Abstract: Pleioblastus amarus is widely distributed in the Nanling Mountain of Guangdong, and has great potential in carbon sink. Constructing aboveground biomass allometry model of P. amarus is important for the calculation of forest carbon stock and the assessment of carbon sink function. In this study, 45 individuals of P. amarus were randomly sampled, and the biomass of each organ was determined to construct allometry models for different organs and total aboveground biomass (AGB). Diameter at breast height (DBH) was involved in the models as the main predictor and tree height (H) as an additional predictor. The accuracies of different models with or without H were compared. The results showed that 76.1%±0.8% of AGB was allocated to culm, 14.5%±0.5% to twig and 9.4%±0.4% was allocated to leaf. Culm biomass was positively related to DBH (P<0.05), while the biomass of twigs and leaves were negatively related to DBH. The univariate model involving DBH and the bivariate model involving DBH and H had very high accuracies for predicting both culm biomass and AGB, with R² exceeding 0.95, and the accuracies for predicting twig biomass and leaf biomass were also high, with R² being 0.888 and 0.684, respectively. Adding H as an additional predictor into the model improved the accuracy of the model prediction to a small extent but resulted in the problem of multicollinearity. There were small differences between AGB estimates using a single model and the sum of different organs. The results suggest that P. amarus tends to allocate biomass to twigs and leaves in the early stage to ensure rapid photosynthesis and lately to culms for stabilizing its status in community. Univariate models with DBH as a single variable are recommended in the estimation of P. amarus AGB in order to reduce workload while having a high accuracy in calculation.

Key words: Pleioblastus amarus, biomass, distribution characteristics, regression model