Foliar sampling with an unmanned aerial system (UAS) reveals spectral and functional trait differences within tree crowns

Imaging spectroscopy is currently the best approach for continuously mapping forest canopy traits, which is important for ecosystem and biodiversity assessments. Ideally, models are trained with trait data from fully sunlit leaves from the top of the canopy. However, sampling leaves at the top of the canopy is often difficult, and sunlit foliage from the crown periphery is collected instead, assuming minimal within-crown trait variation among sunlit leaves. We tested the degree to which crown position affects foliar traits and spectra using mixed-effects models comparing sun leaves from the crown centre of mature sugar maple (Acer saccharum Marsh.) trees collected with DeLeaves, a novel twig-sampling unmanned aerial system device, with sun leaves from the crown periphery collected with a pole pruner. Sun leaves from the crown centre differed from sun leaves from the crown periphery in absorption, reflectance, transmittance, and a series of foliar traits, including leaf thickness, leaf mass, and leaf nitrogen content per unit area, demonstrating differences in resource allocation depending on sun exposure. Our study highlights the importance of accurately matching the location of foliar samples and spectral data and of sampling across gradients of intra-individual variation for the accurate prediction of foliar trait distributions across and within canopies using imaging spectroscopy.