مدیریت بیابان (May 2022)

Prediction of Potential Distribution of Prosopis Farcta L. in Marginal Rangelands of Niatak River of Sistan

  • Vahid Azimi,
  • Hossein Piri Sahragard,
  • Peyman Karami,
  • Morteza Saberi

DOI
https://doi.org/10.22034/jdmal.2022.548628.1374
Journal volume & issue
Vol. 10, no. 1
pp. 53 – 66

Abstract

Read online

The present study aimed at preparing the potential distribution map and identifiying the environmental requirement of Prosopis farcta L. using tree-based and regression methods in the marginal rangeland of Niatak river in Sistan region. For this purpose, species presence data was recorded randomly. Environmental variables were prepared using field sampling and digital elevation model. In order to achieve the pseudo-absence points, first habitat modeling was performed using the domain model, then pseudo-absence points were prepared using the prediction map obtained from this method. Species distribution modelling was conducted using random forest (RF), classification and regression trees (CART) and generalized additive model (GAM). The accuracy of the models used was evaluated using the area under curve criterion. Result showed that the RF with area under curve 0.98 has the highest accuracy. Generalized additive models and classification and regression trees were ranked after RF. The highest and lowest values of kappa index were assigned to the RF with 0.75, and GAM with 0.43 Kappa value. Accordingly, the RF model is the most accurate model in predicting the potential habitat distribution. Analysis of the variable’s importance showed that in the studied scale, edafic factors and distance from the river have greater effect on species distribution than other factors. So that, in all models used, acidity and electrical conductivity were identified as the most important variables. In general, it is suggested that habitat development plans for Prosopis farcta should be planned in the central and marginal parts of the Niatek river due to better suitability of these regions for species distribution.

Keywords