AbstractAimWe present a mechanistic niche model that integrates the demography of competing plant species in a metabolic, stochastic framework. In order to explore the model's ability to generate multiple species and community patterns, we assessed trait composition, richness gradients and spatial distributions of species ranges and abundances of simulated communities.LocationHypothetical, sloped plane.MethodsStage‐structured populations of species differing in traits and habitat requirements competed for space in a grid‐based model. Demographic processes (recruitment, reproduction, mortality, dispersal) and resource competition were explicitly simulated. Demographic rates and carrying capacity followed metabolic constraints. We simulated 50 species pools until reaching stable communities. Species pools were initialized with 400 species that had random traits and habitat requirements. The habitat requirements generated potential distributions of richness, range and abundance, whereas the simulations yielded realized distributions.ResultsThe communities assembled in the simulations consisted of species spread non‐randomly within trait space. Potential species richness peaked at mid‐elevations, whereas realized richness was slightly shifted towards higher elevations. For 11% of all species, the highest local abundances were found not in the most suitable habitat, but in suboptimal conditions. 53% of all species could not fill the climatically determined potential range. The ability to fill the potential range was significantly influenced by species traits (e.g. body mass and Allee effects) and species richness.Main conclusionsSpatial and trait properties of surviving species and of equilibrium communities diverged from the potential distributions. Realized richness gradients were consistent both with patterns observed in nature and those expected from null models based on geometrical constraints. However, the divergences between potential and realized patterns of richness, ranges and abundances indicate the importance of demography and biotic interaction for generating patterns at species and community levels. Consequently, bias in correlative habitat models and single‐species mechanistic models may arise if competition and demography are neglected. Additionally, competitive exclusion provides a mechanistic explanation for the low transferability of single‐species niche models. These results confirm the usefulness of mechanistic niche models for guiding further research integrating ecological niche, community ecology and biogeography.