6 7

23 24 Beside biotic interactions, habitat preference and dispersal ability of species play a prominent 25 role in the building of animal species assemblages. However, these traits are usually very poorly 26 documented for soil organisms. A soil transfer experiment was designed to study habitat preference 27 (including land-use and soil preference) and dispersal ability of soil springtail species living in a 28 meadow and in an adjoining deciduous forest. The study was performed in the Morvan Regional 29 Natural Park (Central France), using untreated or defaunated soil blocks, transferred to another land30 use or replaced in their original land-use. Land-use preference was quantified in untreated and 31 untransferred samples from meadow and forest. Dispersal ability was estimated from the time at which 32 species colonized defaunated samples in their own habitat. Soil preference was estimated from the 33 colonization rate of defaunated samples by comparing transferred and untransferred soil blocks. 34 Results showed that in the community, 6% of species were land-use generalists, 30% were soil 35 generalists and 36 % recolonized defaunated soil blocks within a week. Land-use preference, soil 36 preference and dispersal ability were largely independent components of species characteristics. 37 Although our experiment dealt only with small-scale colonization, comparisons between species 38 showed that the dispersal type based on anatomical features (legs, antenna, furcula, visual apparatus) 39 does not allow predicting the dispersal ability of these species. Discrepancies between land-use 40 preference and soil preference suggest that other habitat features must be relevant for Collembola, and 41 that a trade-off exists between eco-physiological and biotic interactions (including food requirements). 42