Soil aggregate stability and soil organic matter fractions under agropastoral systems established in native savanna

An area of native savanna on an Oxisol in the Eastern Plains of Colombia was opened and sown to various rotations of grass or grass-legume pasture with rice. After 4.5 years, the soil was sampled for studying the effect of land conversion on soil aggregation and on the distribution of total and particulate soil organic matter across the aggregate size classes. The size distribution of undisturbed aggregates did not vary among treatments. Five different methods were used to measure wet aggregate stability (WAS). The choice of method affected the WAS average across treatments as well as the differences among treatments. The only consistent observation was the lower WAS under monocropped rice compared with the other treatments. Inclusion of a legume in a pasture hardly affected aggregate stability. In contrast to the WAS measurements, which were carried out with soil aggregates of 1-2 mm, wet sieving of whole-soil samples revealed additional differences among treatments: large macroaggregates (>2 mm) proved less stable under those treatments that involved soil disturbance through ploughing and harvesting. Total soil C and N content did not vary among treatments, despite considerable differences in plant production levels. The C concentration, but not the N concentration, declined with decreasing aggregate size. The distribution of whole-soil C and N content across aggregate size classes depended more on the amount of soil in a certain size class than on the size class's C or N concentration. Those treatments that involved frequent soil disturbance had a smaller fraction of large macroaggregates (>2 mm) and, as a consequence, less C and N in the large macroaggregate fraction. The particulate organic matter (POM) fraction accounted for only 6.2-8.5% of total soil carbon. The small size of this pool makes it unlikely that POM can serve in these Oxisols for estimating the amount of soil organic matter with medium turnover rate, as suggested by others.