Information on net primary production in tropical forests is needed for the
development of realistic global carbon budgets, for projecting how these e
cosystems will be affected by climatic and atmospheric changes, and for eva
luating eddy covariance measurements of tropical forest carbon flux. Howeve
r, a review of the database commonly used to address these issues shows tha
t it has serious flaws. In this paper we synthesize the data in the primary
literature on NPP in old-growth tropical forests to produce a consistent d
ata set on NPP for these forests. Studies in this biome have addressed only
a few NPP components, all aboveground. Given the limited scope of the dire
ct field measurements, we sought relationships in the existing data that al
low estimation of unmeasured aspects of production from those that are more
easily assessed. We found a predictive relationship between annual litterf
all and aboveground biomass increment. For 39 diverse tropical forest sites
, we then developed consistent, documented estimates of the upper and lower
bounds around total NPP to serve as benchmarks for calibrating and validat
ing biogeochemical models with respect to this biome. We developed these es
timates based on existing field measurements, current understanding of abov
eground consumption and biogenic volatile organic carbon emissions, and our
judgment that belowground production is bounded by the range 0.2-1.2 x ANP
P (aboveground NPP). Across this broad spectrum of tropical forests (dry to
wet, lowland to montane, nutrient-rich to nutrient-poor soils), our estima
tes of lower and upper bounds on total NPP range from 1.7 to 11.8 Mg C.ha(-
1).yr(-1) (lower bounds) and from 3.1 to 21.7 Mg C.ha(-1).yr(-1) (upper bou
nds). We also showed that two relationships that have been used for estimat
ing NPP (the Bray-Gorham relationship based on leaf litterfall and the Miam
i model based on temperature or precipitation) are not valid for the tropic
al forest biome.