@article{hodge_acosta_unda_woodbridge_mansfield_2018, title={Global near infrared spectroscopy models to predict wood chemical properties of Eucalyptus}, volume={26}, ISSN={0967-0335 1751-6552}, url={http://dx.doi.org/10.1177/0967033518770211}, DOI={10.1177/0967033518770211}, abstractNote={ Global near infrared spectroscopy models (multiple-species, multiple-sites) were developed to predict chemical properties of Eucalyptus wood. The sample data set included 186 samples from four data sets (five species) originating from six countries: Eucalyptus urophylla from Argentina, Colombia, Venezuela, and South Africa; Eucalyptus dunnii from Uruguay; Eucalyptus globulus and Eucalyptus nitens from Chile; and Eucalyptus grandis from Colombia. The 186 samples were all preselected from larger collections of 400 to nearly 1800 samples to represent the range of chemical and spectral variation in each data set. The chemical traits modeled were total lignin, insoluble lignin, soluble lignin, syringyl–guaiacyl ratio (S/G), glucose, xylose, galactose, arabinose, and mannose. Single-species models and global multiple-species models were developed for each chemical constituent. For the global model, the R2cv for total lignin, insoluble lignin and syringyl–guaiacyl ratio were 0.95, 0.96, and 0.86, respectively. An alternate expression of the syringyl–guaiacyl relationship (S/(S+G)) resulted in better near infrared calibrations (e.g., for the global model, R2cv = 0.95). The global models for sugar content were also very good, but were slightly inferior to those for the lignin related traits, with R2cv = 0.74 for glucose, 0.89 for xylose, and from 0.72 to 0.91 for the minor sugars. To investigate the utility of the global models to predict chemical traits for species not included in the calibration, three-species calibrations were used to predict each trait in a fourth species data set. The prediction fit statistics ranged from excellent to poor depending on the species and trait, but in general the predictions would be at least moderately useful for most species-trait combinations. For some species-trait combinations with poor initial predictions from the global model, the inclusion of 10 samples from the “new” species into the calibration global model improved the fit statistics substantially. The global calibrations will be useful in tree breeding programs to rank species, families, and clones for important wood chemical traits. }, number={2}, journal={Journal of Near Infrared Spectroscopy}, publisher={SAGE Publications}, author={Hodge, Gary R and Acosta, Juan Jose and Unda, Faride and Woodbridge, William C and Mansfield, Shawn D}, year={2018}, month={Apr}, pages={117–132} }
 @article{hodge_woodbridge_2010, title={Global near infrared models to predict lignin and cellulose content of pine wood}, volume={18}, ISSN={["1751-6552"]}, DOI={10.1255/jnirs.902}, abstractNote={ Global near infrared models to predict lignin and cellulose content of pine wood were developed using 517 samples for lignin and 457 samples for cellulose. Samples came from seven different pine species, including tropical species ( Pinus caribaea, P. oocarpa, P. maximinoi, P. patula and P. tecunumanii) and temperate species ( P. radiata and P. taeda) from five different countries (Brazil, Colombia, Chile, South Africa and the USA). The global models were tested on an independent validation data set and had excellent fits for lignin [correlation coefficient ( r2)=0.97 and standard error of prediciton ( SEP) = 0.44] and good fits for cellulose ( r2 = 0.82 and SEP = 1.08). Subsets of the data were used to develop smaller multi-species, multi-site calibrations that could be tested on independent datasets containing different species not included in the calibration model. For calibrations based on four or more species, predictions from those models on independent datasets were generally good, with only slight degradation in r2 and SEP relative to the calibration R2 and SECV. The results suggest that global calibrations could be valuable in tree breeding programmes to rank species and genotypes for lignin and cellulose content. Species-specific models were developed for two species ( P. tecunumanii and P. taeda) which had sufficient numbers of observations; the global calibrations gave predictions as good as the species-specific calibrations. }, number={6}, journal={JOURNAL OF NEAR INFRARED SPECTROSCOPY}, author={Hodge, Gary R. and Woodbridge, William C.}, year={2010}, pages={367–380} }
 @article{schimleck_hodge_woodbridge_2010, title={Toward global calibrations for estimating the wood properties of tropical, sub-tropical and temperate pine species}, volume={18}, ISSN={["0967-0335"]}, DOI={10.1255/jnirs.900}, abstractNote={ Near infrared (NIR) spectroscopy is increasingly being used to replace traditional methods of wood property assessment and, as a result, multi-site, multi-species (or global) calibrations are of interest to organisations who assess wood properties on a large-scale. In this study, the development of global wood property calibrations for tropical, sub-tropical and temperate pines is explored. In a three-stage process, samples provided by ten forest industry companies and representing 14 pine species (two species had two varieties) and one hybrid, grown on 17 different sites in five countries (Argentina, Brazil, Chile, Colombia and South Africa) were used to develop calibrations for air-dry density, microfibril angle (MFA) and stiffness. Initial calibrations, based on samples from individual co-operators, had variable statistics; for example, R2 for density ranged from 0.51 to 0.89. In the second stage, multi co-operator calibrations were obtained from two independent data sets that included samples from sites in each country. For the three properties, calibration statistics compared well to those obtained in stage 1, with stiffness having the best fit statistics ( R2 = 0.917, RPDc=3.3). MFA and stiffness calibrations showed the presence of nonlinearity in the data. The calibrations performed reasonably well when used to predict the wood properties of the alternate set, with density having the weakest predictions. Stage 3 calibrations were based on all available samples and were similar to those developed in stage 2, demonstrating that it is possible to build multi-site, multi-species calibrations for pines. }, number={6}, journal={JOURNAL OF NEAR INFRARED SPECTROSCOPY}, author={Schimleck, Laurence R. and Hodge, Gary R. and Woodbridge, William}, year={2010}, pages={355–365} }
 @article{hodge_woodbridge_2004, title={Use of near infrared spectroscopy to predict lignin content in tropical and sub-tropical pines}, volume={12}, ISSN={["1751-6552"]}, DOI={10.1255/jnirs.447}, abstractNote={ Near infrared (NIR) spectroscopy was used to predict lignin content for trees from five species of tropical and sub-tropical pines ( Pinus caribaea, P. maximinoi, P. oocarpa, P. patula and P. tecunumanii) grown in Brazil and Colombia. Breast height disks were taken from 174 trees and wedges from the disks were sectioned to sample juvenile and mature wood. The sections were ground into woodmeal and NIR reflectance spectra were measured on both unextracted woodmeal and woodmeal with extractives removed. Klason lignin content was measured on the woodmeal samples and partial least squares were used to fit calibration equations to predict lignin content from the reflectance spectra. Good prediction models were obtained regardless of which data set (i.e. combinations of species and regions) was used for the model calibration. A model using reflectance spectra for woodmeal with extractives removed and combining data for all species across both regions had an R2 of 0.90 and standard error of cross-validation of 0.43% lignin for the calibration data set and an R2 = 0.91 and standard error of 0.40% for the validation data set. Calibration equations developed using only Brazilian or Colombian data were tested on the other data set. Predictions were very good, with prediction R2 ranging from 0.83 to 0.90 and standard errors of prediction from 0.43 to 0.54% lignin. }, number={6}, journal={JOURNAL OF NEAR INFRARED SPECTROSCOPY}, author={Hodge, GR and Woodbridge, WC}, year={2004}, pages={381–390} }
 @inbook{dvorak_jordan_hodge_romero_woodbridge_2000, title={The evolutionary history of the Mesoamerican Oocarpae}, ISBN={0620264608}, booktitle={Conservation and testing of tropical and subtropical forest tree species by the CAMCORE Cooperative}, publisher={Botha Hill, South Africa : Grow Graphics}, author={Dvorak, W. S. and Jordan, A. P. and Hodge, G. R. and Romero, J. L. and Woodbridge, W. C.}, year={2000}, pages={1} }