According to Liu et al. [8] the main difficulty in building such models is associated to the dissimilarity among biomasses (different NIR spectra). It is s not practical to develop a NIR calibration model with species showing large dissimilarity. So, to ensure a good prediction and reliable result, principal component analysis (PCA) was performed [11] to justify the development of a single calibration model containing three different biomasses. Besides, the usual statistical parameters (calibration and validation plots, calibration and validation errors, among others) were used to ensure the confidence of the models.