No abstract

Important biochemical traits may be robustly mapped by optical spectroscopy, allowing inferences about phylogenetic conservation in plant species. However, distinct types of data processing might lead to distinct patterns of phylogenetic signal in the foliar spectra. Thus, investigate the standard analytical approaches in order to understand their influence over the phylogenetic signal is essential. In this context, this study investigates how untransformed and transformed foliar-spectral data affects the phylogenetic signal of plant species located in regenerating forest gaps. Spectroscopic measurements from the adaxial surface of leaf samples were taken under standard light and temperature conditions for 53 regenerating plant species with a field spectroradiometer. Then, the average spectral signature for each specie was considered under two types of data processing: untransformed (raw reflectance spectra) and transformed (normalization by first derivative). Examined spectral regions for untransformed and transformed wavelengths were VIS (visible region: 400–700 nm), NIR (near infrared region: 701–1349 nm), SWIR-1 (short-wave infrared region part one: 1551–1849 nm) and SWIR-2 (short-wave infrared region part two: 2051–2450). Evolutionary conservation was evaluated through Blomberg (K) and Pagel (λ) metrics, which were calculated for all 1649 bands considering the average specie’s spectra. Thus, the percentage of wavelength with significant phylogenetic signal (K and λ) was quantified in both types of spectral processing. Significant phylogenetic signal was found for transformed spectra in NIR and SWIR-1 regions, along with reduced portions in SWIR-2. For untransformed spectra, there was significant signal mainly in SWIR-2. In conclusion, main results indicate that normalization by first derivative performs better in disentangling overlapping wavelengths. Thus, the transformed spectra can highlight the phylogenetic signal of plant features that are underemphasized in untransformed foliar spectra