Quantifying the relative importance of how local (environmental or niche‐based) and regional (dispersal‐related or spatial) processes regulate the assembly of communities has become one of the main research avenues of community ecology. It has been shown that the degree of isolation of local habitats in the landscape may substantially influence the relative role of environmental filtering and dispersal‐related processes in metacommunities. Dendritic stream networks are unique habitats in the landscape, where more isolated upstream sites have been predicted to be primarily structured by environmental variables, while more central mainstem rivers by both environmental and spatial variables (hereafter the network position hypothesis, NPH). However, the NPH has almost exclusively been tested for stream macroinvertebrates, and therefore its predictions warrant confirmation from multiple taxa. We examined the validity of the NPH for benthic diatoms, macrophytes, macroinvertebrates and fish in the Pannon Ecoregion, Hungary. Following the NPH we predicted a clear dominance of environmental over spatial variables in headwaters, and a larger effect of spatial variables in rivers compared to headwaters. We tested these predictions using variance partitioning analyses separately for the different taxa in headwater and in riverine habitats. We found large differences in the explained community variance when the impact of environmental (physical and chemical) and spatial (overland and watercourse distance) variables for various taxa was studied. In general, total explained variance was lower for the more passively dispersing plant taxa than for animal taxa with more active dispersal in both streams and rivers. However, similar to other studies, the total explained variance was low for both headwater streams and rivers. Community structure of diatoms could be best explained by both environmental and spatial variables in streams, whereas their community structure could not be explained by either variable group in rivers. The significance of environmental and spatial variables depended on the distance measure (overland versus watercourse) in the case of macrophytes. Community structure of macroinvertebrates could be explained by environmental variables in streams and by both environmental and spatial variables in rivers. Moreover, variation was explained by different predictors when macroinvertebrate taxa were divided into flying and non‐flying groups, suggesting the importance of dispersal mode in explaining community variation. Finally, community structure of fishes could be explained by both environmental and spatial variables in streams and only by environmental variables in rivers. In conclusion, we found no clear evidence of the NPH in our multi‐taxa comparison. For example, while patterns in macroinvertebrate communities seem to support the NPH, those in fish communities run counter with the predictions of the NPH. This study thus shows that different taxa may behave differently to isolation effects in stream ne...

Summary 1. Quantifying the relative importance of environmental filtering versus regional spatial structuring has become an intensively studied area in the context of metacommunity ecology. However, most studies have evaluated the role of environmental and spatial processes using taxonomic data sets of single snapshot surveys. 2. Here, we examined temporal changes in patterns and possible processes behind the functional metacommunity organization of stream fishes in a human‐modified landscape. Specifically, we (i) studied general changes in the functional composition of fish assemblages among 40 wadeable stream sites during a 3‐year study period in the catchment area of Lake Balaton, Hungary, (ii) quantified the relative importance of spatial and environmental factors as determinants of metacommunity structure and (iii) examined temporal variability in the relative role of spatial and environmental processes for this metacommunity. 3. Partial triadic analysis showed that assemblages could be effectively ordered along a functional gradient from invertebrate consuming species dominated by the opportunistic life‐history strategy, to assemblages with a diverse array of functional attributes. The analysis also revealed that functional fish assemblage structure was moderately stable among the sites between the sampling periods. 4. Despite moderate stability, variance partitioning using redundancy analyses (RDA) showed considerable temporal variability in the contribution of environmental and spatial factors to this pattern. The analyses also showed that environmental variables were, in general, more important than spatial ones in determining metacommunity structure. Of these, natural environmental variables (e.g. altitude, velocity) proved to be more influential than human‐related effects (e.g. pond area, % inhabited area above the site, nutrient enrichment), even in this landscape with relatively low variation in altitude and stream size. 5. Pond area was, however, the most important human stressor variable that was positively associated with the abundance of non‐native species with diverse functional attributes. The temporal variability in the relative importance of environmental and spatial factors was probably shaped by the release of non‐native fish from fish ponds to the stream system during flood events. 6. To conclude, both spatial processes and environmental control shape the functional metacommunity organization of stream fish assemblages in human‐modified landscapes, but their importance can vary in time. We argue, therefore, that metacommunity studies should better consider temporal variability in the ecological mechanisms (e.g. dispersal limitation, species sorting) that determine the dynamics of landscape‐level community organization.