ObjectiveThe quality of wood determines its value in production, and high-quality wood tends to earn higher profits. However, the size of the knot on the trunk can seriously affect the quality of wood, and the knot is formed after the death of branch. Therefore, by studying the properties of branches, looking for suitable ways of controlling the forest is of great significance to improve the quality of wood.
MethodBased on the branch analysis data of 70 trees selected from 10 sample plots of Larix olgensis plantations in Mengjiagang Forest Farm of Jiamusi City, Linkou Forestry Bureau and Dongjingcheng Forestry Bureau, Heilongjiang Province of northeastern China, this paper develops classic Logistic model using traditional method and generalized linear mixed models (GLMM) to predict the branch survival of planted Larix olgensis tree. Goodness of fit tests and independence tests were implemented for all models.
ResultThe branch survival was affected by many factors, like the degree of natural pruning of trees, the growth position of branches and the competition among trees.In the model, CR reflected the degree of natural pruning of trees, and the parameter value positive indicated that the natural pruning degree of trees was low and the branches were mostly in the living state.BRH and WHOLE reflected the growth position of branches in trees, and those negative parameter values showed that the branches in the upper part of the canopy grew well due to the sufficient light, and the branches in the lower part of the canopy died early due to mutual shadowing.HD reflected the competition between trees, and the negative parameter value showed that the intense competitive environment could reduce the survival probability of branches.AICs, RMSEs, AUCs and the correctness of model judgment can be used to compare the prediction effect of basic model and GLMM. The calculated AIC = 801.67, RMSE = 0.126, AUC = 0.9975 and the correctness of model judgment was 97.9% of GLMM. GLMM had obvious smaller AIC, smaller RMSE, larger AUC and larger correct rate. Thus, GLMM could efficiently solve the problem of the variation among different individuals, and improve the accuracy of predicting the branch survival status. The accuracy of the model was good in the independence test.
ConclusionThis study would provide the theoretical basis for determining reasonable management measures and improving the timber quality for Larix olgensis plantation.