Aims The aim of the study was to discover what set of variables best explains the transition from warm to mesic forest vegetation. Based on various variables grouped into sets (geomorphological, ecological, structural, soil characteristics and chorological), six models were built and tested by generalized additive mixed models (GAMMs). We assumed that each set of variables has different explanatory power. Our aim was to compare the six different models (sets of variables), to test which model best explains the species turnover in forest communities along the transition between warm and mesic temperate forests and to try to find reasons for the different explanatory power of the models.
Methods The research took place in the southern part of the Balkan Peninsula. Field sampling was done according to standard methods. The gradient from warm to mesic forests was defined as the turnover of species and evaluated by projection of samples on the first unconstrained DCA axis. Geomorphological, ecological, structural and soil characteristics, together with chorological sets of variables, were regressed on the turnover of species composition. Based on the five sets of variables, six models were constructed and tested by generalized additive mixed models.
Important findings Ecological conditions best explain the change of forest communities along the gradient; evolution and the development of vegetation reflected in chorotypes are also of high importance; geomorphology and structure seem not to change so dramatically and soil shows the least significant differences of all. Ecological variables are the most important set of variables in the transition between warm and mesic temperate forests but eco-evolutionary dynamics after the Pleistocene should also be taken into consideration.