Spiders are a functionally important taxon in forest ecosystems, but the determinants of arboreal spider beta diversity are poorly understood at the local scale. We examined spider assemblages in 324 European beech (Fagus sylvatica) trees of varying sizes across three forest stands in Würzburg (Germany) to disentangle the roles of tree architecture, spatial distance, and dispersal capacity on spider turnover across individual trees. A large proportion of tree pairs (66%) showed higher compositional dissimilarity in spider assemblages than expected by chance, suggesting prominent roles of habitat specialization and/or dispersal limitation. Trees with higher dissimilarity in DBH and canopy volume, and to a lesser extent in foliage cover, supported more dissimilar spider assemblages, suggesting that tree architecture comprised a relevant environmental gradient of sorting spider species. Variation partitioning revealed that 28.4% of the variation in beta diversity was jointly explained by tree architecture, spatial distance (measured by principal coordinates of neighbor matrices) and dispersal capacity (quantified by ballooning propensity). Among these, dispersal capacity accounted for a comparable proportion as spatial distance did (6.8% vs. 5.9%). Beta diversity did not significantly differ between high- and low-vagility groups, but beta diversity in species with high vagility was more strongly determined by spatially structured environmental variation. Altogether, both niche specialization, along the environmental gradient defined by tree architecture, and dispersal limitation are responsible for structuring arboreal spider assemblages. High dispersal capacity of spiders appears to reinforce the role of niche-related processes.

Point process models performed better when life stages are included, indicating the importance of considering life stage when modeling spatial distributions for understanding community assembly. The finding that habitat acts weakly and non-habitat processes act dominantly in determining spatial distributions of species suggests a strong dependence of spatial patterns on non-habitat processes.