A well-known model of community assembly predicts that only certain co
mbinations of species can coexist, and those that do differ in critica
l aspects of morphology or size that allow them to exploit different r
esources, The model has been controversial, because it has been nearly
impossible to confirm experimentally and because many of the reported
patterns are no different than expected by chance. I addressed an asp
ect of this model by investigating whether very similar species of str
eam fishes co-occurred less frequently than more dissimilar species. I
sampled 219 assemblages from the Red River basin in Texas, Oklahoma,
and Arkansas; calculated an index of co-occurrence between all possibl
e pairs of cyprinid (minnow) species; and estimated their morphologica
l similarity and phylogenetic relatedness. Only sympatric or parapatri
c pairs were analyzed to allow for the possibility of ecological inter
actions and to make the analysis independent of the size of the area s
ampled, Randomization tests demonstrated a significantly low degree of
co-occurrence between morphologically very similar pairs compared wit
h less similar pairs of species. However, there was no significant dif
ference in co-occurrence between the species most related phylogenetic
ally and more distantly related species. Of six alternative hypotheses
, interspecific competition best explained the pattern.