Issue: Approaches to predicting species assemblages through stacking individual niche-based species distribution models (S-SDMs) need to account for community processes other than abiotic filtering. Such constraints have been introduced by implementing ecological assembly rules (EARs) into S-SDMs, and can be based on patterns of functional traits in communities. Despite being logically valid, this approach has led to a limited improvement in prediction, possibly because of mismatches between the scales of measurement of niche and trait data. Evidence: S-SDM studies have often related single values of a species’ traits to environmental niches that are captured by abiotic conditions measured at a much finer spatial scale, without accounting for intraspecific trait variation along environmental gradients. Many evidences show that omitting intraspecific trait variation can hinder the proper inference of EARs from trait patterns, and we further argue that it can therefore also affect our capacity to spatially predict functional properties of communities. In addition, estimates of environmental niches and trait envelopes could vary depending on the scale at which environmental and trait measurements are made. Conclusion: We suggest that to overcome these limitations, surveys sampling both niche and trait measurements should be conducted at fine scales along wide environmental gradients, and integrated at the same scale to test and improve a new generation of spatial community models and their functional properties.