Espectroscopia de campo para a discriminação de espécies exóticas invasoras e nativas em floresta subtropical

Abstract

Alien species have adaptive and competitive advantages compared to coexisting native species. These advantages can be expressed in their leaf functional traits and should be used in order to discriminate invasive plants in a complex forest environments. For this reason, it is necessary to understand the relationship of these characteristics with their spectral signatures. We propose a methodological framework to assess discrimination between native and invasive species based on their leaf functional traits associated with field spectroscopy. For this purpose, four tree species were analysed, two invasive (P. guajava and H. dulcis) and two native species (P.cattleianum and L. divaricata). First we evaluated the leaf functional traits and the spectral behavior of invasive and native plants. Photosynthetic pigments were responsible for the greatest interspecific variability especially in the green region around 550nm. Afterwards we applied a Linear Discriminant Analysis (LDA) to analyze the reflectance spectra and we achieved accuracy of 97% in discrimination of species. The most informative spectra are concentrated in the visible light close to the main pigment absorption features. Finally, we aim to understand the functional significance of the optical properties of invasive species by using a technique based on the combination of narrow bands and linear regression models. P. guajava had better functionally optical properties for: specific leaf area (R822/R801; R²=0,95); Car/Clh (R839/R785; R²=0,93) and relative water content (R817/R802; R²=0,83). H. dulcis had the best results for relative water content (R706/R531; R²=0,85), specific leaf area (R706/R531; R²=0,77) and Clha/Clhb (R818/R769; R²=0,47). The spectral responses of invasive species in this study can be used to develop canopy models to increase spatio-temporal resolution through the use of reflectance data extracted from images acquired from suborbital and orbital platforms. Overall, we propose that this methodology is appropriate for discriminating invasive plants. However, parameterization by species is necessary to allow it to be expanded to other species of interest.