Null‐Model Tests for Competitive Displacement: The Fallacy of Not Focusing on the Whole Community

Abstract

Null—model tests for competitive displacement involve comparisons of species' relationships in an observed community with those of randomly constructed communities. I examine the performance of several parameters that have been used to characterize community structure for the purpose of this comparison. These parameters have in common the fact that they focus on selected species pairs within the community rather than considering the community as a whole. Fleming and Partridge (1984) proposed a parameter for examining flowering phenologies that focuses on one species pair at a time. Fleming and Partridge compared the observed overlap between members of a particular species pair with the distribution of possible overlap values for this pair from randomly constructed communities. A second parameter, the overlap between a particular species and all other species combined, was employed in a similar way. Using hypothetical phenologies I show that using these parameters (a) will always result in acceptance of the null hypothesis when in fact a segregated flowering pattern exists, and (b) may lead to acceptance of the competition hypothesis when in fact the null hypothesis is true. A community—level parameter, mean pairwise overlap, is shown to make neither or these errors. Simberloff and Boecklen (1981) employed a family of parameters to examine the competition hypothesis that there is a constant difference in log body size (=a constant ratio of body sizes) between adjacent species ranked according to body size. Their parameters focus on the relative magnitude of the differences between two pairs of adjacent species, such as the two pairs with the smallest and largest log body size differences. Using hypothetical communities I show that these parameters may not always recognize a community with competitive displacement, and may mistake a clumped community for one with displacement. Several community—level parameters are shown to have greater power in discerning competitive displacement.