Assessing species abundance unevenness within and between plant communities

A simple to calculate and statistically testable method is proposed for assessing species abundance unevenness or dominance concentration within plant communities. Dominance rather than evenness is considered the preferred measure, because perfect evenness can be defined and used as a benchmark for comparing communities regardless of floristic richness. Four dominance (Simpson, Berger-Parker, McIntosh, and a proposed [D]) and four evenness (Carmargo, Gini, Shannon, and Williams) indices are comparatively analyzed. The Simpson, McIntosh, Gini, and Williams indices were correlated with species richness, the Berger-Parker index was correlated with total species cover, the Shannon index over-estimated evenness, the Simpson index under-estimated dominance concentration, and a nonlinear relationship occurred between the Simpson and all of the evenness indices. Only DW and the Carmargo indices fulfilled the technical requirements established for evenness with appropriate reversals of criteria for assessing a dominance concentration index. Determination of DW was based on the maximum difference between cumulative proportion (also referred to as Lorenz curve or partial order) and perfect evenness values. The conventional assumption that dominance (D) is the complete opposite of evenness (E), as assessed by indices, was found in practice to be lacking, without the inclusion of an error term (i.e., 1 = E + D + error). Therefore, both dominance concentration and evenness should be reported when characterizing plant communities