Aim The development of metacommunity theory inspired a series of studies exploring the importance of environmental and spatial effects on the composition of biotic assemblages. However, the comparison of different groups of organisms has been hampered by differences in sampling design, spatial scales or the environmental variables involved. Our aim was to test how dispersal ability affects metacommunity structure and associated species distributions by sampling different species groups in the same plots to avoid these problems. Location Western Carpathian Mountains (Europe). Methods In 191 fens we sampled the composition of diatom, bryophyte, vascular plant and mollusc assemblages, water chemistry, and macroclimatic data. We then generated spatial variables covering all relevant spatial scales using analysis of principal coordinates of neighbour matrices (PCNM). We applied the adjusted variation partitioning algorithm to quantify the effects of environment and space. Results Pure effects of water chemistry and space were highly significant for all groups of organisms. Spatial effects were stronger for groups with larger propagules (vascular plants, molluscs) than for those with smaller propagules (diatoms, bryophytes). Assemblages of macroscopic bryophytes were structured slightly less by geography and much more by environment than were those of microscopic diatoms. Vascular plant and mollusc assemblages turned out to be more spatially structured (as compared to diatom and bryophyte assemblages), with small differences between the two groups. Coarse-scale spatial effects dominated in the bryophyte metacommunity, while in the other groups, including diatoms, finer-scale effects were also important. Main conclusions Given that our analyses are based on a standardized sampling and analytical framework, our findings provide strong support for the hypothesis that both environmental and spatial variables structure metacommunities of organisms with very different dispersal abilities, including microscopic diatoms. In addition, we show for the first time that the strengths of these effects and their scale dependence may be predicted using important trait differences between organisms, for example differences in propagule size.