A comparison of logistic regression and maximum entropy for distribution modeling of range plant species (A case study in rangelands of Western Taftan, Southeastern Iran)

Abstract

This study aimed to compare the efficiency of logistic regression and maximum entropy models for distribution modeling of plant species habitats in the rangelands of western Taftan, southeastern Iran. Vegetation cover was sampled using a systematicrandomized method. Soils were sampled at 0-30 and 30-60 cm depths through digging of eight soil profiles. The agreement between predictive maps generated by models with documented maps of habitats indicated that logistic regression was able to predict the distribution of Artemisia aucheri and Artemisia sieberi habitats at excellent (kappa value = 0.95) and weak (kappa value = 0.39) levels, respectively. On the other hand, the agreement between predicted maps generated by maximum entropy with documented maps was very good for Amygdalus scoparia and Artemisia aucheri habitats (kappa value = 0.82 and 0.76, respectively), and weak for Artemisi aucheri (kappa value = 0.55). This study indicates that logistic regression and maximum entropy methods had the same efficiency in distribution modelling of plant species with a limited ecological niche. However, the maximum entropy model can receive priority in distribution prediction of plant species with a limited ecological niche because it uses only presence data of plants and a small dataset.