Résumé
High-throughput techniques for the compositional analysis of lignocellulosic biomass are essential to allow the genetic analysis and genetic improvement of bioenergy feedstocks. In this study, we investigated the feasibility of using near-infrared (NIR) spectroscopy for rapid assessment of wood chemical traits in a large sample of Populus nigra L. individuals evaluated in clonal trials at two contrasting sites. Spectra were acquired from 5799 wood samples collected in 3 harvests corresponding to two coppice rotations at one site and one coppice rotation at the second. Calibrations were developed and validated using 120 reference samples, representing spectral and chemical variations in the samples. The resulting global and site specific calibrations for most of the traits were at least good enough for ranking of genotypes, demonstrating the usefulness of NIR analysis for phenotyping the studied population. Clonal repeatability (View the MathML sourceHc2) estimates of the studied traits based on all samples were moderate to high (View the MathML sourceHc2 ranging from 0.57 to 0.89 in the 3 harvests). When data were pooled over coppice rotations or sites, the genotype × environment interaction was more evident across sites than across rotations. However, the interaction was smaller than the genotype main effect for all traits, except for glucose and extractives contents. Importantly, the interaction resulted mainly from re-ranking of a few genotypes leaving a substantial amount of stable and performant genetic material, which may encourage breeding for improved main wood components. Optimization of the NIR analysis for assessing clonal trials would facilitate the exploitation of standing genetic variation of energy or chemical related traits in tree breeding program.