Generalised Canonical Correlations Analysis for Explaining Macrofungal Species Assemblages

Canonical correlations analysis uses a secondary set of multivariate observations to correlate with, and thereby help to explain, differences among a primary set of multivariate observations. In ecology, the variables of the primary data set may be of the “presence/absence” type, or they may be the abundances of individual species in a species list, whereas the secondary set often involves environmental variables. The nonGaussian distribution of the primary data set precludes the use of traditional multivariate statistical methods, but recent developments in the application of canonical analysis enable similarities between the primary variates to be calculated using measures of similarity in common use in ecology. In this paper, we illustrate how one of these generalisations, the canonical analysis of principal coordinates (CAP), may explain differences in species assemblages of macrofungal survey data obtained from experimental units of differing floristic composition, or how lists of species obtained during different seasons may reflect weather variables. The canonical correlation analyses incorporate non-parametric permutation tests that give exact probability values, as well as ordination diagrams that display the differences in the experimental units that mirror the differences in the mycota. D. A. Ratkowsky and G. M. Gates (2008). Generalised canonical correlations analysis for explaining macrofungal species assemblages. Australasian Mycologist 27 (1): 33-40.