Individual Tree Mortality Models for Six Plantations in Taiwan

• Main Authors: Yang, Ming-Xin, 楊銘鑫
• Other Authors: Wang, Chao-Huan
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/99256638856727631016

碩士 === 國立宜蘭大學 === 森林暨自然資源學系碩士班 === 99 === As the social and economic development, people are increasing to emphasis on management of plantations. In order to predict the growth and development of plantation, Logistic regression models are presented to construct individual tree mortality models for Chamaecyparis formosensis, Taiwania cryptomerioides, Cryptomeria japonica, Calicedrus Formosans, Acacia confuse and Fraxinus groffothii plantations using the permanent plots data from Taiwan Forestry Bureau. The independent variables are diameter at breast height (D), site index (SI), crown ratio (CR), stand basal area in trees larger than subject tree (BAL) and the average dominant tree height (Hd). The dependent variable is a binomial variable for survival and death, using the SAS Proc Logistic for analysis. After all the variables through Variance inflation factor (VIF) test, then chi-square(χ2) test, the AIC value and the SC value are used for the independent variables selection. For those plots without planting year records, Hd is used to replace the SI to establish models. The results show that the mortality models of all species has D or D2, so the main impact factor of the mortality models is the diameter at breast height. The regression coefficient of D is negative, the representative of mortality rate decreasing as D increasing. The regression coefficient of D2 in Chamaecyparis formosensis and Fraxinus groffothii is positive, the representative of initial mortality rate decreasing as D increasing, but after the minimum point, it increasing as D increasing. Hd in Cryptomeria japonica model is not significant, but in the other species models are positive significant. SI in Chamaecyparis formosensis, Taiwania cryptomerioides and Calicedrus Formosans models is positively significant, but in Acacia confuse model is negatively significant. All the models through Chi-square (χ2) test, so they can have good prediction and explanation. When the lack of the planting year record, we can use Hd instead of SI to establish models.