AUTHOR=Xu Anyang , Wang Dongzhi , Liu Qiang , Zhang Dongyan , Zhang Zhidong , Huang Xuanrui 

TITLE=Incorporating stand density effects and regression techniques for stem taper modeling of a Larix principis-rupprechtii plantation

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.902325

DOI=10.3389/fpls.2022.902325

ISSN=1664-462X

ABSTRACT=Stand density is the main factor affecting tree stem form. Based on the results of variable-exponent models exploring the effect of different features in the stem taper model, we found that stand density plays an important role in economic and ecological function evaluation of stem volume and stem form quality. The dataset included 396 analytical trees from 132 standard plots of larch (Larix principis-rupprechtii) plantation in Saihanba, Hebei Province. Based on 12 different forms of variable-exponent models, we explored the optimal basic equation for plantation and the effects of stand density, basal area, canopy density and different forms of stand density on the prediction accuracy of the variable-exponent models; The variable-exponent taper equation including stand density was constructed by nonlinear regression, nonlinear mixed effect model and nonlinear quantile regression method. The results indicated that Kozak (2004) variable-exponent taper equation was the best basic model to describe the change of stems form of larch plantation, and the density factor in the form of √sd improved the prediction accuracy of the basic model; Based on nonlinear regression, nonlinear mixed effect model and nonlinear quantile regression, the variable-exponent taper equation including stand density was constructed. In terms of the prediction accuracy of the models, the order we found was: nonlinear quantile regression > nonlinear mixed effect model> nonlinear regression. When the quantile was 0.5, the nonlinear quantile regression model exhibited the best performance, based on the following metrics: coefficient of determination (R2) =0.9766, the root mean square error (RMSE) =1.0367, mean absolute bias (MAB) =0.7636, and the mean percentage of bias (MPB) =4.9282. The variable-exponent taper model of larch plantation based on nonlinear quantile regression fit well with stem growth and development, and provides a scientific basis for the rational management of larch plantations.