Abstract: In this study,the data consisted of 702 individuals of Pinus sylvestris var.mongolica in 22 temporary sample plots of pure plantation in Zhanggutai of Liaoning Province,northeast China,were used to develop the canopy-DBH models,i.e.the basic model (fitted by ordinary least squares,OLS),generalized model (fitted by OLS),nonlinear mixed-effect basic model and nonlinear mixed-effect generalized model.The goodness-of-fits and prediction accuracy of the four models were compared.For the mixed-effect model,four sampling strategies,i.e.,random sampling,large-DBH tree sampling,small-DBH tree sampling and medium-DBH tree sampling,were designed to calculate the random parameters.The effects of individual factor and stand level variables on the canopy-DBH relationship were simulated.Model evaluation indices included the determination coefficient (R²),mean absolute error (MAE) and root mean square error (RMSE).Results showed that the height to live crown base (HCB),relative spacing index (RS) and stand age (A) were the dominant factors in CW-DBH models.The goodness-of-fits of mixed-effect CW-DBH models (R²,MAE and RMSE) were 0.703 0,0.386 6 and 0.515 4,and those of mixed-effect generalized model were 0.705 1,0.382 2 and 0.513 6,respectively,which were better than those of the OLS models.However,the difference in goodness-of-fit between the mixed-effect basic and generalized models was not significant (below 1%).Canopy decreased with the increase of A and height of HCB,but increased with the increase of RS.

Key words: Pinus sylvestris var.mongolica, sand-fixing plantation, canopy-DBH, mixed-effect basic model, generalized model