Modelling forest bird community richness using CORINE land cover data: a study at the landscape scale in Hungary

In this study we (1) examined the applicability of the widely available CORINE land cover map of Europe in predicting several components of the richness of forest breeding bird community, and (2) analysed how different ecologically meaningful species groups respond to the differences in landscape composition and how these differences are reflected in the relationships between total species richness and richness of these species groups at the 2.5 x 2.5 km(2) scale. We found that landscape composition had only moderate success in predicting the richness components of the forest bird community. The predictive power of the applied general linear models differed very much: roughly 60% of the observed variance was explained when the dependent variables (species richness and abundance) were calculated using data of all the 21 studied forest bird species or that of common forest birds. However, species richness and abundance of groups of more vulnerable species were predicted with much less success (30% variance explained), suggesting that CORINE is not an adequate tool in predicting the conservation status of these sensitive forest birds. Forest cover explicated 90 to 100% of the explained variance in the models suggesting that forest bird community was much less sensitive to the type of land cover occuring in the surroundings. We showed that richness and abundance of selected species groups had different non-linear relationships with total species richness, suggesting that - if used alone -, total species richness is a weak predictor of other richness components of the forest bird community.