Above-ground dry biomass of living trees including palms was estimated in 6
5 l ha plots spanning a 1000 km(2) landscape in central Amazonia. The study
area was located on heavily weathered, nutrient-poor soils that are widesp
read in the Amazon region. Biomass values were derived by measuring the dia
meter-at-breast-height (DBH) of all greater than or equal to 10 cm trees in
each plot, then using an allometric equation and correction factor for sma
ll trees to estimate total tree biomass. Detailed information on soil textu
re, organic carbon, available water capacity, pH, macro- and micro-nutrient
s, and trace elements was collected from soil surface samples (0-20 cm) in
each plot, while slope was measured with a clinometer. Biomass estimates va
ried more than two-fold, from 231 to 492 metric tons ha(-1), with a mean of
356 +/- 47 tons ha(-1). Simple correlations with stringent (p < 0.006) Bon
ferroni corrections suggested that biomass was positively associated with t
otal N, total exchangeable bases, K+, Mg2+, clay, and organic C in soils, a
nd negatively associated with Zn+, aluminum saturation, and sand. An ordina
tion analysis revealed one major and several minor soil gradients in the st
udy area, with the main gradient discriminating sites with varying proporti
ons of clay (with clayey soils having higher concentrations of total N, org
anic C, most cations, and lower aluminum saturation and less sand). A multi
ple regression analysis revealed that the major clay-nutrient gradient was
the only significant predictor, with the model explaining 32.3% of the tota
l variation in biomass. Results of the analysis suggest that soil-fertility
parameters can account for a third or more of the variation in above-groun
d biomass in Amazonian terra-firme forests. We suggest that, because the co
nversion of forest to pasture tends to reduce the nitrogen, clay, organic c
arbon, and nutrient contents of soils, forests that regenerate on formerly
cleared lands may have lower biomass than the original forest, especially i
n areas with low soil fertility. (C) 1999 Elsevier Science B.V. All rights
reserved.