Preliminary study of colorimetry as an auxiliary tool for Manilkara spp. wood discrimination

This study aims to collect and calculate colorimetric parameters (CIELab) to group samples using a traditional colour classification chart. It also aims to test the efficiency of visible spectroscopy in the discrimination of wood identified as maçaranduba and maparajuba, from authorized exploitation in a conservation unit in the Brazilian Amazon, as well as to classify those results by a model for the discrimination between genera of the Sapotaceae family. The colour parameters and visible spectra of both RESEX and xylotheque samples were acquired for colour classification, Principal Component Analysis (PCA), and Linear Discriminant Analysis (LDA) to build and validate a model. The RESEX samples were categorized into 3 classes of brown and all subjects had darker and less saturated tones. PCA of the RESEX based on raw and treated data with Savitzky-Golay 1^st derivative presented a colour similarity among the samples, generating some groupings of Manilkara as well as dispersion of samples from the same subject. The LDA based on the raw xylotheque data with surface refreshment had an accuracy of 85.33% and classified all the samples as Pouteria. The LDA based on the raw xylotheque samples without surface refreshment had an 81.33% accuracy level and correctly identified most of the RESEX samples as Manilkara, however still presented misidentification as Pouteria and inconclusive/overlapping genus classification. This work corroborates how complex the relationships within the Sapotaceae family are and indicates a preliminary and potential use of VIS spectroscopy as an auxiliary tool for Manilkara spp. wood discrimination.