A new method of species-reduction for ecological data

Abstract

Existing strategies for species-reduction in data matrices from species-rich communities are critically reviewed, and a new method is proposed. It is based on a two-parameter model; each species can be associated with an ‘eident value’, which is a measure of the extent to which that species discriminates between site-groups. The method is applied to a small segment of agronomic data, and to a rainforest survey, and in both cases is shown to produce meaningful results. Application of the method promises greatly to reduce the computing time currently needed for two-parameter analyses, and therefore to allow more extensive search techniques to be employed.