Rapid prediction of the chemical traits of hybrid poplar with near infrared spectroscopy

Short-rotation crops such as hybrid poplars are likely to play an increasing role in the development of a bio-based refinery due to their potential for classical selection and breeding for desirable chemical traits. To efficiently monitor and improve the chemical traits of hybrid poplars for an application, a rapid measurement of their chemical composition is needed. Near infrared spectroscopy (NIRS) and partial least square (PLS) methods were used to develop chemometric models on hybrid poplar clones. Models for glucan, lignin, xylan, galactan, arabinan, mannan, extractive and ash contents were attempted on a series of 40 hybrid poplar clones. Models for the lignin syringyl/guaiacyl (S/G) ratio were also undertaken on a population of 21 hybrid poplar clones. Calibration models were thus developed and evaluated using the original NIR spectra and the 1st and 2nd spectral derivatives. For the chemical traits assessed in this study, the 1st derivative spectral treatment provided the best calibration models. When evaluated on an independent set of poplar samples the correlation coefficients (r) for predicting lignin, galactan, mannan contents and the lignin S/G ratio were high ranging from 0.92 to 0.95, while those for xylan and ash contents were acceptable at 0.80 and 0.82, respectively. In contrast, extractives, glucan and arabinan contents could not be predicted with any of the NIR models. For those chemical traits that were well modeled with NIRS (r > 0.90), the ratio of performance to deviation (RPD) varied between 2.1 and 2.5 demonstrating that these models could be used as initial screening methods to assist selection and breeding programs of hybrid poplars for applications in bioenergy and bioproducts.