Many studies suggest that aside from environmental variables, such as topography and climate, species-specific ecological traits are relevant to explain the geographic distribution of intraspecific genetic lineages. Here, we investigated whether and to what extent incorporating such traits systematically improves the accuracy of random forest models in predicting genetic differentiation among pairs of localities. We leveraged available ecological datasets for birds and tested the inclusion of two categories of ecological traits: dispersal-related traits (i.e., morphology and foraging ecology) and demographic traits (such as species survival rate and generation length). We estimated genetic differentiation from published mitochondrial DNA sequences for 31 species of birds (1,801 total genetic samples, 526 localities) in the Atlantic Forest of South America. Aside from the aforementioned ecological traits, we included geographic, topographic and climatic distances between localities as environmental predictors. We then created models using all available data to evaluate model uncertainty both across space and across the different categories of predictors. Finally, we investigated model uncertainty in predicting genetic differentiation individually for each species (a common challenge in conservation biology). Our results show that while environmental conditions are the most important predictors of genetic differentiation, model accuracy largely increases with the addition of ecological traits. Additionally, the inclusion of dispersal traits improves model accuracy to a larger extent than the inclusion of demographic traits. Similar results are observed in models for individual species, although model accuracy is highly variable. We conclude that ecological traits improve predictive models of genetic differentiation, refining our ability to predict phylogeographic patterns from existing data. Additionally, demographic traits may not be as informative as previously hypothesized. Finally, prediction of genetic differentiation for species with conservation concerns may require further careful assessment of the environmental and ecological variations within the species range.