SPECIES ABUNDANCE PATTERNS - THE PROBLEM OF TESTING STOCHASTIC-MODELS

1. Tokeshi (1990) proposed a goodness-of-fit test to distinguish among a general class of stochastic species abundance models using field da ta. This test is a good first step, but it is impaired by some shortco mings: there is no adjustment for the number of species (the rejection rate increases with the number of species in a data set); the varianc e of the generated abundance distributions are not taken into account (data sets with higher variance than a model are not rejected). 2. We propose an alternative Monte Carlo test that overcomes these problems. This is a versatile test that can be used with any stochastic model g enerating distributions.