Canonical correlation analysis (CCA) was used to interpret the
relationships between soil microbiological and soil biochemical
variables (Nash and Chaloud, 2002). The CCA was intended to derive
a linear combination (canonical variate) of the two sets of variables
and to indicate which specific variable was responsible for
discriminating among the forest management practices. Therefore,
the variance of the original sets of variables was redistributed into a
few pairs of canonical variates tomaximize the correlation between
them. These canonical correlation coefficients were interpreted
as ordinary Pearson correlation coefficients and their significance
evaluated via a chi-square statistic