The study of ecological communities through time can reveal fundamental ecological processes and is key to understanding how natural and human pressures will affect biodiversity. Most studies of ecological communities through time consider only one or a few summary measures (e.g. species richness, total abundance), which might neglect important aspects of community structure or function. We studied temporal variation in several measures of species diversity, size diversity, and species composition in an intensively sampled bird community to determine whether different biodiversity measures change synchronously. We used a novel function regression model, which supports the study of diversity measures that are distributions (e.g. species abundance distributions) alongside measures that are scalar values (e.g. species richness). Most diversity measures changed predictably within years, but inter-annual changes in size diversity and species composition were not reflected in species diversity. Within and among years, there was considerable variation in distributional measures that was not captured in scalar measures. Predictable variation within years probably was related to seasonal variation in weather patterns or food availability, but variation in size diversity among years probably resulted from stochastic changes in species composition. These results suggest that species and size diversity may be decoupled, and that inferences on scalar diversity measures might not reflect fundamental changes to community structure or function. Our method supports the inclusion of size-based measures and distributional measures in ecological analyses, and broader uptake of our approach is likely to provide new insight into the processes structuring ecological communities, and inform the links between structure and function in ecological communities.