Abstract Background and Aims Species diversity is expected to increase with environmental heterogeneity. For plant communities, this pattern has been confirmed by numerous observational studies. Yet, experimental studies yield inconsistent results potentially because of how experiments create soil heterogeneity or because seeds were sown homogeneously. Using a field experiment, we tested how soil heterogeneity, plant spatial aggregation via seed arrival, and grain size influence plant species richness in a restored grassland. Methods We manipulated soil heterogeneity and seed arrival in 0.2 × 0.2 or 0.4 × 04 m patches within each 4.0 × 4.6 m plot and allowed community assembly to occur for 4 growing seasons. Results Despite quantifiable soil differences, soil heterogeneity did not impact total or sown species richness, but did weakly influence non-sown richness. Richness differences were driven by non-sown plant species that likely exhibited higher establishment in aggregated plots due to decreased interspecific competition and conspecific facilitation. Conclusion Our results suggest that fine-scale soil heterogeneity weakly affects prairie plant diversity, but heterogeneous plant spatial structure can have a stronger effect on diversity. These results suggest that plant colonization may be the primary source of environmental heterogeneity and may explain inconsistent results from soil heterogeneity experiments.