Are habitat models transferable in space and time?

Habitat models have proved to be efficient tools assessing the habitat suitability of species. Before applying such a model, its scope and transferability in time and space need to be tested. This paper shows ways of validating habitat models by transferring them in time and space and by testing the significance of the resulting confusion matrices, comparing data and prognoses. The importance of cutoff value choice for the classification results is discussed. Species of interest is the bush cricket Conocephalus dorsalis, which was surveyed in two fens in Northern Germany between 1995 and 1997.