Predictive Habitat Suitability Models for Teucrium Polium L. Using Boosted Regression Trees

Abstract

The aim of this study was to evaluate the association between the occurrence of Teucrium polium species and environmental variables to forecast habitat suitability. For species distribution modeling (SDM), three important steps have been taken: (I) identification of locations of Teucrium polium in the field, (II) preparing thematic predictors in GIS, and (III) SDM and its evaluation. A total of 113 positions of Teucrium polium have been reported for this project and their geographical locations have been registered using GPS. Fifteen variables were chosen and their layers were mapped using ArcGIS. Boosted regression trees (BRT) were modeled for habitat suitability. The findings showed that in the studied region, the highest effects on the distribution of T. polium were mainly correlated with altitude and silt percentage. Metrics such as Cohen’s kappa statistics (0.78), an area under the curve AUC (0.94), and true skill statistic TSS (0.8) indicated that the BRT model had high efficiency and accuracy for SDM. Increasing knowledge and encouraging the effective usage of medicinal plants, as well as their protection in natural areas, would be critical for their sustainability.