R. 2002. Forest habitat characteristics affect balance between sexual reproduction and clonal propagation of the ectomycorrhizal mushroom Hebeloma cylindrosporum.-Oikos 99: 25-36. We investigated how habitat characteristics and in particular the level of disturbance was affecting colonisation processes and population dynamics of an ectomycorrhizal fungus. The basidiomycete Hebeloma cylindrosporum associated with Pinus pinaster trees occurs in two distinct habitats within the forest dune ecosystem along the Atlantic coast of France. The 'dune habitat' (D habitat) is characterised by scattered trees growing in almost pure sand at the extreme fringe of the forest marking the transition with the bare dune. The 'forest habitat' (F habitat) is within the mature forest in places of high human frequentation which prevents litter and humus accumulation. We studied over several years (up to ten) populations of H. cylin-drosporum present in three geographic sites. Within each site we studied two areas corresponding each to one of the two possible habitats. In all sampled areas fruit bodies occurred within discrete patches of ground. Although an interaction between habitat and geographic site existed, habitat had significant effects on patch sizes and patch internal densities. Compared to the F habitat, patches in the D habitat were significantly larger and presented a lower density of fruit bodies. Patches very seldom occurred at the same place from one year to the next in the F habitat when in most cases they were perennial and increased in size in the D habitat. Fruit body size, which reflects reproductive output of the genets, differed between the two habitats (means of the cap diameters of 25.6 and 40.4 mm in respectively the F and D habitats). Almost all patches in the F habitat were made of several annual genets each producing on average 1.6 fruit bodies while most patches in the D habitat were dominated by one perennial genet which could simultaneously produce up to 60 fruit bodies. Under-representation of small sized genets in the D habitat suggests a low level of sexual recruitment in this habitat despite the large number of fruit bodies (and therefore of spores) produced by each of the existing genets. We therefore suggest that H. cylindrosporum could be considered as a fugitive species. Genets of this species survive in the D habitat in which few ectomycorrhizal fungal competitors occur. This habitat could represent the optimal habitat for this species. In the F habitat frequent recruitment could occur in 'empty patches' resulting from the local and transient elimination of competitors induced by human disturbance.