Search for a command to run...
Summary Most metacommunity studies indicate that dispersal processes play a minor role compared with species sorting in explaining metacommunity organisation, in particular, in stream systems. However, the role of dispersal could vary with environmental harshness, as a result of frequent resetting of community succession by disturbances and the selection of generalist species from regional species pools. The importance of dispersal may also be mitigated by species dispersal ability. In this study, we explored how species sorting and dispersal shaped invertebrate and fish metacommunities across streams in three tropical headwater catchments in Bolivia with contrasting environmental harshness, including flow regime, altitude and climate conditions. We addressed the hypothesis that the relative roles of dispersal and species sorting vary with environmental harshness: we predicted that the role of species sorting would predominate in benign conditions, whereas that of dispersal would predominate under moderate environmental harshness, and that neither dispersal nor species sorting would be relevant to explain metacommunities under high environmental harshness. We also hypothesised that the role of dispersal would decrease with increasing species dispersal ability. Although there was little or no spatial autocorrelation of environmental distances (i.e. environmental differences) across the headwater catchments, community similarity correlated more strongly with environmental than spatial distances among headwater sites that had low environmental harshness, but the opposite pattern was observed among sites with moderate environmental harshness. Under high environmental harshness, neither environmental harshness nor spatial distances between sites explained community similarity. Under moderate environmental harshness, the correlation between community similarity and spatial distances was the strongest for moderate dispersers of both invertebrates and fish. Yet, in contrast to fish, strongly dispersing invertebrate taxa were spatial structured, suggesting that they were not able to reach all sites as predicted. Our results suggest the role of dispersal might be underestimated, notably in systems prone to environmental harshness. Better proxies for dispersal, along with the use of spatial distances to account for resistance to animal movements in river systems and that account for flow magnitude and directionality, slope, riparian vegetation, wind and streambed roughness, may promote a more realistic integration of dispersal processes in basic and applied metacommunity research.