Long-term soil organic carbon dynamics in a subhumid tropical climate:: 13C data in mixed C3/C4 cropping and modeling with ROTHC

Scanty information on long-term soil organic carbon (SOC) dynamics hampers validation of SOC models in the tropics. We observed SOC content changes in a 16-year continuously cropped agroforestry experiment in Ibadan, south-western Nigeria. SOC levels declined in all treatments. The decline was most pronounced in the no-tree control treatments with continuous maize and cowpea cropping, where SOC levels dropped from the initial 15.4 to 7.3-8.0 Mg C ha(-1) in the 0-12 cm topsoil in 16 years. In the two continuously cropped alley cropping (AC) systems, one with Leucaena leucocephala and one with Senna siamea trees, SOC levels dropped to 10.7-13.2 Mg C ha(-1). Compared to the no-tree control treatments. an annual application of an additional 8.5 Mg ha(-1) (dry matter) of plant residues, mainly tree prunings, led to an extra 3.5 Mg C ha(-1) (similar to0.2% C) in the 0-12 cm top soil after I I years, and 4.1 Mg C ha(-1) after 16 years. The addition of NPK fertilizer had little effect on the quantities of above-ground plant residues returned to the soil, and there was no evidence that the fertilizer affected the rate of SOC decomposition. The fact that both C-3 and C-4 plants returned organic matter to the soil in all cropping systems, but in contrasting proportions, led to clear contrasts in the C-13 abundance in the SOC. This C-13 information, together with the measured SOC contents, was used to test the ROTHC model. Decomposition was very fast, illustrated by the fact that we had to double all decomposition rate constants in the model in order to simulate the measured contrasts in SOC contents and delta(13)C between the AC treatments and the no-tree controls. We hypothesized (1) that the pruning materials from the legume trees and/or the extra rhizodeposition from the tree roots in the AC treatments accelerated the decomposition of the SOC present at the start of the experiment (true C-priming), and/or (2) that the physical protection of microbial biomass and metabolites by the clay fraction on this site, having a sandy top soil in which clay minerals are mainly of the 1: 1 type, is lower than assumed by the model. (C) 2004 Elsevier Ltd. All rights reserved.