Diversity: A Sampling Study

TLDR

The rarefaction method of Sanders (1968) compares diversity across samples of differing sizes, while two indices relate to intra‑ versus interspecific encounter probabilities as indirect evenness measures and two others directly assess evenness. The study proposes scaling diversity indices to their possible value ranges for comparability and recommends choosing an index based on the desired property to measure diversity. Four diversity indices, including a new one, are compared by evaluating their sensitivity to changes in individual and species counts, individual distribution among species, and spatial aggregation. Rarefaction overestimates expected species richness, with the overestimation magnitude driven by individual distribution among species and moderate spatial aggregation.

Abstract

The rarefaction method proposed by Sanders (1968) to allow one to compare the diversity of samples composed of different numbers of individuals and species is compared with the results that would be obtained if smaller random samples were taken from a universe defined by the observed sample. The rarefaction technique overestimates the number of species that would be expected by sampling. The magnitude of the overestimation is strongly influenced by the distribution of individuals among species in the original sample and by the presence of even moderate spatial aggregation. Four diversity indices, one of them newly proposed here, are compared in terms of their response to changes in the numbers of individuals and species, in the distribution of the individuals among the species, and in spatial aggregation. To increase comparability, it is suggested that scaled values be used; the scaling should be done in terms of the possible range of values of the index for the given numbers of individuals and species. Two indices can be related to probabilities of intra- versus interspecific encounters, and are only indirect measures of the evenness of distribution of individuals among species. The other two are direct measures of evenness. Which index is selected should be determined on the basis of what property is to be used as a measure of diversity.