Tree damage, allometric relationships, and above-ground net primary production in central Amazon forest

The loss of tree mass over time from damage can lead to underestimates in above-ground net primary productivity (ANPP) if not accounted for properly. Bias in the allometric relationship between trunk base diameter (Db, at 1.3 m height or above the buttresses) and mass can also lead to systematic errors in ANPP estimates. We developed an unbiased model of the relationship between Db and tree mass using data from 315 trees (5 cm Db) harvested in the central Amazon. This model was compared with other theoretical (n=1) and empirical models (n=4). The theoretical model, and one empirical model, made predictions that differed substantially form our central Amazon model. The other three empirical models made predictions that were consistent with our model despite being developed in different tropical forests. Models differed mostly in predicting large tree mass. Using permanent forest inventory plot data, our Db versus tree mass model, and a bole volume model, we estimated that tree damage amounts to 0.9 Mg ha-1 per year (dry mass) of litter production. This damage should be included as a mass loss term when calculating ANPP. Incorporating fine litter data from published studies, we estimated that average ANPP for central Amazon plateau forests is at least 12.9 Mg ha-1 per year (or 6.5 Mg C ha-1 per year). Additional sources of error as described in the text can raise this estimate by as much as 4 Mg ha-1 per year. We hypothesize that tree damage in old-growth forests accounts for a significant portion of age related decline in productivity.