Across-crown and canopy variations of plant functional traits and spectra for deciduous broadleaf and evergreen coniferous species in a temperate forest

Rapid and large-scale characterizations of forest canopies using remote sensing and modelling techniques are necessary, but the extent to which leaf-level traits and spectra can be upscaled to validate larger scale remote sensing data and vice versa is still not well understood. For one, spatial variations of leaf-level traits across tree crowns and canopies can bias results. While differences in extreme crown variations (e.g., shade and sun leaves) have been observed, across crown variation (i.e., sunlit leaves from top and sides of crown), which is of particular importance to increasingly higher resolution optical airborne and satellite remote sensing applications, have not been as widely documented.Here, we investigated the across-crown and canopy variations of leaf-level plant functional traits and spectral reflectances and transmittances in a temperate forest in northern Japan. Our study period spanned the growing and senescence periods from July to October 2023 as well as deciduous broadleaf and evergreen conifer tree species. For each tree, we collected and processed sunlit leaves from two different crown positions visible to overhead aircraft: the top of the tree crown (TC) and at the tree crown periphery (CP).We found differences in certain trait values (e.g., carbon (C) concentration,  d13C, leaf mass per area (LMA), chlorophyll, and total carotenoids) and spectra between TC (average height of 16.7 m above ground level) and CP (average height of 12.5 m above ground level) positions. We also show that differences between trait values and spectra between crown positions varied across tree species and over the growing and senescence periods. Overall reflectance and transmittance spectra across species showed increasing difference between TC and CP, especially in the near-infrared (NIR) region, from the growing to the senescence periods. Collectively, results suggest that crown sampling positions should be considered in the estimation of functional traits from spectral information, reflecting the dynamics of crown architecture. Particularly in spectral regions such as the NIR or short-wave infrared (SWIR), as well as during periods of senescence, where greater difference between crown positions were found.