Cross-scale variation in the density and spatial distribution of an Amazonian non-timber forest resource

Successful management of tropical forest resources depends upon an understanding of their patterns of density and spatial distribution, since these affect the potential for harvesting. The variation in these patterns across different spatial scales has rarely been explored. We assessed the extent to which different spatial scales are useful in understanding resource distribution, using the example of an economically-significant tropical tree genus, Copaifera, which is valued across Brazilian Amazonia for its medicinal oleoresin. We mapped the spatial distribution of Copaifera trees at three nested spatial scales: basin-wide (across Brazilian Amazonia), landscape (across two contiguous extractive reserves) and local (within a 100-ha plot). Using data from our own study and an Amazon-wide forest inventory (Projeto RADAMBRASIL), we quantified the population distribution, density and size structure at the genus and species level at all three scales, relating these to two environmental variables - forest type and elevation. Spatial statistics were used to further characterize the resource at the landscape and local levels. The distribution, density and adult population structure differed between species and forest types at all three spatial scales. Overall tree densities ranged from 0.37ha ^-1 (basin-wide scale) to 1.13ha ^-1 (local scale) but varied between forest types, with várzea containing a Copaifera tree density just 43% of that in terra firme forest at the landscape scale. Spatial distribution analyses showed significant clumping of some species, especially C. multijuga which averaged 61m between neighbouring trees. We compare our cross-scale density estimates and discuss the relative merits of studying the distribution of non-timber forest products (NTFP) at more than one spatial scale. Our results have implications for the management and extraction of this important Amazonian forest resource.