If interspecific competition is a strong structuring force of communities, ecologically similar species should tend to have spatial ranges at local scale that do not overlap. Experimental testing of this hypothesis becomes impracticable with large communities. One possibility to tackle this issue is a correlational approach, by comparing the matrix of niche overlap with that of spatial overlap. The use of the standard Mantel test is however impaired by the non-linearity in the relationship of the two descriptors: in a competitively structured assemblage, species with high niche overlap are expected to be segregated spatially, but species with small niche overlap may or may not exhibit high spatial overlap. To overcome this problem, we devised an original randomization test, which was run for three data-sets comprising frogs, lizards, and birds along altitudinal gradients. The test yielded intriguing results: reptiles and birds revealed an adjustment that would reduce the potential for interspecific competition, while amphibians showed the opposite trend, that is, ecologically similar species co-occurred more than expected by chance. Frogs may be more constrained by resource requirements, possibly breeding sites, than by competition. Our test will help to assess the generality of this pattern with other data-sets.