Measurements of trait community composition are known to be sensitive to the way species abundance is assessed, but not to what extent. This was investigated by considering two of the most commonly used indices of community trait composition, trait averages and functional diversity, in bee communities along a post-fire environmental gradient. The indices were computed using three different species abundance measurements (log and unlog number of individuals and species occurrence only) and 5 traits. For certain traits, the responses of the indices to fire varied according to how species abundance was measured. The measurements that took species abundance into account in the most distinct way (e.g., occurrence vs. unlog data) produced the least similar results for all traits. Species were then grouped into different classes on the basis of their relative abundance (i.e., dominants, subdominants, and rare species). As a result, the measure that attaches the highest importance to the abundance of species (unlog data) related mostly to the dominant species traits, while the measure attaching the lowest (i.e., species occurrence) related more to rare species traits. Species diversity was mostly independent of trait averages and functional diversity, regardless of the measure of species abundance used. We also quantified functional redundancy (i.e., the potential minus the observed functional diversity in each community). When more weight was attached to species abundance, redundancy decreased and tended to be less correlated with species diversity. Overall, the way species abundance is taken into consideration in indices of functional composition offers promising insights into the way community assembly mechanisms respond to environmental changes.