13a: Distance-Based Redundancy Analysis (db-RDA)

Task H

Author

Affiliation

A. J. Smit

University of the Western Cape

Practice Task

Work through these exercises after reading the Distance-Based Redundancy Analysis chapter. It rehearses the workflow you will need for the integrative project, here on the familiar Doubs and mite data. Four exercises are hands-on calculations and two are short conceptual questions.

1. Using the Doubs fish data as the response and the Doubs environmental data as predictors, run a db-RDA with capscale() on a Bray-Curtis dissimilarity, constraining composition by a sensible subset of the environmental variables. Report the permutation test of the whole model (anova(...)), the adjusted \(R^2\) (RsquareAdj()), and the proportion of total inertia that is constrained.

2. Check multicollinearity among the predictors with vif.cca(). Remove predictors with a VIF above 10 one at a time, refitting after each removal, and report the final predictor set.

3. Test the significance of the individual constrained axes (by = "axis") and of the individual terms (by = "terms"), and produce and interpret a triplot (sites, species, environmental arrows), relating the constrained axes to the upstream-downstream gradient.

4. Apply the same workflow to the oribatid mite data (data(mite); data(mite.env)); briefly discuss what constrains the mite community.

5. Contrast what the db-RDA told you with what an unconstrained PCoA or nMDS of the same fish data would have told you. What does the constraint add, and what does it cost?

6. Why use db-RDA (a constrained ordination on a chosen dissimilarity) rather than classical RDA or CCA for these community data? Relate your answer to the choice of dissimilarity and to the assumptions of RDA and CCA.

Assessment Criteria

This Task is not formally assessed. It is built around four hands-on analyses (Exercises 1–4) and two short conceptual questions (Exercises 5–6); work through all six and bring your annotated Quarto document to class for discussion.