Search

To be adequate, evenness indices must satisfy three criteria: (1) independence of species richness; (2) independence of sample size; and (3) unambiguous relationship to evenness. I examine Bulla's evenness index in the light of these criteria, logically implying a failure to satisfy the first, and experimentally proving failures to satisfy the second and third. Therefore, this index is inadequate. I also examine Bulla's heterogeneity index. Being based on his evenness measure, it is also inadequate. Moreover, the index is not the correct heterogeneity measure under his own approach. I derive the correct index and demonstrate its kinship to the evenness index. Even this correct heterogeneity measure lacks a future because no adequate diversity indices can be obtained from Bulla's overlap. By re-examining the savanna data with which Bulla attempted to show the superiority of his index, 1 prove his analysis to be an artifact. In response to Bulla's criticism on G 2,1 , I demonstrate that his argumentation is wrong, and reassure that G 2,1 is an adequate evenness index. To conclude, I offer three general considerations on index testing that, if followed, will prevent a repetition of Bulla's errors.