资源环境科技发展态势分析平台

An individual-tree-based growth-and-yield model (SILFOR) was developed to evaluate the long-term effects of silvicultural treatments beyond logging in a tropical forest is described. Here we focus on the model's approach to prediction of the magnitude and duration of silvicultural treatment effects on growth of the residual trees that will contribute most of the timber available to the next planned harvest in 25-40 years. Data from 30 1-ha permanent sample plots monitored for up to 20 years in dipterocarp forest in Kalimantan, Indonesia were used to develop the model. The treatments were different logging intensities with or without post-logging silvicultural treatments. Each species was assigned to one of five groups based on their ecological traits and the merchantability of their timber. A mixed-effects model was used to account for the spatially and temporally autocorrelated permanent plot data.

The mixed-effects approach improved model performance substantially compared to a fixed-effects approach; specification of the variance function and correlation structure of the error term further improved model fit. Patterns and rates of tree diameter increment varied substantially, as indicated by the large differences among species groups in terms of the level of random-effects, number of parameters assigned as mixed-effects, and the covariates that define the best model. Similarly, silvicultural treatment effects also varied among species groups, as indicated by differences in the treatment dummy variables in the final model. Among-treatment differences diminished over time. We also found that species group representation is effected differentially by the silvicultural treatments. Overall, the study presents a novel and hopefully useful approach to the analysis of growth-and-yield data from tropical forests under intensified management for timber. (C) 2017 Elsevier B.V. All rights reserved.