Dental ecometric traits in large herbivores have been used to reconstruct palaeoenvironments, given the known relationships that these traits have to modern environments (such as the negative correlation between hypsodonty and precipitation). These techniques have largely focused on environments in North America and particularly both Eurasia and Africa, and consequently ecometric models have been trained on groups of herbivores that are most significant in those regions today (Artiodactyla, Perissodactyla, Primates and Proboscidea). The extent to which these relationships are persistent in communities with other dominant herbivores has never been tested. Because South America has been isolated for most of its history, fossil assemblages in the continent contain other clades, which likely have different trait–environment relationships due to their evolutionary history. Quantitative testing of these relationships and tailored regional models for South America, particularly those incorporating dietary information, will improve palaeoenvironmental reconstructions in the continent using mammal communities.