Spectranomics is an emerging field of research that links the spectral-optical properties of leaves with trait and species diversity. Leaf spectra can be described as integrated foliar phenotypes that capture a wide range of functional traits which can provide insight into ecological processes. It has been proposed that leaf traits, and therefore leaf spectra, may reflect below-ground processes such as mycorrhizal associations. If such a relationship exists, this would enable the remote sensing of underground traits via the spectral properties of canopies, as has been claimed by recent studies. However, evidence for the relationship between leaf traits and mycorrhizal association is mixed, and few studies account for shared evolutionary history when testing these correlations. In this study, the relationship between leaf spectra and mycorrhizal associations was evaluated using leaf spectra for 92 vascular plant species measured through the Canadian Airborne Biodiversity Observatory. We modeled the evolution of leaf spectra and used phylogenetic comparative methods, implemented in a penalized likelihood framework for these highly dimensional data, to assess whether species associated with arbuscular mycorrhizas differ in spectral properties from species associated with ectomycorrhizas.
Results/Conclusions
We found that spectra evolve under the Ornstein-Uhlenbeck model with a phylogenetic half-life of approximately 46 million years, indicating that there is fairly strong selection on leaf spectra, as expected for an integrated phenotype of functional traits. Although non-phylogenetic methods identify differences in spectra corresponding to mycorrhizal type, using multivariate phylogenetic models of the high-dimensional spectral data, we found that arbuscular mycorrhizas and ectomycorrhizas do not differ in spectral properties. However, we found that univariate models of individual axes of variation representing specific spectral regions identify multiple optima corresponding with mycorrhizal association, suggesting that arbuscular and ectomycorrhizal plant species experience different selective regimes which may result in different spectral properties at these wavelengths. In conclusion, we demonstrate that spectra are not significantly associated with mycorrhizal association and that methods which do not account for phylogeny may be misattributing variation to mycorrhizal type rather than evolutionary history. However, for this dataset, mycorrhizal type may be predicted using non-phylogenetic methods including PLS-DA, enabling the use of these methods to predict underground traits using remote sensing technology.
