Modelling the spatial distribution of the White Stork (Ciconia ciconia) breeding populations in the Southeast Europe (CROSBI ID 189566)
Prilog u časopisu | izvorni znanstveni rad | međunarodna recenzija
Podaci o odgovornosti
Radović, Andreja ; Kati, Vassiliki ; Perčec Tadić, Melita ; Denac, Damijan ; Kotršan Dražen
engleski
Modelling the spatial distribution of the White Stork (Ciconia ciconia) breeding populations in the Southeast Europe
With this research we wanted to quantify the suitability of the territory of three western Balkans countries (Slovenia, Croatia and Bosnia and Herzegovina) for sustaining breeding population of white storks (Ciconia ciconia). Using spatial modelling on 1km resolution we defined environmental variables that have premier influence on spatial position of breeding pairs in the region. Variables that explained the greatest part in the density of breeding pairs according to the significance of the regression coefficients are the elevation, the total precipitation during spring period and the topographic wetness index of the terrain. Elevation enters the regression equation with the negative loading while other two variables enter the equation with positive loadings. Other important variables for the density of the breeding pairs are minimal temperature during early breeding period, slope, amount of settlements in grid cell and settlement distance. Variables that describe the presence of the rivers or lakes or their proximity (total amount of water in a cell or distance to the nearest water) or the amount of forests in the cell did not proved to be significant at 0.05 level. The multiple regression model explained 71.05% of total variance of the nest densities. Even though it seamed that the addition of pseudo- absence points did not improved the total variation explained with the multiple regression model due to regarded species is at the edge of its distribution, presence only locations are not suitable for the detecting limiting factors for the species in the region. Further improvement of the multiple regression model using presence and absence locations was done using kriging of the residuals. On the validation data the accuracy improved from 68.88% for the multiple regression model to 82.87% for the regression kriging model. Another indicator of the sustainability of the Balkans for the breeding of the white stork population, namely presence of the white stork’s nest in 1 km grid cells, was predicted with the binomial GLM model almost perfectly.
regression-kriging; habitat suitability; Western Balkans; monitoring; climate change
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano