Detecting competitive interactions is important for predicting species responses to environmental change, but it remains challenging, especially over large scales. Based on classical coexistence theory, competition should be most important between species with similar ecological traits (i.e., diet, habitat). However, species with life-history traits that allow them to be very prolific can have competitive effects even with subtle ecological overlap. Such species may be able to achieve such high abundances that they dominate all other competitors (i.e., “dominant entities”). We tested for competition and the importance of dominant entities in small mammal communities (n = 68 species) at 44 sites in 18 regions across the United States using the US National Ecological Observatory Network (NEON). We based inference on changes in abundance over time while also accounting for weather and habitat factors using a dynamic generalized joint attribute modeling framework. We compared relative interaction strengths inferred from our model to predictions expected under coexistence theory based on ecological and life history traits. We used three different cutoff levels to classify a species (or set of species) as dominant entities: >= 45 % of the total interaction strength, > 50 % of the total abundance at a site and a relatively large number of strong interactions (> 10 % of total interaction strength at the site). Predictions of competition based on coexistence theory were not well supported by model results (mean = 0.25, SE = 0.03). Our model indicated more sites with dominant entities than predicted by coexistence theory. In particular, our model identified three generalist Peromyscus species as the dominant entities at more sites (53.1 % vs. 0 %), associated with higher mean interaction strength at a site (58.0 % [SE = 7.5 %] vs 13.8 % [SE = 1.4 %]) and a higher proportion of strong interactions with other species at a site (46.2 % [SE = 4.4 %] vs. 26.7 % [SE = 1.7 %]). Model results more closely matched large-scale observed abundance patterns than did theory-based predictions, suggesting that accounting for dominant entities is important for characterizing competitive interactions. Our study shows the potential of using changes in abundance over time to make inference on community change while accounting for competitive interactions, where reliance on presence/absence and/or ecological traits alone could miss important dynamics.