Characterization of Forest Tree Seed Quality with Near Infrared Spectroscopy and Multivariate Analysis

摘要: The thesis presents a rapid and non-destructive method for characterizing the genetic, physiological technical qualities of both temperate tropical tree species on single seed basis. It is based ‘cross fertilization’ near infrared technology multivariate analysis. result demonstrated that sources, mothers fathers Pinus sylvestris could be identified using spectroscopy (NIRS) with 100%, 93% 71% classification accuracies, respectively. NIRS was employed to detect internal insect infestation in Cordia africana 100% sound infested seeds were achieved basis cuticular components moisture difference between two fractions. An extension this study performed Picea abies differing origin year collection. Detection uninfested found insensitive subtle differences proteins, lipids provided between-seed lot spectral variability removed priori appropriate pretreatment technique or calibration model takes into account natural variability. Sound insect-damaged Albizia schimperiana also successfully separated relative water content. gradient intentionally created by soaking fractions at room temperature specified time owing fact hard impermeable coat intact does not allow diffusion water. application discrimination viable empty patula evaluated discriminated accuracy divergence lipid protein contents. A further made simultaneously discriminate filled, insect-infested three Larix species. recognition while rates filled ranged 90% 100%; highest being sukaczewii followed decidua gmelinii, In vigour analysis, it possible distinguish vigorous aged accuracy. results reported demonstrate capability combined analysis as tool several quality attributes. As establishment new forest plantations shows an increasing trend globally, will play pivotal role upgrading through sorting unproductive seeds, hence facilitating sowing containerised seedling production nurseries and/or direct out field. Therefore, continued emphasis should given towards developing simple, cost-effective automated system future.