Cross-Comparison of Modeling Methods for Ancient Tree Age Prediction: A Case Study on Six Species in Huangshan City, China

Ruijun Wang^1*Xukun Han¹Peichu Liu¹Xiaohan Zhang¹Jinzi Zhang¹Qinghe Hou²Lyu Xiaoqian¹

• ¹Hefei University of Technology, Hefei, China
• ²China University of Mining and Technology, Xuzhou, China

Ancient trees represent vital natural and cultural assets for a nation or region, embodying values across ecological, historical, and landscape dimensions. Accurate determination of age is a cornerstone of effective ancient tree conservation and management. This study focuses on Huangshan City, China, investigating six regionally predominant species: T. grandis, T. mairei, C. sclerophylla, C. officinarum, L. formosana, and A. aspera. We established a cross-comparison framework encompassing these six species and four modeling methods (MLR, GWR, RF, GWRF) to conduct an in-depth analysis of model performance as influenced by method choice and predictor composition. The findings reveal: (1) GWR effectively addresses the spatial heterogeneity inherent in ancient tree distributions, while RF excels at capturing complex nonlinear relationships. The GWRF model, which integrates both approaches, achieved the highest prediction accuracy. (2) Model performance is closely linked to species-specific ecological strategies. Growth in long-lived species (e.g., T. grandis and T. mairei) is manifested more through the accumulation of morphological traits, whereas species with a younger population age structure (e.g., L. formosana and A. aspera) are more constrained by environmental factors; (3) DBH was consistently the key morphological factor across all species, while Altitude and MAP were the most common key environmental factors. The identification of these key factors and their interspecific differences can provide precise guidance for the census, conservation, and management of ancient trees. This study not only provides an optimized solution for predicting ancient tree age but also underscores a deeper principle: scientific conservation must begin with understanding their unique growth logic, thereby establishing a solid theoretical and practical framework for precision management.