The species pool concept has advanced our understanding for how biodiversity is coupled at local and regional scales. However, it remains unclear how species pool size, the number of species available to disperse to a site, influences community assembly across spatial scales. We provide one of the first studies that assesses diversity across scales after experimentally assembling grassland communities from species pools of different sizes. We show that species pool size causes scale‐dependent effects on diversity in grasslands undergoing restoration by altering the shape of the species–area relationship (SAR). Specifically, larger species pools increased the slope of the SAR, but not the intercept, suggesting that dispersal from a larger pool causes species to be more spatially aggregated. This increased aggregation appears to be caused by sampling effects due to fewer individuals arriving per species, rather than stronger species sorting across variation in soil moisture. These scale‐dependent effects suggest that studies evaluating species pools at a single, small scale may underestimate their effects, thereby contributing to uncertainty about the importance of regional processes for community assembly and their consequences for ecological restoration.