Assessing functional diversity and composition using multispectral images in Atlantic Forest restoration sites

Abstract Introduction By focusing on the specific functional traits of species, we can better understand how ecosystems function and respond to environmental change. However, field measurements of functional traits can be time‐consuming and costly, whereas remote sensing data can serve as viable alternatives for studying biodiversity at larger scales. Objectives We used unmanned aerial vehicle (UAV)‐acquired multispectral images to quantify the composition and functional diversity in ecological restoration projects using a trait‐based approach in the Atlantic Forest, Brazil. Methods We analyzed relationships between functional composition and diversity parameters measured in 24 field plots (20 × 20 m) and vegetation indices determined based on UAV‐collected images (UAV‐VIs). Additionally, two weighting approaches for the field data were evaluated. We then fitted and validated regression models to estimate functional composition and diversity using selected UAV‐VIs. Results Spearman correlations between field and remote data ranged from −0.85 to 0.72. Functional diversity indices showed higher correlations when weighted by abundance, while Community Weighted Mean measurements showed higher correlations when weighted by the product of canopy area and height. Simple linear regression models achieved coefficients of determination ( R 2 ) of up to 0.69. Multiple linear regression models allowed better fits ( R 2 up to 0.81). Conclusions There are still challenges and opportunities associated with measuring plant biodiversity using multispectral images. These results may support the development of multispectral remote sensing approaches as a proxy for ecosystem functional diversity and composition.