A spectral library of native Caatinga vegetation to improve species discrimination and ecosystem monitoring in tropical dry forest

The Caatinga biome is part of the seasonally tropical dry forest system and represents a unique Brazilian biome due to its distinctive functional adaptations to extreme climatic conditions. In this semi-arid environment, rainfall is a primary driven to control plant functioning, influencing phenology and biomass production. Reflectance spectroscopy provides continuous measurements across a broad range of the electromagnetic spectrum and captures integrated signals related to leaf structural, biochemical, and physiological properties, as well as environmental conditions, making it a powerful tool for vegetation characterization and ecosystem monitoring. Within a hyperspectral framework, this study aims to develop a spectral library of native Caatinga species to support applications in ecology, remote sensing, conservation, restoration, and climate change research. The study was conducted in a 600 ha area of well-preserved native Caatinga in Petrolina, northeastern Brazil (09°52′32″ S, 40°05′10″ W), characterized by a BSWh’ semi-arid climate with a mean annual temperature of approximately 26 °C and an average annual precipitation of 498.5 mm. Leaf-level spectral reflectance (350–2500 nm) was measured for 13 dominant native shrub and tree species using a portable ASD FieldSpec®3 spectroradiometer. Measurements were performed on healthy, fully developed leaves using a leaf clip with an internal halogen light source. Field campaigns conducted over three years captured both rainy and dry-season conditions under maintained foliation. Distinct spectral responses were observed in specific wavelength regions, although overall patterns were broadly similar across most species, reflecting measurements on the healthiest leaves. Notable interspecific differences were detected in the visible region (450–650 nm), particularly for Croton conduplicatus, Handroanthus spongiosus, Sapium glandulosum, Schinopsis brasiliensis, Senegalia piauhiensis, and Spondias tuberosa, with reflectance peaks shifting from green toward red, indicative of leaf senescence. Seasonal contrasts were also evident, with changes in reflectance peaks across the visible spectrum between rainy and dry conditions. This study establishes a comprehensive spectral dataset of dominant native Caatinga species across seasonal and hydrological gradients, providing a robust foundation for linking leaf-level spectral variability to environmental conditions. The resulting spectral library represents a critical contribution for improving species discrimination, ecological monitoring, and hyperspectral remote sensing applications in tropical dry forests, particularly within one of the most climatically vulnerable and understudied biomes in the world.

Keywords: Caatinga, Hyperspectral reflectance, Leaf traits, Species discrimination

Acknowlegments: The authors thanks to the financial support from the São Paulo Research Foundation (FAPESP) Grant #2022/07735-5, the Pernambuco State Research Support Foundation (FACEPE) (Grant # BFP-0103-5.01/23), the National Council for Scientific and Technological Development (CNPq) (Grant #403692/2024-5), and the Brazilian Agricultural Research Corporation (EMBRAPA) (Grants #10.23.00.111.00.00 and 10.25.00.144.00.00).