Effects Of Thinning On The Growth Of Planted Larix Olgensis And Dynamic Simulation

The purpose of this study was to quantify the effects of thinning on growth of planted Larix olgensis,optimize stand structure,develop a reasonable thinning program and manage forest scientifically.Based on the data of 5 permanent plots remeasured 35 times in Mengjiagang forest farm,5 permanent plots remeasured 18 times in Jiangshanjiao forest farms,the effects of thinning on diameter distribution,mortality of plot and individual tree,individual tree diameter growth were analyzed.The effects of thinning on gross growth,net growth,basal area and volume growth of living tree per hectare and average DBH were analyzed.Based on the results of analysis for thinning effects,a series of models were established including parameter prediction model of tree diameters,mortality models of plots and individual trees,individual tree diameter growth model,models of basal area and volume per hectare and model of stand average diameter at breast height.Using those models,a thinning program was made using non-linear optimization method for managing forest scientifically.The main contents and results were as follows:(1)Based on the data of 578 temporary plots for planted Larix olgensis in 3 forest farms(Mengjiagang,Jiangshanjiao and Songjianghe),the dual regression model of dominant height and average height was established.The dominant height and the average height can be estimated by the dual regression model.(2)The dynamic changes and the effects of thinning on the diameter distribution curves and the three parameters of Weibull function were analyzed.Based on the thinning effects,the parameter prediction model(PPM)of simultaneous equations system was developed and the parameters were estimated by three-stage least squares method(3SLS)and error-in-variable method of simultaneous equations.The results showed that the diameter distribution curves with stand age were changing from higher peak with narrow DBH range to lower peak with large DBH range for unthinned stand(control plot).For thinning plots,the skewness was more right-hand with higher peak,narrow DBH range,and increasing the asymmetry of the curves after thinning.Increasing with thinning intensity,more trees of small or middle size were cut and the curves of diameter distribution were obvious left-truncted higher peak with narrow DBH range than before thinning.The parameter "a" of the Weibull function increased with thinning intensity and the parameter "b" was decreasing.The parameter "c" was not influenced by thinning with lower intensity,but it was decreased for higher thinning intensity.The fitting results for the PPM of simultaneous equations system showed that predicting models for the parameter "b" were best(R2>0.99)and the R2 for model of the parameter "a"was larger than 0.93.The method of 3SLS was as good as the error-in-variable method,but there was no significant difference.The results of validation for the two methods showed that,the precision of the three prediction models were>97%,and the fitting efficiency(EF)were>0.93.The parameter prediction models developed based on the effects of thinnings performed well and could be used to predict the dynamic change of diameter distribution of planted Larix olgensis.(3)The thinning effects on motality at plot and tree level were analyzed and then models for the probability of mortality on sample plot and individual trees within a mortality plot were developed to modelling tree motality for Larix olgensis plantation.A two-step approach was used which consists of(1)estimating the probability of mortality on a sample plot;(2)estimating the mortality among individual trees within a mortality plot.Logistic regression models with "logit" as the link function were specified for the binary and binomial responses of plot mortality and the mortality rate of individual trees within a mortality plot.To overcome problems in modelling of clustered and longitudinally correlated data,the generalized estimating equations(GEE)approach for binomial responses were adopted to estimte the parameters of models.The data were divided into four groups according to thinning times and four models(model(1)-model(4))were developed in the plot level.The significant explicatory variables in model(1)were site index,the logarithm of stand age,thinning age and thinning intensity.There was a problem of multicollinearity between the significant variables for model(2)-model(4).Principal component analysis was used to solve multicollinearity and the primal variables were replaced by principal components for modeling the probability of mortality on plots.The significant primal variables for model(2)?model(4)were stand age,tree numbers per hectare,mean square diameter at breast height and thinning factors.In the tree level,the mortality model used trees in the mortality plots as research objects and the significant variables were planting density,age,the inverse of diameter at breast height and the basal area of all trees larger than the subject tree.An optimal cutpoint was determined for each model by plotting the sensitivity curve and the specificity curve and choosing the outpoint at which the specificity and sensitivity curves cross.The thinning variables were significant for plot mortality models but not significant for the mortality rate model of individual trees within the mortality plots.Hosmer and Lemeshow goodness of fit tests were not significant for models at both plot and tree levels(P>0.05).The areasunder the receiver operating characteristic curve(AUC)of the models were all above 0.91,the accuracies were all above 80%.The fitting results of the models were performed very well.(4)The effects of thinning on individu,al tree 5-year growth after thinned.A distance-independent individual tree diameter growth model based on thinning effects for Larix olgensis was developed to quantify the effects of thinning.The diameter growth models for Larix olgensis was difined as a single-level linear mixed model with random plot effect.The AIC,BIC statistics and likelihood ratio rest were used for model comparison and selection,and the models were tested using independent sample data.The result show that:the individual tree diameter growth was increased with the thinning intensity for youny stand under the same site index,but the thinning effects for individual tree diameter growth was not significant for stand of greater age.The individual tree growth was increased with diameter classes for smaller trees,but it showed opposite rule for bigger trees.The significant independent variables of the individual tree diameter growth models were the square of diameter at breast height(DBH),the natural logarithm of DBH,the basal area of trees larger than the subject tree,site index,thinning age and thinning intensity.The AIC,BIC statistics and root mean square error(RMSE)of the best individual tree diameter growth mixed model were less than that of general linear model for Larix olgensis.The best linear mixed model and general linear model of individual tree diameter growth for Larix olgensis achieved an efficiency of 0.678 and 0.624,respectively,when tested using independent sample data.The fitting and testing results showed that the best linear mixed model was better than the general linear model for of individual diameter growth for Larix olgensis.(5)The effects of thinning on gross growth,net growth,basal area and volume growth of living tree per hectare and average DBH were analyzed.Based on the results of analysis for thinning effects,the models of gross growth and net growth of basal area and volume and the model of average DBH were established.Coefficient of determination of those models were all>0.97.The precision of those models were all>99%tested by independent data.Taking plotl as examples,net growth of volume as objective function,thinning age and intensity as decision variable,an optimal thinning program was developed using no-linear optimization method.The optimal thinning times was 2 and 4.16%,15.74%higher than unthined stand when harvested at 40 and 51.The twice thinning ages were 21 and 30,thinning densities were 20.79%and 19.61%respectively,when harvested at 40.The twice thinning ages were 21 and 34,thinning densities were 27.72%and 27.19%,respectively,when harvested at 51.