Incorporating climate effects in Larix gmelinii improves stem taper models in the Greater Khingan Mountains of Inner Mongolia, northeast China

Abstract

Estimating timber volume and carbon stock in forests is fundamental for silviculture and for accurate estimation of national and global carbon budgets. Taper models are important tools for predicting diameter at any height along a tree bole. Mean annual temperature (MAT) and mean annual precipitation (MAP) influence tree growth, but their precise effects on stem shape are still poorly understood and climatic factors are seldom included in taper models. To evaluate the effect of climate on tree stems, we incorporated MAT and MAP as covariates in the Kozak (2004) model to improve model performance in goodness-of-fit. The Kozak (2004) model with the incorporation of MAT and MAP was refitted using nonlinear mixed-effects (NLME) modeling techniques to account for within-sample tree heteroscedasticity and autocorrelation structure in residuals from data measured at different points along the same individual tree stem of Larix gmelinii (Rupr.). Results showed that the predictive accuracy of the Kozak (2004) model was improved by incorporating MAT and MAP as covariates. The Kozak (2004) model incorporating both MAT and MAP had the highest prediction accuracy for stem diameter, closely followed by the model incorporating only MAT and then the model incorporating only MAP. MAT effect on tree stem shape was stronger than that of MAP. The NLME Kozak (2004) model incorporating MAT and MAP with exponential variance function and first-order continuous autoregressive correlation structure (CAR(1) model) removed the heteroscedasticity and autocorrelation in the residuals, had the best prediction performance. Therefore, such refined model is recommended for planning and management of natural L. gmelinii forests. In conclusion, incorporating the effect of climate variables in stem taper equations could significantly improve timber volume and biomass estimations, particularly in harsh environments, such as natural boreal forests.