Background/Question/Methods Metacommunity theory incorporates space and dispersal into community and evolutionary ecology. A metacommunity is defined as a collection of local communities connected by dispersal, yet there has been little consideration of how the arrangement of local communities within a metacommunity can affect its properties. Nearly all studies either have not explicitly incorporated the arrangement of local communities, or they have conceived unrealistic dispersal networks, with communities randomly/uniformly distributed in space and/or with equal connectivity. We contend that many metacommunities are composed of highly structured dispersal networks with constraints on connectivity that have significant consequences for metacommunity properties. Riverine ecosystems are one common example where dispersal should be highly constrained by a linear-type network structure. We predicted that dispersal limitation should manifest as declining community similarity with distance (i.e., the distance-decay relationship) while a lack of change with distance would indicate that community structure is driven by mechanisms other than dispersal limitation (e.g., species sorting). For riverine systems, dispersal limitation should increase moving downstream and be stronger when connectivity is network-constrained (i.e., distances calculated along the network). Upstream sites should be less dispersal limited, since such sites are farthest from other sources of colonists along the network and, over evolutionary time, communities in these locations are likely highly adapted to local habitat conditions. In terms of metacommunity predictions, a linear network structure should impose species-sorting upstream, and patch-dynamics downstream if dispersal depends on network connectivity. Using large-scale survey data for 91 fish and invertebrate communities on the Appalachian Plateau in western Maryland (USA), we asked: Do the constraints imposed by linear-type dispersal networks affect metacommunity properties?

Results/Conclusions Our analysis of the distance-decay relationship revealed dispersal limitation for both fish and invertebrate communities, and limitation was influenced by network structure. The slope of the distance-decay relationship was steeper and stronger when based on network versus Euclidean distance between communities. Furthermore, the distance-decay relationship differed with a community's position in the network. Slopes of the distance-decay relationship were steeper and stronger for downstream versus headwater sites. Finally, dispersal limitation was stronger for fish than invertebrates - a result that is biologically appealing since many aquatic invertebrates have terrestrial adult stages allowing them to circumvent network-constrained connectivity, versus fish that must disperse along aquatic routes. We suggest that the metacommunity concept is a useful framework within which to consider linear-type networks, but the structuring mechanisms might change depending on location within such networks.