ABSTRACT
This paper reports on the results of analyses of 28 forest vegetation plots (50×50 meters) situated along a transverse transect in Medvednica Nature Park, Croatia. The purpose of the research was to establish a correlation between the plant diversity and environmental factors with an emphasis on concentration of chemical elements in the topsoil. Models were developed to demonstrate a possible approach for defining priority areas for the protection of plant diversity with the use of GIS. Multiple linear regression was used for the development of a predictive model of the number of plant species, with soil acidity and concentrations of cobalt, chromium, magnesium and yttrium as independent variables. For rare species, logistic regression and equal weight predictive models were developed using the altitude, soil acidity and concentrations of nickel, cobalt and potassium extracted as independent variables by canonical correspondence analysis (CCA). Developed models were incorporated in GIS, and areas with highest plant diversity and highest number of rare plant species were defined. Based on the results of the cluster analysis of the species composition, a dataset of vegetation type was derived and used for selection of an additional area in order to protect the complete plant diversity of the surveyed area. Appropriateness of selected statistical methods were discussed, as well as advantages of GIS-based approach in analyses of plant diversity and distribution, and selection of areas important for preservation of plant diversity.
