Evaluating as precisely as possible the genetic variability of quantitative traits is crucial in the context of breeding. It requires accurate, fast and cost-effective evaluation tools. We hypothesized that near infrared spectroscopy (NIRS), which is traditionally used to indirectly evaluate the physical and chemical properties of biological samples, could capture genetic variation and thus be used to predict the heritable variation of quantitative traits. To test this hypothesis, we collected NIRS spectra on wood samples harvested in a cloned association population of Populus nigra. After verifying that NIRS could capture a significant amount of genetic variation, we used this information to predict various phenotypes of interest related to growth, branching, phenology and rust resistance. The prediction accuracies estimated within a cross-validation scheme varied greatly depending on the phenotype, reaching a maximum of 0.74 for the circumference of the stem whose wood was collected for NIRS analyses. More surprisingly, accuracies around 0.7 were also found for traits related to bud flush despite a low correlation between bud flush and stem circumference (|rho] ~ 0.25-0.3). In addition, the accuracies of NIRS-based predictions were in the range of those obtained in the same dataset through genomic prediction with ca. 8,000 SNPs, demonstrating the usefulness of NIRS for predicting quantitative phenotypes. Moreover, phenomic and genomic selection accuracies were correlated together and to both the broad- and narrow-sense heritabilities of the corresponding traits, validating our initial hypothesis that NIRS is able to capture a significant amount of genetic information which can in turn indirectly be used to predict the heritable variation of quantitative traits.