Plant communities often exhibit high diversity, even though pairwise experiments usually result in competitive hierarchies that should result in competitive exclusion. Such experiments, however, do not typically allow expression of spatial traits, despite theoretical studies showing the potential importance of spatial mechanisms of diversity maintenance. Here we ask whether, in a clonal plant model system, spatial trait variation is more likely than growth trait variation to maintain diversity. We used a field-calibrated, spatially explicit model to simulate communities comprising sets of four simulated species differing in only one of a suite of architectural or growth traits at a time, examining their dynamics and long-term diversity. To compare trait manipulation effects across traits measured in different units, we scaled traits to have identical effects on initial productivity. We found that in communities of species differing only in an architectural trait, all species usually persist, whereas communities of species differing only in a growth trait experienced rapid competitive exclusion. To examine the roles of equalizing and stabilizing mechanisms in maintaining diversity, we conducted reciprocal invasion experiments for species pairs differing only in single traits. The results suggest that stabilizing mechanisms cannot account for the observed long-term co-occurrence. Strong positive correlations between diversity and similarity both in monoculture carrying capacity and reciprocal invasion ability suggesting equalizing mechanisms may instead be responsible.