The effective number of species (ENS) or Hill Number has been proposed as a robust measure of species diversity that overcomes several limitations of both diversity indices and species richness. However, it is not yet clear if ENS improves interpretation of biodiversity monitoring data, which has implications for resource management decisions. We calculated ENS as λ1/(1-q) with λ=∑(pi)q and pi = proportional abundance (0 ≤ q ≤ 2 at intervals of 0.1) for two large datasets: 937 fish and 917 mussel samples collected from Illinois, USA streams and rivers. Fish assemblages were sampled with electric seines over reaches 20-times long as wide, and mussel assemblage samples were standardized based on 4-person-hour searches. With increasing q-value, ENS increasingly weights abundance and thus evenness. At q = 0, ENS is simply species richness; at q converging to 1, ENS is eH(H = Shannon Index); and at q = 2, ENS is 1/D (D = Simpson Index). We used random forest regression to model ENS-environment relationships, where environmental predictors included variables describing climate, geology, watershed topography, land-use, soil, and spatial connectivity.
Results/Conclusions
The amount of variation in ENS across the fish and mussel assemblages that was associated with environmental gradients (pseudo-R2) steadily decreased with increasing q as did the relative importance values of key predictors. This result indicates that weighting species abundance reduces our ability to detect, and interpret, how physical habitat likely influences species diversity, which compromises the utility of ENS as a monitoring tool for assessing the impacts of human disturbances such as land-use and climate change. We suggest that researchers and managers estimate species richness and evenness separately instead of combining them into a single measure as occurs in both ENS and traditional diversity indices.