AbstractAimWe aimed to test whether contributions of individual species (SCBD) and contributions of single sites (LCBD) to overall beta diversity can be predicted by species metrics and species characteristics and also by community metrics and ecological variables, respectively.LocationA mainland‐island landscape in the southern Brazilian Atlantic Forest domain.MethodsDung beetles were sampled along 100 sampling units within four large forest fragments. We partitioned beta diversity into SCBD and LCBD aiming to explore their relationships to species and site characteristics, respectively. We then used a combination of multivariate methods and beta regression analyses to examine patterns in SCBD and LCBD. The occupancy, total abundance, niche position, niche breadth and biological traits of species were used as predictor variables for SCBD values. Community metrics, environmental and spatial variables, and temporal patterns in the beta diversity components of turnover and nestedness were used as predictor variables for LCDB values.ResultsWe found that SCBD was strongly related to various species characteristics, such as occupancy, abundance and niche position, but was not related to biological traits of species and niche breadth. In particular, occupancy and its quadratic term showed a very strong unimodal relationship with SCBD, suggesting that intermediate species in terms of site occupancy contribute most to beta diversity. LCBD was mostly explained by variation in species richness, with a negative relationship being detected. Litter height and large‐scale spatial variables were also important in explaining variation in LCBD.Main conclusionsSCBD and LCBD were highly predictably related to species occupancy and species richness, respectively. Environmental conditions and large‐scale spatial variables also correlated with LCBD values. Understanding the determinants of SCBD and LCBD may thus hold a key to various general ecological, bioassessment and conservation issues. Protecting sites with high LCBD values may be a suitable approach to practical biodiversity conservation.