Abstract
Paraponera clavata (Fabricius 1775) (Formicidae: Paraponerinae) is a widely distributed Neotropical ant whose large size has attracted the attention of numerous collectors. Working from museum specimens, a georeferenced database of collection localities was developed. This database then served as the source for computer generated predictive distribution maps. Annual rainfall was the most important variable chosen by the computer model to predict the distribution of P. clavata, both on the scale of the neotropics and at a finer scale at the northern end its distribution in Costa Rica and Nicaragua. When the model was forced to use vegetation as the first predictive variable, the Neotropical model used temperature and rainfall variance as additional variables, while the Mesoamerican model used both climatic and soils variables. Overall, the modeling suggests that P. clavata is more sensitive to abiotic factors (rainfall, temperature, soils) than to biotic factors (vegetation type) in its distribution, although this conclusion comes with the caveat that the vegetation types used in the model are quite generalized. Predictive distribution mapping holds great promise for generating more precise representations of insect distributions, thereby allowing better tests of the extent of distribution overlaps and other community relationships.