Null Models Reveal Differing Drivers of Multidimensional Beta Diversity Change in Invaded Metacommunities

Aim Invasion‐driven changes in beta diversity (Δβ) are multifaceted (taxonomic, functional, and phylogenetic) and are comprised of three components (total beta diversity, replacement, and richness difference); therefore, they represent a nine‐dimensional phenomenon. Invasion‐driven trajectories of multidimensional Δβ are inconsistent and can vary based on the geographic origin of nonnative species, native community structure, and site attributes. Here, we examine the drivers of multidimensional Δβ to better understand how nonnative species alter biodiversity. Location 1023 wadable‐stream fish communities across the conterminous United States. Methods We characterized nine dimensions of Δβ as changes in local contribution to beta diversity (ΔLCBD) derived from pairwise beta diversity of all 1023 sites between contemporary and native‐only assemblages at each site. We used redundancy analysis and variance partitioning to identify and estimate the relative importance of 18 Δβ drivers related to nonnative species origin, propagule pressure, abiotic factors, and native community structure. We compared results between those from raw and null model standardized LCBD. Results We found an overall trend of taxonomic homogenization due to nonnative species occurrences. However, null model standardization revealed inconsistent directional change in functional and phylogenetic Δβ, and also revealed Δβ patterns and drivers that differed from those of raw values. Sites did not contribute evenly to overall Δβ trends, with nonnative species origin and native community structure explaining the most variation in ΔLCBD (16% and 15% respectively). Nonnative occurrences, regardless of origin, contributed to taxonomic homogenization, with the magnitude of change dependent on origin. Native community structure dictated the direction of phylogenetic Δβ, with phylogenetically distinct sites contributing to homogenization. Conclusions Our findings demonstrate the utility in using LCBD to identify site‐level drivers of Δβ and the need for null model standardization of multidimensional Δβ metrics. Additionally, our results show that spatially explicit definitions of nonnative species are needed to gain robust estimates of Δβ.