Unconstrained ordination methods can be used for hypothesis testing: issues in slides and in the book

[pedrohbraga]

In certain slides (e.g, slide 11) and part of the text, there is a discussion implying that we cannot test hypotheses using unconstrained ordination methods.

Unconstrained methods (such as PCA) can be used for multiple types of hypothesis testing. It may require additional steps other than just testing directly in the initial model, such as what could be done in a simple RDA, but the Eigenvectors and Eigenvalues of unconstrained methods can be used for hypothesis testing.

The intention perhaps was to refer to our ability to directly test hypotheses of the effects of a group of variables in another group of variables in a direct analysis.