Modeling the wood properties of loblolly pine (Pinus Taeda L.) growing in southern United States
MetadataShow full item record
This dissertation is divided in to two broad sections. In the first section, models are developed that explain specific gravity (SG) variation within planted loblolly pine (Pinus taeda L.) from the southern United States. A three segmented quadratic model and a semiparametric model were proposed to explain the longitudinal variation of SG within-trees. Based on both models, regional variation in mean disk SG was observed. SG was highest for trees from the South Atlantic and Gulf Coastal Plain and lowest for trees from the Upper Coastal, Hilly Coastal, North Atlantic Coastal Plain and Piedmont. Maps explaining the regional variation in SG at specified heights within a tree were also developed based on the semiparametric model. A multivariate model system for disk SG and moisture content (MC) was also proposed. The proposed model system took account of the contemporaneous correlation between the two properties and was utilized to improve the prediction of one property given that information on the other property was available. Two subject specific prediction approaches commonly used in forestry (Generalized Algebraic Difference Approach - GADA and Nonlinear Mixed Models - NLMM), were evaluated empirically using a subset of disk SG data. The NLMM approach was found to perform better than GADA in terms of root mean square error (RMSE), and mean absolute residual (MAR). In the second section the effects of midrotation fertilization on various growth and wood properties were evaluated and the influence of fertilization on latewood SG was modeled. The effects of midrotation fertilization on growth and wood properties of loblolly pine in thinned and unthinned stands were explored. It was observed that both growth and wood property responses were higher in magnitude in the stand that received midrotation fertilization following thinning. Finally, the response of latewood SG following midrotation fertilization in a thinned stand followed a consistent pattern with the rate of nitrogen applied and was modeled successfully.