Abstract

Tree height and diameter at breast height (DBH) are key survey factors in forest inventory. Compared to measuring DBH, the process of measuring tree height is time-consuming, labor-intensive and costly, especially in natural mixed forests. Tree height is therefore commonly predicted using height-diameter (H-D) models. Nevertheless, developing H-D models for natural mixed forests are hampered by the large diversity of tree species and the small number of rare tree species observed. In this study, a nonlinear mixed effects H-D model was established for a mid-montane humid evergreen broad-leaved forest in southwest China. For this purpose, we classified tree species by K-means clustering according to the specific plant functional traits. Data were collected from 5, 737 trees in 100 subplots with 86 species. The results revealed that the nonlinear mixed effects model of tree height accounted for 75% of the variation in tree height without significant heteroscedasticity, and that the model parameters had biological significance. In the 10-fold cross validation, no clear fluctuation in the model evaluation metrics was observed, indicating the absence of overfitting in the model. Of the 15 stand variables, dominant height (Hd) and Margalef’s index of DBH (DMg), which contributed the most to the height-diameter model, were selected to expand the model. Our findings suggest that tree height increased with increasing values of Hd and DMg, and vice versa. This study highlights the feasibility of categorizing natural mixed forests by utilizing plant functional traits and the influential role of tree size inequality on tree height variation.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call