOMAIN Logistic Regression Tree in Fereydunshahr Region Isfahan Province Marjan Saky Email Adrdress m saky@na iut ac ir 30 July 2012 Department of Natural Resources Isfahan University of Technology Isfahan 84156 83 111 IranDegree M Sc Language FarsiSupervisors Mahdi Bassiri Isfahani Bassirim@ cc iut ac ir Mostafa Tarkesh Isfahani tarkesh@cc iut ac ir Mohammad Reza Vahabi vahabi@cc iut ac ir AbstractThe relationship between plant species and environmental factors has always been a central issue inplant ecology Defination of these numerical relationships is main core of potential habitatmodelling of plant species These models are applied to achieve information about spatialdistribution of species habitat suitability and evaluation of climate change effects on the speciesspatial distribution Current study aimed to characterization of effective environmental factors onspatial distribution of Astragalus verus and produces potential habitat map using two statisticaltechniques logistic regression tree LRT and DOMAIN in a local scale A stratified randomsampling was applied to pick up 100 sites 50 presence 50 absence of given species Environmental maps were produced using geostatistics methods Kriging and Inverse DistanceWeight by Arc GIS software for the whole study area For each sampling site presence or absenceof given species as well as the environmental variables were recorded Two statistical techniques Logistic regression tree and DOMAIN were used to examine environmental variables related to thecurrent species distribution Maps of potential habitat of Astragalus verus were generated usingthese models Both statistical techniques produced strong and useful models LRT model indicatedthat species occurrence has high correlation with some environmental factors such as mean dailytemperature clay Electrical Conductivity and percentage of organic carbon of the soil Also theresults of PCA analysis to determination of the input factors to DOMAIN model show that 5environmental factors clay Electrical Conductivity percentage of organic carbon precipitation ofdriest season minimum temperature of coldest month are the most important factors related tospatial distribution of this species According to prediction map derived from the LRT and speciesresponse curves Astragalus verus is highly probable in terms of climate condition in regions withmean daily temperature is more than 14 31oC and in terms of edaphic conditions in soils withElectrical Conductivity is less than 0 3 ds m clay 42 to 46 percent organic carbon equivalent to 1to 2 percent Model evaluation was performed both by independent data set and model data Cohen s kappa value kappaLRT 0 9 kappa DOMAIN 0 95 and area under receiver operatingcharacteristics curve AUCLRT 0 96 AUCDOMAIN 0 97 based on independent data set indicated thehigh power of models to create potential habitat map of Astragalus verus in local scales Usingoptimal threshold from evaluation of models habitat suitability map were generated for the species Habitat suitability map derived from the DOMAIN and LRT demonstrate that 14118 hectare 14 of Study area and 29000 30 of study area are suitable for species occurrence According tokappa and AUC of ROC plot Domain model has higher power prediction The result of this studycan be applied in management and rehabilitation programs Keywords Prediction Models Potential habitat of Astragalus verus Logistic Regression TreeModel DOMAIN Model Geographic Information Systems

مهدي بصيري، مصطفي تركش

محمدرضا وهابي

حميد متين خواه، حسين بشيري