Analysis of Energy Traits of Populus spp. Clones by Near-Infrared Spectroscopy

Abstract

The application of near-infrared spectroscopy (NIRS) and multivariate analysis for determining the calorific value and specific gravity of Populus spp. clones was assessed. Projection to latent structure (PLS) models of calorific value and specific gravity were developed from NIR original spectra and also from the first and second spectral derivatives. The best calibration models were built from the NIR second spectral derivative with good calibration statistics for both calorific value (r = 0 97, RMSEC = 0 05 kJ/g) and specific gravity (r = 0 98; RMSEC = 0 005). The calibration models from the NIR first spectral derivative were also good for specific gravity (r = 0 92, RMSEC = 0 010) and moderate for calorific value (r = 0 82; RMSEC = 0 11 kJ/g). When evaluated on a validation dataset, the models from the NIR first spectral derivative performed best for both specific gravity (r = 0 84; RMSEP = 0 021) and calorific value (r = 0 81; RMSEP = 0 13 kJ/g). In both cases, the standard errors of prediction (SEP) obtained from the NIRS calibration models were less than twice those of the corresponding laboratory measurement. The NIRS models were therefore useful for quickly determining calorific value and specific gravity of hybrid poplars but with a lower accuracy than the corresponding laboratory measurements. The study also helped delineate parentage as a factor of choice for manipulating wood specific gravity and thus biomass yield in hybrid poplars. On the other hand, calorific value was uniform within the population evaluated, indicating that little improvement in calorific value can be expected from selecting for it in hybrid poplar programs.