Can spectral phenotypes discriminate subspecies? A case study at two European and Oriental beech forest stands

Assisted migration programs, introducing new better adapted species at critical locations in our forests, have the potential to mitigate the adverse effects of climate change through the increase of forest diversity and resilience. While such measures entail ecological risks related with invasiveness of exotic species or outbreaks of new diseases, introducing close relatives of native species or populations from different parts of the species range is seen as the ecologically safer option. However, due to the similar appearance of closely related species, monitoring based on the external phenotype becomes difficult and leaves genetic screening as the only reliable, yet expensive option, limiting our ability to monitor large geographic areas. Reflectance spectroscopy has emerged as an important tool to assess plant functional trait distributions and taxonomic diversity, representing a rapid, scalable and integrated measure of the plant external and internal phenotype.

Here, we examine the potential of leaf-level reflectance spectroscopy to discriminate between the subspecies European beech (Fagus sylvatica L.), and Oriental beech (Fagus sylvatica spp Orientalis (Lipsky) Greut. & Burd), which has been proposed as a potential candidate for assisted migration in European forests due to its greater genetic diversity and potentially higher drought tolerance. We investigated two European beech forests in France and Switzerland where Oriental beech from the Caucasus was introduced over 100 years ago next to European beech. Over the summers of 2021 and 2022, we measured leaf spectral reflectance and leaf morphological and biochemical traits from previously genotyped adult trees.

Using least squares discriminant analysis (PLS-DA), we found that leaf spectral reflectance allowed the accurate discrimination of the two beech subspecies. In particular, we found that the short-wave-infrared (SWIR) region between 1450-1750 nm from top-of-canopy leaves provided the most accurate subspecies discrimination (BA = 0.86±0.08, k = 0.72±0.15). To provide a mechanistic basis of our findings, we estimated a suite of leaf traits based on spectra-derived indices and standard field and lab protocols. Phenotyping confirmed significant subspecies differences between traits that are known to govern light-plant interactions in the SWIR, including lignin, nitrogen in proteins, leaf mass per area and leaf thickness.

Our study provides a basis for crown-level subspecies classifications from airborne or satellite-based imagery in the genus Fagus. Our findings provide an important starting point for the interpretation of variability in tree crown reflectance and the superior discrimination capacity we found for leaves at the top as compared to leaves at the bottom of the canopy, holds promise for the upscaling of the method using remote sensing.