Competition has been shown to be an important factor in natural system
s. Yet, its effect on community structure remains poorly understood an
d is often disputed. Null models have been employed by researchers in
recent years, but these generally suffer from a lack of power estimati
on and unclear distinction between statistical significance and biolog
ical importance, Models that employ an appropriate alternative hypothe
sis allow for power estimates and for quantitative evaluation of the e
ffects of competition. We have developed methodologies to (1) allow fo
r testing the significance of competition in natural communities, (2)
determine the power of these estimates, and (3) produce maximum likeli
hood estimates of interaction parameters for these communities. Using
these methodologies we show that small mammal communities at the inter
face between Valdivian temperate rain forest and Patagonian steppe in
southern South America appear significantly and importantly structured
by competition, although we cannot distinguish between resource compe
tition, interference competition, and apparent competition. Secondaril
y, we show that the geographic origin of the component species (Valdiv
ian vs. Patagonian) is relatively more important than their habitat af
finities in structuring these communities.