Chemometric Modeling of the Structural Integrity, Chemistry and Bioenergy Potential of Elite Loblolly Pine Families and Forest Biomass
Type of DegreePhD Dissertation
Forestry and Wildlife Science
MetadataShow full item record
Pinus taeda L. (loblolly pine) is the most economically important tree species in the USA. With 30 million acres in plantations in the southern United States alone, it accounts for over 50% of the standing pine volume of the region. Over the past sixty years however, Southern Pine Decline (SPD) has been causing the premature death of this species. As a management strategy, stakeholders are selecting elite loblolly pine families that are currently being screened for tolerance to SPD for deployment. However, before deploying these elite families, important wood traits that dictate the quality of this essentially new feedstock towards different end uses must be known. This study examined the rapid screening of elite loblolly pine families for important wood properties and optimum utilization pathways using near infrared spectroscopy (NIR). In addition, the genetic variation, site and genotype by site interaction for the wood traits were investigated. Apart from its contribution to the conventional forest products industry, loblolly pine will play a role in the emerging bioeconomy. As such, this study also demonstrated the potential of NIR, as well as Fourier transform infrared spectroscopy (FTIR) and thermogravimetric analysis (TGA) in the rapid assessment of the heterogeneous loblolly pine biomass. NIR was used to rapidly determine the density, strength properties, chemical composition and bioenergy potential of the elite loblolly pine families. Three to five latent variables were used in the development of NIR-based partial least squares (PLS) models that had R2 values (cross-validation) of 0.58 to 0.88, and RPD values of 1.54 to 2.48. Validated models were employed in the screening of the families. The effects of family, site and family by site interaction were tested for the properties. Genotype of the loblolly pine families affected all the studied traits. In addition, the interaction term was significant for all the properties except for MOE. As such, tree breeders should bear in mind that desired traits of the elite families might be unstable on different sites. Further studies with more sites would be useful in estimating the extent of the genotype by site interactions. Nonetheless, desirable properties of some families remained high on the two forest sites. For example, A1, A26, A15, A2 and A9 which had consistently high cellulose contents on the two sites also had higher density, modulus of rupture (ultimate strength) and modulus of elasticity (stiffness). In addition, the amount of cellulose will affect the yield of pulp or ethanol. On the flip side, the strength-related properties of A33 and A21 remained low on both study sites. Apart from these two being undesirable for structural applications, their low strength properties could also make them more vulnerable to inclement weather on site. For the studies on the loblolly pine biomass, NIR and FTIR were used to classify forest residue into the plant part components of wood, wood & bark and slash (i.e. tops and limbs). Linear discriminant models that were developed with raw NIR and FTIR spectra had classification accuracies of over 96% for both tools. With respect to the quantitative assessment of biomass, full-cross-validated PLS and principal components regression (PCR) models were used. This study demonstrated that TGA coupled with chemometrics can be used for the compositional analysis of lignocellulosic biomass. The developed methodology enabled the simultaneous prediction of both the chemical and proximate properties from a single thermogram. According to the literature, this has not been attainable by the conventional deconvolution of TG data. In addition to its rapidity and simplicity, this alternative technique allowed the prediction of some monomeric sugars. Further studies will however be necessary to validate the capability of chemometrics to model the thermal degradation and quantitative prediction of the individual monomeric sugars. Comparing the predictive performance of the three analytical tools investigated in this study, the NIR models generally had better diagnostics relative to the FTIR and TGA models in predicting the chemical composition and thermal reactivity properties of the heterogeneous loblolly pine biomass.