Holzforschung, Vol. 60, pp. 29–31, 2006 • Copyright  by Walter de Gruyter • Berlin • New York. DOI 10.1515/HF.2006.006 Article in press - uncorrected proof Short Note Calibration of NIR to assess lignin composition (H/G ratio) in maritime pine wood using analytical pyrolysis as the reference method Ana Alves 1 , Manfred Schwanninger 2 , Helena Pereira 1,3 and Jose ´ Rodrigues 3, * 1 Centro de Estudos Florestais, Instituto Superior de Agronomia, Universidade Te ´ cnica de Lisboa, Lisboa, Portugal 2 BOKU-University of Natural Resources and Applied Life Sciences, Vienna, Department of Chemistry, Vienna, Austria 3 Tropical Research Institute of Portugal (IICT), Forestry and Forest Products Group, Lisboa, Portugal *Corresponding author: Jose ´ Rodrigues, Tropical Research Institute of Portugal (IICT), Forestry and Forest Products Group, Tapada da Ajuda, 1349-017 Lisboa, Portugal E-mail: jocarod@isa.utl.pt Keywords: analytical pyrolysis; H/G ratio; lignin compo- sition; maritime pine; NIR; partial least-squares regres- sion (PLSR). Introduction Gymnosperm (softwood) lignins such as from maritime pine are complex aromatic heteropolymers composed of p-hydroxyphenyl (H), guaiacyl (G) and, if detectable (Saito et al. 2005), negligible (Obst and Landucci 1986) amounts of syringyl (S) phenylpropanoid units. The importance of lignin composition in terms of H/G ratio in softwoods for the pulping industry has led to attempts at genetic modification (MacKay et al. 1999; Wadenba ¨ ck et al. 2004). Nevertheless, very little is known about the natural variation of lignin composition. The lack of a suit- able method to assess lignin composition in a large number of samples can in part explain this situation. Ana- lytical pyrolysis is being increasingly used to assess lignin composition (Obst and Landucci 1986; Faix et al. 1993; Choi et al. 2001; Rodrigues et al. 2001; Yokoi et al. 2001; Kuroda et al. 2002; del Rio et al. 2005; Meier et al. 2005) but even if time requirements for the analytical procedure are well below those involved in wet-chemical analysis (Meier and Faix 1992) it is still demanding when large screening programs are necessary. The simplicity, rapidity, and high reproducibility ofnear- infrared spectroscopy (NIRS) have widened its use to the determination of the chemical composition of ligno- cellulosic materials as a substitute for wet-chemistry techniques (Schwanninger and Hinterstoisser 2001; Gier- linger et al. 2002; Raymond and Schimleck 2002; Schim- leck et al. 2003). The aim of this work was to use analytical pyrolysis data to develop an NIR method for determining H/G ratios, with a focus on assessment of the natural variation of lignin composition in maritime pine wood. Materials and methods A total of 68 maritime pine (Pinus pinaster Aiton) wood discs (da Silva Perez et al. 2005) were ground with a Thomas-Wiley ED-5 mill to pass a 1-mm sieve, screened in a vibratory sieving appa- ratus and the 40–60-mesh wood meal fraction was retained for analysis. The samples were extracted for 16 h with water, fol- lowed by 12-h acetone extraction in a Soxhlet apparatus, and were then dried at 608C overnight. Klason lignin was determined using the sulfuric acid method (Schwanninger and Hinterstoisser 2002). Analytical pyrolysis was performed using a CDS Pyroprobe 1000 with a coil filament probe connected to a GC (HP 5890 FID) via a heated interface (2808C). The pyrolysis was carried out at 6508C for 10 s, using 85 mg of the extractive-free samples. At least duplicate analyses were performed. Details of the con- ditions and quantification procedures have been published else- where (Rodrigues et al. 1999, 2001). NIR spectra of the extractive-free wood meal samples were recorded on a Bruker Vector 22/N using a spinning cup module with 50 scans per sample at spectral resolution of 8 cm y1 and two or three spectra per sample. Samples were randomly divided into two groups (1 and 2). Each group was used for cross-validation (CV) and test set val- idation (TS), with group 1 as CV and group 2 as TS, and then the other way round, to evaluate if the model statistics were identical, or at least very similar, leading to the same rank. The sample spectra were pre-processed using first derivative-multi- plicative scattering correction (MSC), and regressed against the calibration components using OPUS Quant 2 Software (Bruker). Then a partial least-squares regression (PLSR) model including all calibration spectra was calculated (Gierlinger et al. 2002). Results and discussion The H/G ratio of the 68 maritime pine wood samples determined by analytical pyrolysis ranged from 0.041 to 0.111, with an average of 0.064 and a standard deviation of 0.016. The precision of the method was 0.005 based on the pooled standard deviation of the replicates. This data set was used for calibration of the NIR spec- tra. The wavenumber range (6100–5760 cm y1 ) and the pre-processing method (first derivative-MSC) gave the best model for prediction of the H/G ratio (Table 1). The same parameters for wavenumber range and pre-pro- cessing method were applied for the determination of the