Litter fall, litter stocks and decomposition rates in rainforest and agroforestry sites in central Amazonia

The sustainability of agroforestry systems in Amazonia was assessed from their litter dynamics and decomposition. Litter fall and litter stocks were determined from July 1997 to March 1999 in four sites in central Amazonia: a primary rainforest, a 13- year- old secondary forest, and two sites of a polyculture forestry system which consisted of four planted tree species of commercial use amidst upcoming secondary growth. The average annual litter fall in the undisturbed primary rainforest (FLO) was 8.4 t ha(-1) year(-1), which is within the range of litter fall in other rainforests in the region. It was similar in one of the two polyculture sites ( 8.3 t ha(-1) year(-1) ), but lower in the secondary forest and in the second polyculture site. In the litter fall in secondary forest and agroforestry sites, the leaf portion was higher ( 76 - 82% of total litter fall ) than in FLO, due to reduced fine matter and wood fall. Leaf litter fall variability was much lower in the plantation sites than in the forests, which is explained by the much more homogeneous stand structure of the plantations. The quality of the produced litter, measured as C/ N ratio, differed significantly between the primary forest site and one polyculture and the secondary forest site. The cumulative input of nitrogen through litter fall was 144 kg ha(-1) year(-1) in FLO, and 91 - 112 kg ha(-1) year(-1) in the polycultures and the secondary forest. Litter fall was not correlated with soil parameters, but had a significant linear regression with canopy closure. For the primary rainforest, litter fall was also (inversely ) correlated with monthly rainfall. Litter fall was higher in the first year (1997 - 1998; an El Nino period) than in 1998 - 1999. Litter stocks on the forest floor were highest in the secondary forest (24.7 t ha(-1) ), and much lower in the polyculture sites (15.1 - 16.2 t ha(-1) ) and the primary forest (12.0 t ha(-1) ). There were no differences in the relative N content (C/ N ratio) of the litter stocks between the sites, but the larger stocks led to higher absolute N contents in the litter layer in the secondary forest. From the monthly values of litter stocks ( S ) and litter fall ( P ), the decomposition coefficient k(e) = P/ S was calculated, which was, on average, highest for the primary forest ( 0.059 ), followed by the polyculture systems ( 0.040 - 0.042 ), and by the secondary forest ( 0.024 ). Thus, due to low decomposition rates, the secondary forest site showed large litter accumulations in spite of a relatively low litter fall. In contrast, the primary forest showed high litter fall but low stocks, due to high decomposition rates. The decomposition coefficients of the polyculture systems ranged between the primary and the secondary forest. The reduced decomposition rates in the man- managed agroecosystems indicate quantitative and/ or qualitative changes in the decomposer communities of these systems that lead to a higher build- up of litter stocks on the forest floor. However, the decomposer systems in the polyculture sites still were more functional than in the site of non- managed secondary growth. Thus, from a soil biological viewpoint, ecologically sustainable low- input agroforestry in Amazonia will benefit from the application of these polyculture systems.