Pedro Segurado; William E Kunin; Ana F Filipe; Miguel B Araújo.

Inferring biotic interactions from the examination of patterns of species occurrences has
been a central tenet in community ecology, and it has recently gained interest in the context of singlespecies
distribution modelling. However, understanding of how spatial extent and grain size affect such
inferences remains elusive. For example, would inferences of biotic interactions from broad-scale
patterns of coexistence provide a surrogate for patterns at finer spatial scales? In this paper we
examine how the spatial and environmental association between two closely related species of
freshwater turtles in the Iberian Peninsula is affected by the geographical extent and resolution of the
analysis. Species coexistence was compared across spatial scales using five datasets at varying spatial
extents and resolutions. Both similarities in the two species' use of space and in their responses to
environmental variables were explored by means of regression analyses. We show that a positive
association between the two species measured at broader scales can switch to a negative association at
finer scales. We demonstrate that without examination of the effects of spatial scale when investigating
biotic interactions using co-occurrence patterns observed at coarse resolutions, conclusions can be
deeply misleading.