:: Volume 15, Issue 2 (8-2021) ::
مرتع 2021, 15(2): 195-213 Back to browse issues page
Habitat potential modeling of Thymus kotschyanus Boiss. & Hohen. in the northern of Ardabil Province rangelands
Sahar Ghafari, Ardavan Ghorbani , Mehdi Moameri, Raoof Mostafazadeh, Mahmood Bidar Lord, Azad Kake Mami
Department of Range and Watershed Management, Faculty of Agricultre and natural resource, University of Mohaghegh Ardabibli, Ardabil
Abstract:   (391 Views)
This study aims at comparing the performance of MaxEnt and logistic regression in preparing the predictive habitat distribution map of Thymus kotschyanus and determining the factors affecting in the northern of Ardabil Province rangelands. 28 sites were selected and at each site, three transects with a length of 100 m and on each transect ten 1m2 plots were established. Soil samples were taken digging of nine soil profiles and samples taked from 0-15 and 15-30 cm depths. Geostatistical techniques were used to provide the soil maps. Digital elevation model was used to prepare the topography maps as a data layer. The Landscape metrics using Fragstats was calculated. Model accuracy in MaxEnt method was evaluated by using the AUC. By jackknife method and response curve, the most important environmental predictor variables were found. Model accuracy in logistic regression method is evaluated by using Hosmer and Lemshow Statistic and ROC. Then predictive distribution maps were produced. The Kappa coefficient index was used to evaluate the accuracy of the distribution maps. Based on logistic regression model, potassium of the 2th depth, aggregation index (positive correlation), slope and bare soil were the negative correlations and based on MaxEnt model, clay of the first depth, patch area (coefficient of variation), slope are the most influential factors affecting the presence of this species in this habitat. T.kotschyanus is widespread on light-textured soils (clay<15%), potassium (<18 meq/L), Silt (<20%) and high elevation (300-2700 meters above sea level), and slope (35-55%). Based on kappa coefficient, logistic regression model was able to predict the habitat distribution of studied species at the good level (kappa= 0.64) and MaxEnt had in intermediate level (kappa 0.42). These results indicate that the logistic regression model is more accurate.
Keywords: Modeling, Habitat Desirability, Environmental factors, Logistic regression, Maximum entropy (MaxEnt)
Full-Text [PDF 861 kb]   (143 Downloads)    
Type of Study: Research | Subject: Special
Received: 2021/08/6 | Accepted: 2021/08/1 | Published: 2021/08/1

XML   Persian Abstract   Print

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Volume 15, Issue 2 (8-2021) Back to browse issues page