Some Comments on Niche Analysis in Canonical Space

摘要: Discriminant functions have been used to identify axes of niche separation, and dispersion locations on these have, in turn, interpreted as representing width specialization. We discuss the influence sampling scheme choice measures results an analysis conducted canonical space. Specifically, habitat variables measured at random should provide a more representative measure available than would recording only where animals are observed. Species presence information alone tends bias mean toward one most common species is found. If specialization be by distance from overall habitat, each station receive equal weighting. The straightforward breadth seems variance or standard deviation scores. squared centroid can then comparable multidimensional measure. Finally, we suggest that overlap via joint probability density function rather area concentration ellipses.