Aims Changes in the biomass of tropical forests play an important role in the global carbon cycle, but the biomass of these forests has been poorly quantified. A strategy for regional biomass estimation should supplement previous surveys with new data. Accurate data are necessary for reducing the uncertainty in the carbon budget of tropical regions.

Methods Biomass and its allocation were estimated for the tropical seasonal rain forest in Xishuangbanna, southwest China. Regression models relating tree biomass to DBH (diameter at breast height, 1.3 m) were developed, and a power-law allometric relationship W=aD^b was used to estimate the tree biomass, where W is the biomass of a tree (kg of leaves, branches, stems or roots), a and b are constants and D is the DBH (cm). Other biomass components were sampled in different quadrats in three 1 hm² permanent research plots: shrubs (ten 25 m² quadrats), herbs (ten 4 m² quadrats), dead wood (the whole plot), large fallen branches (twenty-five 25 m² quadrats) and litterfall (twenty-five 1 m² quadrats). This method was used for estimating above- and below-ground biomass of live and dead plants (trees, seedlings, shrubs, herbs, woody lianas, epiphytes, coarse woody debris and litterfall).

Important findings Total biomass for the three plots was 370.163, 550.119 and 351.442 Mg·hm^-2, with an average of (423.908±109.702) Mg·hm^-2 (95% confidence interval). Living biomass made up 95.28% of the total biomass, with coarse woody debris and litterfall comprising the rest. Most living biomass (98.09%±0.60%) (Mean±SD, n=3) of the seasonal rain forest was concentrated in the tree layer. In the allocation of total biomass, stems accounted 68.33% and roots, branches and leaves made up 18.91%, 11.07% and 1.65%, respectively. The biomass allocation among different DBH classes was concentrated in the middle and largest classes, with large trees (D>70 cm) accounting for 43.67% ± 12.67%. The most important ten species, in terms of biomass, made up 63.43% of the tree layer. Leaf area index (LAI) of the tree layer for the three plots was 5.73, 7.35 and 6.08, with an average of 6.39. Estimated aboveground biomass in our study sites fell within the range of published values for tropical moist forests and was lower than that of Malaysian and Cameroon rain forests, but higher than some neotropical rain forests. In terms of total biomass, Xishuangbanna tropical seasonal rain forest is also higher than moist forest in Brazil.