Hyperspectral Analysis of Tree Foliar Chemical and Physiological Responses to Abiotic and Biotic Stress

Monitoring forest health is crucial to understanding function and managing productivity of forest systems. However, traditional estimates of tree health are time consuming and challenging to collect because of the vertical and spatial scales of forest systems. This study evaluated the ability of a novel application of hyperspectral data to estimate foliar functional trait responses to multiple biotic and abiotic stressors and to classify different stress combinations. In a greenhouse environment, we exposed one-year-old black walnut (Juglans nigra L.) and red oak (Quercus rubra L.) seedlings to multiple stress factors, alone and in combination. We collected reference measurements of numerous leaf physiological traits and paired them with spectral collections to build predictive models. The resulting models reliably estimated most black walnut and red oak leaf functional traits with external validation goodness-of-fit (R2) ranging from 0.37 to 0.90 and normalized error ranging from 7.5% to 18.3%. Spectral data classified different individual stress groups well, but the ability of spectral data to classify stress groups depended on if the stress events were applied individually or in combination. High-dimensional spectral data can provide information about plant stress, improve forest monitoring in future predicted environments, and ultimately aid in management efforts in forest systems