Unfragmented, biophysically complex river networks are rare, but their study is vital to understanding the processes by which imperiled, diverse riverine organisms are sustained. In such a network in the wilderness of central Idaho, USA, we studied the increased biophysical heterogeneity that occurs when multiple, adjacent tributaries enter the mainstem in close proximity, and investigated the effect of such ‘tributary-confluence complexes’ on biodiversity and metacommunity dynamics, as well as food-web characteristics that may be linked to maintenance of this diversity. Throughout the 6th-order river network (including four confluence complexes and three mainstem sites not associated with confluences), we measured habitat, surveyed benthic macroinvertebrates, fishes, and amphibians, sampled gut contents of the latter two groups, and collected organism tissues for stable isotope analyses. In addition, to assess macroinvertebrate metacommunity dynamics we combined this spatially extensive sampling with a decadal-scale analysis of metacommunity dynamics in one confluence complex that has been annually monitored. From comparisons and aggregation analyses, we found increased habitat heterogeneity associated with tributary confluence complexes, and that this network complexity supported increasingly diverse macroinvertebrate metacommunities. Additionally, we observed emergent metacommunity phenomena at confluence complexes, and that high inter-annual variation among habitats within a metacommunity contributed disproportionately to the maintenance of diversity at the decadal scale. We also found greater fish and amphibian diversity, and meta-food web complexity associated with tributary confluence complexes. Furthermore, we observed as much variation in diets of ecospecies among habitats in tributary confluence complexes as among all the assemblage within any single habitat, pointing to high feeding plasticity in these habitat mosaics. Metacommunity dynamics, as well as trophic variability and meta-food web complexity, may result from heterogeneous mosaics shaped by network topology and heterogeneity, and such complexity may be important to biodiversity maintenance in river networks.