Recent studies highlight linkages among the architecture of ecological networks, their persistence facing environmental disturbance, and the related patterns of biodiversity. A hitherto unresolved question is whether the structure of the landscape inhabited by organisms leaves an imprint on their ecological networks. We analyzed, based on pyrosequencing profiling of the biofilm communities in 114 streams, how features inherent to fluvial networks affect the co-occurrence networks that the microorganisms form in these biofilms. Our findings suggest that hydrology and metacommunity dynamics, both changing predictably across fluvial networks, affect the fragmentation of the microbial co-occurrence networks throughout the fluvial network. The loss of taxa from co-occurrence networks demonstrates that the removal of gatekeepers disproportionately contributed to network fragmentation, which has potential implications for the functions biofilms fulfill in stream ecosystems. Our findings are critical because of increased anthropogenic pressures deteriorating stream ecosystem integrity and biodiversity.stream networks | hydrological regime S treams and rivers sculpt the continental surface, forming fluvial networks (1), within which the biodiversity ranks among the highest on Earth (2). The dendritic nature of fluvial networks has been shown to affect the spatial and temporal patterns of microbial, invertebrate, and fish biodiversity (3-8). Ecological theory and observations posit that the local environment governs the dynamics and diversity of ecological communities in headwaters, the smallest and most abundant streams in fluvial networks. In contrast, dispersal ensures that communities further downstream are shaped not only by their immediate environment but also by upstream processes (3-9). Thus, the dynamics of the metacommunity, which comprises all interconnected communities in a landscape (10), are inextricably linked to the organization and hydrology of the fluvial network (5-8). This perception is essential to understand, predict, and manage streams and rivers and their resistance and resilience to human alterations across scales (that is, from patches to the catchment) (11).Ecological interactions are often usefully represented as networks (12). For example, analyses of food webs and mutualistic (e.g., pollination) networks have demonstrated that network organization can be linked to network persistence, to disturbance (12-16), or to species coexistence and richness (17). Microbial communities are so diverse and poorly studied that mapping out the interactions on the basis of biological knowledge is currently impossible for all but the simplest of habitats. Therefore, cooccurrence networks are increasingly used to infer microbial interactions (18,19)
in soils (20), oceans (21), lakes (22), and even in global genomic surveys (23).A key question is whether the organization of microbial cooccurrence networks and their response to disturbance reflect physical characteristics inherent in fluvial networks such as geo...