Identifying the factors controlling local community structure is a central
problem in ecology. Ecologists frequently use regression to test fur a nonl
inear saturating relationship between local community richness and regional
species pool richness, suggesting that species interactions limit the numb
er of locally coexisting species. However, communities in different regions
are not independent if regions share species. We present a Monte Carlo tes
t for whether an observed local-regional richness relationship is significa
ntly different from that expected when regions are nonindependent and speci
es interactions do not limit community membership. We illustrate this test
with data from experimental microcosm communities. A conventional F-test su
ggests a significant saturating relationship between realized community ric
hness and species pool richness. However, the Monte Carlo test fails to rej
ect the null hypothesis that species interactions do not affect community r
ichness. Strong species interactions do not necessarily set an absolute upp
er limit to the number of locally coexisting species.