The PCA technique is used to transform the original variables to
a new orthogonal set of Principal Components (PCs) such that the
first PC contains most of the data variance and the second PC contains
the second largest variance and so on. Objects that exhibit
similar variances to the analysed variables have similar PCA scores
forming a cluster when plotted on a biplot. Additionally, strongly
correlated variables have the same magnitude and orientation
when plotted, whereas uncorrelated variables are orthogonal to
each other. Detailed descriptions of PCA can be found elsewhere
(Massart et al., 1988). Data pre-treatment was carried out to reduce
‘noise’ which interferes with the data analysis (Kokot et al.,
1998). Accordingly, the data matrix was column centred and
standardised (auto scaled). This was done by subtracting the column
mean from each cell value and dividing by the column standard
deviation. This ensured equal significance of variables with
a standard deviation of 1 (Kokot et al., 1998).