Habitat connectivity and dispersal ability drive the assembly mechanisms of macroinvertebrate communities in river networks

Metacommunity studies commonly use spatial distances or, more recently, landscape resistance measures between study sites as a surrogate for connectivity. However, local communities are connected to many other sources of colonisation than the sites included in a study, and the availability and distance to potential colonisation sources may be better surrogates for dispersal than distances to other sampling sites. Here we test the effect of habitat connectivity on the assembly of stream‐riffle macroinvertebrates with different dispersal abilities, after controlling for habitat heterogeneity and among‐site distances (62 km on average). We used a null model approach to compare observed community dissimilarity to random expectation. Significant deviations from randomness were expected due to the hierarchical structure of river networks with their increasing flux of organisms from headwaters to mainstem reaches. We found a gradual shift in dispersal‐based processes driving assembly mechanisms, from dispersal limitation in the isolated headwater streams to randomness in connected headwater and isolated mid‐order streams, and to mass effects in the most connected mid‐order streams. Weak flyers were constrained by dispersal limitation in the most isolated sites, whereas strong flyers were not restricted by the river network structure and were mainly assembled through mass effects. The approach taken was sufficient to unravel the importance of dispersal and habitat connectivity on community assembly and may therefore be particularly well suited to other large data sets with isolated sites (i.e. low geographical density of sites).