The effect of distance measure choice on the power of various permutation-based multivariate tests for difference.
by
Cameron Hurst
Griffith University
Coauthors: Janet Chaseling (Griffith University), Carla Catterall (Griffith University)

Multivariate techniques and in particular ordination methods play a central role in the analysis of ecological communities. One of the main properties distinguishing various ordination methods commonly employed in ecological data analysis is the distance measure underlying the models. While a number of studies have considered the effect of distance measure on descriptive unconstrained ordination techniques (e.g. Faith et al., 1987; Kenkel & Orloci, 1986), little work has been done on examining how distance measure selection might influence the power of inferential canonical ordination techniques and other multivariate tests. This study focuses on how distance measure choice affects the ability of various permutation-based multivariate techniques to detect differences among ecological assemblages. In particular, the ability of the Euclidean, Chi-square and Bray-Curtis distance measures to detect the compositional difference between assemblages represented by species counts, is considered.

References: Faith, D.P., Minchin, P.R. and Belbin, L. (1987) Compositional dissimilarity as a robust measure of ecological distance. Vegetatio 69, 57-68.

Kenkel, N.C. and Orloci, L. (1986) Applying metric and nonmetric multidimensional scaling to ecological studies: Some new results. Ecology 67, 919-928.

Date received: April 2, 2002