@phdthesis{Buchner2025, author = {Buchner, Lisa}, title = {From leaf traits to canopy signatures: a multiscale and multisensory assessment of ash dieback in Fraxinus excelsior L.}, doi = {10.17904/ku.opus-1015}, url = {http://nbn-resolving.de/urn:nbn:de:bvb:824-opus4-10155}, school = {Katholische Universit{\"a}t Eichst{\"a}tt-Ingolstadt}, pages = {X, 95 Seiten : Illustrationen, Diagramme, Karten}, year = {2025}, abstract = {Ash dieback, a disease caused by the fungal pathogen Hymenoscyphus fraxineus, is severely threatening the existence of the European common ash (Fraxinus excelsior L.). The invasive pathogen leads to progressive symptoms such as leaf loss, shoot dieback, and stem necrosis, often resulting in high mortality in affected forest stands. While these visual symptoms are well documented, finer-scale physiological and morphological leaf responses remain largely unexplored. Ash dieback symptom severity is typically assessed through time-intensive field-based ratings; however, remote sensing technologies such as Unmanned Aerial Vehicles (UAVs) offer new opportunities for large-scale, remote assessment of disease impact. Therefore, this dissertation applies a multiscale and multisensory approach to assess the effects of ash dieback. The main research questions were: (1) Does ash dieback, beyond visible symptoms such as leaf loss and shoot dieback, also induce fine-scale morphological and physiological alterations in the leaves of infected ash trees? (2) Can multisensory UAV data and the thereof calculated vegetation indices detect different degrees of damage caused by ash dieback? (3) What level of segmentation accuracy is required to ensure reliable estimation of mean vege-tation index values for individual ash tree crowns? Field investigations were carried out at four study sites in 2022 and 2023, combining visual vitality assessments with detailed analyses of leaf physiology and morphology in ash trees affected by ash dieback. The examined leaf traits included chlorophyll fluorescence, chlorophyll content, Specific Leaf Area (SLA), leaf thickness, and Fluctuating Asymmetry (FA), all of which were evaluated in relation to visually assessed damage severity. In addition, at two of the sites, repeated UAV-based aerial surveys were conducted from May to October over two consecutive years to capture the whole vegetation period of the common ash, using RGB, multispectral and thermal sensors. Complementing this, close-range multispectral images of individual tree crowns were acquired in 2023 to provide higher spatial resolution data for detailed crown analysis. In Publication 1 of this dissertation, physiological and morphological leaf traits were investigated in relation to visually assessed disease severity. Among the measured traits, SLA exhibited the most consistent and significant correlation with increasing damage severity, highlighting a potential link between leaf morphology and ash dieback. In Publication 2 a novel UAV-based monitoring workflow that utilizes RGB and multispectral imagery is introduced to classify ash dieback severity via vegetation index thresholds. The study demonstrated that both RGB and multispectral indices, particularly the Green-Red Vegetation Index (GRVI) and Difference Vegetation Index (DVI), can effectively distinguish between mildly and severely damaged trees. The combination of both multispectral and RGB indices achieved a combined classification accuracy of 77.2 \%. Publication 3 explored the influence of ash tree crown segmentation precision on vegetation index reliability. A newly developed fine segmentation method, based on unsupervised machine learning, successfully excluded non-foliar elements such as ground pixels and canopy gaps, improving spectral data interpretation. Although mean vegetation index values per crown did not differ significantly between coarse and fine segmentation, vegetation index heterogeneity increased with disease severity, emphasizing the added value of detailed crown delineation for detecting subtle stress patterns. Collectively, these studies contribute a scalable and interdisciplinary framework that bridges leaf-level physiological and morphological measurements with crown-level spectral data and machine learning-based crown segmentation. They further demonstrate how UAV-based remote sensing can be effectively integrated into long-term conservation strategies for Fraxinus excelsior L. This work underscores the importance of combining plant physiology, remote sensing, and machine learning to advance forest health monitoring and offers practical insights for the conservation of Fraxinus excelsior L. under ongoing disease pressure.}, subject = {Eschen}, language = {en} }