GIS assumes an increasingly prominent role in niche modelling, particularly for those applications that either predict a species' niche or estimate the environmental factors that determine a species' distribution. The use of GIS to model population-level distributions has received considerably less attention. We developed a Monte Carlo method designed to test for statistical associations between population distribution and different types of spatial data. This method examines the spatial distribution between species presence points and distance to key vegetation, comparing the association for the observed data to that of a random data set. Poison frogs (genus Dendrobates) display a diverse array of complex parental care strategies, which are linked by an elaborate network of interactions between ecological and social factors. Territorial species vary significantly in their associated reproductive strategies, reproductive resources and size. Species with higher levels of cooperation in parental care have been hypothesized to be more strongly territorial. Here we investigate differences in spatial use patterns between two species of poison frogs, D. imitator and D. variabilis, that differ in parental care type (uniparental male care versus biparental care, respectively). We predicted that: (1) D. imitator would be more closely associated with phytotelmata than would D. variabilis; (2) home ranges of D. imitator pairs would be more inclusive and overlap more than those of D. variabilis pairs; (3) D. imitator pairs would maintain smaller home ranges than D. variabilis pairs; (4) D. imitator pairs would maintain spatially exclusive home ranges. Our results strongly support the prediction that differences in parental care of these two species are reflected in their spatial habitat use, particularly with regard to pool fidelity, territoriality, and home range size and overlap. © 2008 The Association for the Study of Animal Behaviour.