Process-based prediction of Chinese fir growth in multi-layered uneven-aged mixed forests using iLand

Accurate prediction of forest growth is key to optimizing silviculture and sustainable management. Traditional empirical models struggle to capture long-term and structurally complex growth dynamics. This study systematically compares the predictive performance of the process-based model iLand with parameterized empirical growth models for Chinese fir ( Cunninghamia lanceolata (Lamb.) Hook) in multi-layered uneven-aged Chinese fir– Phoebe bournei (Hemsl.) Yen C. Yang forests, using data from 30 plots in China. After the models were calibrated with tree-core data, growth projections were conducted for a 7-year validation period (2015–2022) and a 30-year scenario simulation. The results show that iLand (R² = 0.950, RMSE = 0.73 cm) significantly outperforms the best empirical model in predicting diameter at breast height (DBH) and stem volume, and more accurately reproduces timber structure (χ² test P = 0.091). Sensitivity analysis reveals that foliage biomass allometric exponent ( bmFoliage_b ) and specific leaf area ( specificLeafArea ) are the most sensitive parameters in iLand. Significant differences ( P < 0.01) in DBH and volume predictions occur when bmFoliage_b deviates by more than 0.01 or when specificLeafArea deviates by more than ±0.2; changes in tree height beyond ±10%–15% significantly affect volume ( P < 0.05). The model demonstrates good stability in response to uncertainties in height input. This study confirms that iLand, with its solid physiological foundation, effectively captures the growth response of Chinese fir plantations to management practices, offering reliable decision support for developing refined management strategies for multi-layered uneven-aged forests.