Mutualism benefits partner species, and theory predicts these partnerships can affect the abundance, diversity, and composition of partner and non-partner species. We used 16 years of monitoring data to determine the ant partner species of tree cholla cacti (Cylindropuntia imbricata), which reward ants with extrafloral nectar in exchange for anti-herbivore defense. These long-term data revealed one dominant ant partner (Liometopum apiculatum) and two less common partners (Crematogaster opuntiae and Forelius pruinosus). We then used short-term characterization of the terrestrial ant community by pitfall trapping to sample partner and non-partner ant species across ten plots of varying cactus density. We found that the dominant ant partner tended a higher proportion cacti in plots of higher cactus density, and was also found at higher occurrence within the pitfall traps in higher density plots, suggesting a strong positive feedback that promotes ant partner occurrence where plant partners are available. Despite the strong association and increased partner occurrence, ant community-wide effects from this mutualism appear limited. Of the common ant species, the occurrence of a single non-partner ant species was negatively associated with cactus density and with the increased presence of L. apiculatum. Additionally, the composition and diversity of the ant community in our plots were insensitive to cactus density variation, indicating that positive effects of the mutualism on the dominant ant partner did not have cascading impacts on the ant community. This study provides novel evidence that exclusive mutualisms, even those with a strong positive feedback, may be limited in the scope of their community-level effects.