Using the Non-Parametric Classifier CART to Model Lebanon Cedar (Cedrus libani A. Rich) Distribution in a Mountain Mediterranean Forest District
Kürşad Özkan
More details
Hide details
Department of Soil and Forest Ecology, Faculty of Forestry, Suleyman Demirel University,
32260 Isparta, Turkey
Pol. J. Environ. Stud. 2013;22(2):495–501
This study was conducted to define indicator species of Lebanon cedar (Cedrus libani A. Rich) and to build its potential distribution model in the Yukarigokdere Forest District of Turkey’s Mediterranean region. In the study, the data were collected from 119 sample plots. To define indicator species of Lebanon cedar, interspecific correlation analysis (ICA) was employed. The findings obtained from ICA showed that thermo- Mediterranean elements (Pistacia terebinthus subsp. palaestina, Quercus coccifera, and Styrax officinalis) are the most significant negative indicator species of Lebanon cedar, while Its most significant positive indicator plants are supra- and mountain-Mediterranean elements (Acer hyrcanum subsp. sphaerocaryum, Berberis crataegiana, Amelanchier parviflora, Fraxinus ornus subsp. cilicica, and Sorbus umbellate). By using climatic and topographical data as explanatory variables, visual assessment of the potential distribution probability of Lebanon cedar based on classification and regression technique (CART) was performed. Ten-fold cross validation was run for selection of the optimal tree. The variables building the tree model were elevation, head index, annual precipitation, slope degree, landform category, and topographical position index. Among them, elevation was found to be the most significant factor on the distribution of Lebanon cedar.