5.3.2.2 PCA of a Correlation Matrix
# A reminder of the content of the env dataset
summary(env)
# PCA based on a correlation matrix
# Argument scale=TRUE calls for a standardization of the variables
env.pca <- rda(env, scale = TRUE)
env.pca
summary(env.pca) # Default scaling 2
summary(env.pca, scaling = 1)
Hint If you don’t want to see the site and species scores, add argument axes = 0 to
the summary() call.
Note that the scaling (see below) is called at the step of the summary (or, below,
for the drawing of biplots) and not for the analysis itself.
The “summary” output looks as follows for scaling 2 (scalings are explained below);
some results have been deleted:
Call:
rda(X = env, scale = TRUE)
Partitioning of correlations:
Inertia Proportion
Total
11
1
Unconstrained
11
1
Eigenvalues, and their contribution to the correlations
Importance of components:
PC1
PC2
PC3
PC4
PC5 …
Eigenvalue
6.0979 2.1672 1.03761 0.70353 0.35174 …
Proportion Explained 0.5544 0.1970 0.09433 0.06396 0.03198 …
Cumulative Proportion 0.5544 0.7514 0.84571 0.90967 0.94164 …
5.3 Principal Component Analysis (PCA)
155