2.7. Statistical analysis
One-way ANOVA was applied to test the effect of land use change
and age of sugar cane plantation on litter C and N contents, as well as
on soil bulk density, percentage clay, pH-values, bulk SOC and total
soil N contents (nine sugarcane and one forest plots) for 0–10,
10–20 and 0–20 cm depths and MWD of water stable aggregates
(four sugar cane and one forest plots) for 0–10 and
10–20 cm depths. For a given soil depth, differences in: (a) aggregate
weight(inpercentage)amongaggregate sizeclasses and plantations,
(b) aggregate C concentration (in g C kg1 sand free aggregate) and
C/N ratios among aggregate size classes and plantations, (c) SOM C
concentration (in g C kg1 soil) and C/N ratios among SOM fractions
(i.e., total fPOM, total iPOMc, total iPOMf and total mineral-
associated SOM) and plantations, and (d) iPOM C concentration
(in g C kg1 sand-free aggregate) and C/N ratios among iPOM
fractions (i.e., iPOMc and iPOMf in large macroaggregates, small
macroaggregates and microaggregates) and plantations were
analyzed with split plot analysis, with plantation age as main plot
and aggregate size, SOM or iPOM fraction as subplot factors. We did
not include soil depth as a factor in the statistical analysis but
analyzed land use effects on soil variables for the two soil depths
separately and for the 0–20 cm soil cores. For each plot (forest and
sugarcane),we had six replicate soil samples foreach soil depth, each
corresponding to a composite soil sample from a single transect.
Post-hoc multiple mean comparisons were analyzed with Tukey’s
HSD test at P < 0.05. Prior to statistical analyses data of C, and C/N
ratios of all fractions were log transformed to homogenize
variances. All statistical analyses were run using STATISTICA 7.1
(StatSoft Inc., Tulsa, OK, USA).