Potential biomass accumulation in Amazonian regrowth forests

Biomass accumulation in the secondary forests of abandoned pastures and sla sh-and-burn agricultural fallows is an important but poorly constrained com ponent of the regional carbon budget for the Brazilian Amazon. Using empiri cal relationships derived from a global analysis, we predicted potential ab oveground biomass accumulation (ABA) for the region's regrowth forests base d on soil texture and climate data. For regrowth forests on nonsandy soils, the globally derived relationship provided a nearly unbiased linear predic tor of Amazonian validation data consisting of 66 stands at seven sites; th ere was no significant difference between stands that regrew following use as pasture land and those that regrew following slash-and-burn agriculture. For regrowth forests on nonsandy soil, the 1 sigma error range of our ABA model was 58%-171% for the Amazonian validation data. For regrowth forests on sandy soils, the validation data were limited to 19 stands at one site, and the globally derived relationship was substantially biased multiplicati vely and nonlinearly. Hence we developed a regional refinement by adding to our validation data ABA values from the two Amazonian sites with sandy soi l that had previously been included in the global analysis. Based on a cons ervative jackknife goodness-of-fit assessment (leaving out one site at a ti me), we calculated a 1 sigma error range of 42%-158% for our sandy soil Ama zonian regrowth forest ABA model. We present our predictions of potential r egrowth forest ABA as a set of 0.5 degrees resolution maps for the region a t 5, 10, and 20 years following abandonment.