engleski

Predictive distribution model of rocky pastures with respect to the grazing intensity on the Northern Adriatic islands

Rocky pastures are habitats that significantly increase the plant and landscape diversity. Nevertheless they are also endangered because of the frequent overloading of cattle and negligence due to the abandonment of traditional system of sheep farming. This research was conducted from the 2008 to 2010 on abandoned, moderately and intensively used rocky pastures on the northern Adriatic island of Pag, Krk and Cres. Based on the field collected data and LANDSAT ETM+ satellite image we have made a predictive map of rocky pastures with respect to the grazing intensity. Classification tree method (CT) have been used to develop predictive model than can estimate pasture capacity on the northern Adriatic islands. According to the model developed (with 83% accuracy on input data, Kappa statistics = 0.75), highest share of pastures was that of moderately used ones i.e. 40.36%. For latter it seems that on the northern Adriatic islands intensity of 1 to 1.5 sheep per hectare is the optimal intensity, that can improve not only the plant diversity, but “ecologically” based sheep manure as well.

grazing; remote sensing; GIS; northern Adriatic islands

nije evidentirano