Thi Ha Nguyen, Ram Subedi Mukti, Thi Thanh Thuy Phan, Van Viet Nguyen
Modeling the dominant-height growth and site index for Rhizophora apiculata plantations in Ca Mau province, Vietnam
Dendrobiology 2025, vol. 94: 176-186
https://doi.org/10.12657/denbio.094.013
Abstract:
Site classification is crucial aspect of forest management, particularly for assessing the productivity potential of forest stands. Forest productivity potential is often expressed as the dominant height of a stand at a given index age, commonly referred to as the site index (SI). Site index information is critical for forest management and silvicultural decision-making, including determining optimal thinning schedules and clear-cutting timings, as well as predicting stand development patterns. This study aimed to develop an SI for Rhizophora apiculata plantations in Ca Mau province, Vietnam. Fitting data were collected from 97 temporary sample plots (N = 1237), with ages ranging from 5 to 24 years. Using a 10-fold cross-validation method, six well-established growth models were applied to the dominant height-age data. Fifteen trees were destructively harvested for stem analysis and were used for model validation. The models were evaluated using six statistical metrics: a: adjusted R2 (R2adj.), mean bias (MB), the root mean square error (RMSE), mean absolute bias (MAB), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). The results revealed that the Gompertz equation outcompeted other candidate models resulting in the highest adjusted R2 (R2adj. = 0.946) and least mean absolute bias (MAB = 1.125) when used in stem analysis data. Therefore, Gompertz equation with asymptote or shape parameter expansion is recommended for generating family of site index curves.
Keywords: Index age, 10-fold cross validation, guide curve, destructive sampling, Rhizophora apiculata