Near-infrared-based models for lignin syringyl/guaiacyl ratio of Eucalyptus benthamii and E. pellita using a streamlined thioacidolysis procedure as the reference method

Lignin syringyl/guaiacyl (S/G) ratio is a key trait in the cellulose pulping industry. Near-infrared (NIR) spectroscopy data collected with a bench and a portable instrument were used to develop prediction models for S/G ratio from sawdust samples of 1220 trees of two contrasting Eucalyptus species, namely E. benthamii (n = 484), which is a temperate, cold-resistant species, and E. pellita (n = 736), which is a heat- and disease-resistant tropical species. For each species, samples were selected based on maximal NIR spectral variation and analyzed for S/G using a streamlined thioacidolysis method with minimal sample input. NIR models were developed for each species separately and jointly. Instead of just using the coefficient of determination (R²) and the ratio of performance to deviation (RPD), the Spearman’s rank correlation (r_s) and the average of the coefficient of correlation between the references and predicted values (CV_RP) were taken into account to evaluate the models. The bench spectrometer had a better performance than the portable instrument (R² from 0.77 to 0.86 versus 0.31 to 0.47). Species-specific NIR models were better for E. benthamii (r_s = 0.89, CV_RP = 2.75%, R² = 0.86 and RPD = 1.8), while the joint-species model was better for E. pellita (r_s = 0.97, CV_RP = 3.98%, R² = 0.82 and RPD = 2.1). These NIR models should prove useful for high-throughput wood phenotyping in advanced breeding programs.