European Journal of Forest Research Plant Science

Comparison of theoretical and empirical models for branch size prediction of Chinese fir

Lyu, Yanjuan, Wu, Xiangrong, Xiahou, Chengyu, Chen, Wanrong, Ma, Xiangqing, Sun, Shuaichao
Springer Nature · 2026-11-03 · DOI 10.1007/s10342-026-01965-8

Abstract

Branch size prediction supports crown assessment and timber-quality management in Chinese fir ( Cunninghamia lanceolata ) plantations. We compared seven theoretical equation families and an empirical power function under matched calibration conditions using 4,997 branch-diameter and 4,825 branch-length measurements from 121 trees in 46 plots at five forest stations in Fujian Province. Tree age and diameter at breast height ( D ) were evaluated as covariates. Predictive performance was evaluated using five-fold cross-validation grouped by plot. Calibration used one to eight branches selected randomly or stratified by crown position. Calibration improved prediction for both model classes. With D and stratified sampling, theoretical and empirical models achieved root mean square errors (RMSEs) of 4.169 and 4.249 mm for branch diameter using three calibration branches, and 29.005 and 29.416 cm for branch length using five branches. Differences between model classes were small, with conditional 95% intervals for RMSE differences including zero. Theoretical models using D ranked first overall for both responses, although calibrated differences between D and age were small. Gompertz and Mitscherlich were selected as application models for branch diameter and length, respectively, with age-based alternatives where establishment records are reliable. Three calibration branches for diameter and five for length, sampled across crown positions, provided a practical balance between measurement effort and prediction accuracy for accessible crowns.

Read at publisher →