Growth, yields and mineral nutrition of cupuacu (Theobroma grandiflorum) in two multi-strata agroforestry systems on a ferralitic Amazonian upland soil at four fertilization levels

Cupuacu (Theobroma grandiflorum) is an economically important Amazonian tre e crop and a common component of agroforestry systems on the infertile soil s of the Amazon region. To provide farmers with recommendations on nutrient management for this species, information on the relationships between fert ilizer input, nutrient availability in the soil, foliar nutrient levels, gr owth and yield of the trees under the local pedoclimatic conditions are req uired. Cupuacu was grown for seven years in two multi-strata agroforestry s ystems with four fertilization levels on a xanthic Ferralsol in central Ama zonia. Yields were measured during the last four years, and the biomass of the trees was estimated from an allometric regression in the last year. Gro wth and yield data were compared with soil and leaf analyses from both syst ems. In one system, leaf samples of three age classes and four collection d ates during a year were analysed to determine a sampling protocol with maxi mum sensitivity for differences in mineral nutrition for the species. The c upuacu yields were significantly influenced by fertilization level and crop ping system. Soil and leaf analyses suggested that the yield response was m ainly to P fertilizer, whereas N fertilizer did not seem to affect growth a nd yields of the plants. The soil data were easier to relate to the cupuacu yields than the leaf data, apparently because the trees responded to great er nutrient inputs with increased vegetative growth rather than with increa sed nutrient concentrations in the leaves. However, foliar analyses gave us eful additional information on limiting nutrients and suggested that the tr ees in an unlimed treatment developed a latent Mg deficiency which may affe ct crop yields in the future. The pronounced difference in growth and yield of cupuacu between the two multi-strata systems could not be explained wit h foliar analyses, and with soil analyses only by taking into account the f ertility of the soil under previous intercrops planted between the trees. S ampling recommendations for foliar analysis of cupuacu are given. For soil analysis in agroforestry systems with spatially heterogeneous nutrient avai lability in the soil, sampling of a sufficiently large area around the targ et trees is essential.