Partial least squares regression ordination plot, showing the distribution of quadrat samples on the first two axes identified by PLS. PLS attempts to find a new set of components which maximise the covariance between the matrix of soil water chemistry data and a multivariate response matrix of plant species cover data. Also shown, arbitrarily rescaled to fit the axes, are the loadings for the plant species (right-hand panel) and the chemical ions (shown in the three other panels). Quadrats which lie close together on the two PLS axes are similar in terms of species composition and soil water chemistry, as indicated by the loadings for the plant species and the chemical ions.