LAUSR

Summary The metacommunity framework has shed light upon the significance of local and regional processes in shaping ecological communities. However, our knowledge is limited how landscape context (i.e. type and positioning of habitats) influences the structuring of metacommunities. We examined the role of environmental selection (niche-based species sorting), dispersal and drift (i.e. stochastic changes in species abundance) mechanisms on the structuring of fish metacommunities in riverscapes. We used a hierarchical design and examined metacommunity structuring (i) in individual running water habitat types, namely highland streams, lowland streams, highland rivers and lowland rivers; (ii) in landscape types, where patterns in highland and in lowland types were examined separately and (iii) at the level of the whole riverscape, where samples from all running water types were pooled. Elements of metacommunity structure (EMS) analysis, community–environmental relationships in redundancy analysis, and decay of community similarity with distance (distance decay) complemented with partial Mantel tests were used to reveal the mechanisms behind the observed patterns. Both the EMS framework and community–environment relationships revealed the prevalent role of selection (niche-based species sorting mechanisms) on fish metacommunity structuring, although stochastic variability also influenced the results. Metacommunity structures reflecting niche-based mechanisms (Clementsian or quasi-Clementsian) were observed at higher hierarchical levels (i.e. at the riverscape and the landscape type levels), while quasi-Clementsian or random patterns were found within single running water types. Distance decay relationships indicated that dispersal limitation did not clearly affect metacommunity structuring in any combination of running water types. Partial Mantel tests showed that some part of the variation in distance decay could be explained by the correlation between environmental variables and fish assemblage structure in some habitat types, which further suggests the importance of niche-based mechanisms in metacommunity structuring. Our study shows that different metacommunity structures can arise in a hierarchy of habitat types in riverscapes, and niche-based species sorting mechanisms are more influential in their structuring than dispersal ones. Landscape classifications are useful for a better understanding of the structuring of metacommunities in both the freshwater and terrestrial realms.