Background, assumptions, and interpretation of Principal Component Analysis ordination (PCA). Choice of covariance vs. correlation association matrix.