Functional traits provide us with a window to better understand and predict species performance and hence the outcome of the community assembly process, i.e. community structure, as well as its consequences, i.e. ecological function. Trait-based approaches may be highly desirable, but there are numerous obstacles to making progress besides the obvious hurdle of simply measuring a set of traits in a species pool. Understanding the controls on ecological function, such as primary productivity, will most likely involve traits aggregated and averaged across the community. Even so, we may find scaling problems between traits and aerial-based process rates. Average community traits may also define community composition if communities are assembled primarily following Shipley's ecological Entropy Maximization (EM) model, where species with modal trait values are most abundant. Under this model, a restoration plan might simply include species with traits similar to the expected value from a trait-environment model (also required, but rarely known). Species with quasi-extreme trait values may either be omitted or targeted for more intensive management. Alternatively, communities may not be assembled following EM. In this case, trait averages may be of little use and measures such as trait ranges, their maxima and minima, as well as measures of species packing and trait diversity, are likely needed to define composition. Alternatively, we may simply model key aspects of species performance that are key to restoration (e.g. establishment) as a function of traits, but this approach may be of limited generality.