Data ordination through PCA resulted in a 4-component model which cumulatively accounted for
84.7% of the total system variance (TSV). The eigen vectors that relate the PCs to the original variables
along with the proportion of total system variance for each PC are given in Table III. The first component
(PC1) as the most informative presented highly weighed variables (with a loading absolute value of > 0.7)
on basic cations (Ca and Mg), P sorption, exchangeable Al, and NaOH-Pi. Moderate and positive loaded
variables included soil pH, clay, C/N ratio, and residual P. When PCA was run exclusively with the
control treatments that received no P fertilizer, a similar pattern in PC structure was observed (data
not shown), with however a higher weight on available P. The second component was dominated by
soil organic constituents (organic C and total N) and oxalate-extractable Fe as well as exchangeable
K and C/N ratio.