Landscape features and study design affect elements of metacommunity structure for stream fishes across the eastern U.S.A.

Abstract Metacommunity ecology includes connectivity in the investigation of local and regional processes to understand community assembly. The elements of metacommunity structure (EMS) framework classifies metacommunities into categorical archetypes based on patterns of three metrics indexing β diversity: coherence, turnover, and boundary clumping. Although the EMS framework is most commonly used to classify metacommunity types, an elements‐based approach of examining how factors affect each specific continuous EMS variable could provide a more mechanistic understanding of how metacommunities are structured at the landscape scale. Moreover, few studies have sought to quantify how methodological issues such as number and spacing of local communities affect observed outcomes of EMS‐based analyses. Using a large dataset of stream fish occurrences in the eastern U.S.A., we used mixed effects models to investigate how (1) landscape‐scale variables and (2) methodological issues such as number and location of sampling sites affect coherence, turnover, and boundary clumping individually; as well as (3) how landscape‐scale variables influence overall metacommunity patterns derived from the EMS framework. Coherence, turnover, and boundary clumping were related to temperature, density of dams, developed land use, and γ diversity. Interestingly, distance among sampled sites in metacommunities negatively affected turnover, and the number of sampling sites positively affected all three variables. Elevation affected overall observed metacommunity patterns, with metacommunities transitioning from Clementsian to clumped species loss patterns with increasing elevation. Our results suggest that metacommunity structure is affected by both natural and anthropogenic landscape‐scale variables. Observed metacommunity properties are also influenced by sampling density and site location within the catchment. Accounting for important natural, anthropogenic, and methodological issues will be critical for improving the inferential power of metacommunity analyses to begin understanding which landscape‐scale variables should be the focus of conservation and management of fish communities at a catchment scale.