Abstract: Crown width is a key indicator reflecting tree vigor and stand spatial structure, and the accurate prediction of which is important for the structural regulation and scientific management of Larix principis-rupprechtii and Betula platyphylla mixed forests. We used 1316 individual trees of L. principis-rupprechtii and B. platyphylla from 29 plots (20 m × 30 m) in the Saihanba Mechanical Forest Farm, Hebei Province, as research subjects. First, we selected an optimal base model from 12 commonly used crown width-diameter at breast height models. Then, we introduced variables related to tree size, competition intensity, stand size structural heterogeneity, and soil physicochemical properties to develop a nonlinear mixed-effects crown width model by incorporating species dummy variables and plot-level random effects. The results showed that the composite, growth, and exponential functions exhibited the best fitting performance among the candidate base models. The model prediction accuracy was significantly improved after including species dummy variables and plot-level random effects. Compared with the base model, the coefficient of determination increased from 0.29 to 0.60, while the root mean square error and mean absolute error decreased by 24.1% and 27.1%, respectively, explaining approximately 60% of the variation in crown width. Covariate analysis indicated that crown width was significantly positively correlated with the Gini coefficient of basal area at breast height and soil bulk density, but negatively correlated with the competition index and the height-to-diameter ratio. These findings demonstrated that jointly considering biotic and abiotic factors in crown width modeling could effectively improve model predictive performance for mixed forests.

Key words: Larix principis-rupprechtii and Betula platyphylla mixed forest, crown width model, structural diversity, competitive intensity, mixed effects model