11163 research outputs found
Sort by
Advancing Differentially Private Deep Learning: Algorithmic Improvements and Performance Evaluation
No full textIn recent years, machine learning and deep learning technologies have been widely applied across various domains. However, the privacy concerns associated with these technologies have become increasingly prominent. In this context, privacy-preserving training of deep learning models has emerged as a significant area of research. Differentially private Stochastic Gradient Descent (DPSGD), a privacy-preserving Deep Learning (DL) framework based on Differential Privacy (DP), has been extensively adopted to protect the privacy of training data. Nevertheless, the application of DP often leads to a decline in model performance. To address this issue, this paper proposes two enhanced methods: Particle Swarm Optimization-Based DPSGD (PSO-DPSGD) and Advanced DPSGD with Clipping-Ratio-Based Selective Update (ADPSGD). These algorithms aim to improve the performance of differentially private deep learning models through adaptive gradient clipping and optimized training trajectory, respectively. Experimental results demonstrate that under the same privacy budget (ϵ), the proposed methods achieve significantly higher model accuracy compared to the standard DPSGD algorithm. Furthermore, this study systematically summarizes evaluation work on DPSGD and investigates whether different strategies preserve equivalent privacy protection under the same privacy budget. Empirical evidence from membership inference attacks further demonstrates the practical guarantee of DPSGD variants. This study provides effective solutions for balancing model performance and privacy guarantees in differentially private deep learning, offering valuable insights for privacy-sensitive applications in real-world scenarios.2026-01-2
Labour and Vertical Empire: High-Altitude Porters in Himalayan Mountaineering c. 1890s - 1930s
No full textThis dissertation re-examines the history of Himalayan mountaineering from the 1890s to the 1930s, shifting the focus from elite climbers to the indigenous porters whose labour was foundational to these mountaineering expeditions. It analyses how British and German mountaineering expeditions used high-altitude endeavours to assert national prestige, scientific authority, and racial hierarchies. Drawing on official records and expeditionary archives of the German Alpine Club (Munich), the Alpine Club (London) and the Royal Geographical Society (London), the dissertation reveals how groups such as the Sherpas and Bhutias were categorised under colonial racial doctrines and managed within structured, asymmetrical labour systems. The dissertation moves beyond traditional narratives of heroic ascent and evocations of a partnership between climbers and porters, and highlights the agency and resistance of these workers, who navigated complex relationships of dependency and professionalisation. Ultimately, the dissertation illustrates how the vertical frontiers of empire became sites where modern identities of masculinity, race, and nationalism were contested and performed.2026-01-2
Sending AI to Med School: Increasing the Level of Evidence in Evaluation and Development of AI-ECG Models for Cardiovascular Diseases
No full textArtificial intelligence (AI) models trained on electrocardiograms (AI-ECG) can automatically learn highly specific disease pattern for cardiovascular risk assessment and diagnosis. However, it is typically not clear which part of the electric heart signal is responsible for the AI’s decision. This thesis examines how the level of evidence in AI-ECG models can be increased to facilitate clinical application.
Models were trained end-to-end with raw ECG signals as input, evaluated using explainable AI (XAI) methods that reveal the AI’s decision making process, linked to physiologically grounded ECG parameters that already hold high clinical evidence, and its prediction capability was externally validated to test generalization. The pipeline was
tested across three use cases: (I) AI-derived ECG heart age as a potential biomarker for cardiovascular disease, (II) adaptation of 12-lead AI-ECG models to single-lead input for arrhythmia detection in wearable devices in a clinical study with patients suffering from heart failure with reduced ejection fraction, and (III) longitudinal analyses that compare AI prediction and established ECG parameter with respect to predictive capability in disease detection.
XAI results showed that model’s decision making aligned with ECG parameters supported by clinical evidence across population-based, in wearable, and longitudinal settings. The model’s prediction capability was comparable in all external validations indicating that the learned effects can be generalized across different populations and healthcare systems. Additionally, ECG parameters that are known to be closely connected to the investigated diseases also correlate directly with the AI-ECG’s predictions. Overall, the thesis outlines a practical framework for the development and evaluation of AI-ECG models that connect XAI results, established ECG parameter, and model’s prediction capability to increase the level of evidence and potentially support application of AI-ECG in early detection and prevention of cardiovascular diseases.2026-01-2
Phylogenomics and functional genomics of several streptophyte algae
No full textDer Übergang der Pflanzen von aquatischen zu terrestrischen Lebensräumen stellt einen entscheidenden Meilenstein in der biologischen Geschichte der Erde dar und katalysierte die Entwicklung der vielfältigen Ökosysteme unseres Planeten. Streptophyten-Algen, die nächsten Verwandten der Landpflanzen, spielten eine grundlegende Rolle in diesem Transformationsprozess, indem sie Anpassungen entwickelten und verfeinerten, die das Leben an Land ermöglichten. Diese Arbeit untersucht die evolutionäre Entwicklung der Streptophyten-Algen und legt den Schwerpunkt auf ihre innovativen Strategien zur Bewältigung abiotischer Herausforderungen wie ultravioletter Strahlung, Austrocknung und Temperaturschwankungen. Zu diesen Anpassungen zählen photoprotektive Systeme, fortschrittliche Zellwandstrukturen sowie molekulare Mechanismen zur Kontrolle oxidativen Stresses. So ermöglichte beispielsweise die Evolution nicht-photochemischer Quenchmechanismen eine effiziente Dissipation überschüssiger Lichtenergie, während die Reaktionen auf Austrocknungsstress strukturelle Verstärkungen und biochemische Signalwege zur Neutralisierung reaktiver Sauerstoffspezies (ROS) umfassten. Diese evolutionären Innovationen bildeten das grundlegende Werkzeugset für die Terrestrialisierung der Pflanzen.
Zentrale streptophyte Linien wie die KCM- und ZCC-Grade weisen ein bemerkenswertes Spektrum ökologischer Anpassungen auf. Die Klebsormidiophyceae zeigen einzigartige Merkmale, die auf das Überleben in extremen Lebensräumen – von Polarregionen bis hin zu ariden Wüsten – zugeschnitten sind, während die Zygnematophyceae evolutionäre Innovationen in der zellulären Organisation und Stresstoleranz aufweisen und als die nächsten algalen Verwandten der Embryophyten gelten. Eine detaillierte Analyse ihrer Phylogenomik und funktionellen Genomik liefert Einblicke in die molekularen und genetischen Mechanismen, die ihren evolutionären Erfolg ermöglichten und die Grundlage für die Evolution der Landpflanzen schufen.
Im Zentrum dieser Studie steht die Entwicklung und Anwendung eines robusten phylogenomischen Frameworks, PhyloRSeq++, das zentrale Herausforderungen der phylogenetischen Rekonstruktion adressiert. Diese Pipeline löst Probleme wie Inkongruenzen zwischen Genbäumen, Kontamination von Datensätzen und die Identifikation von Orthologen und ermöglicht so die Generierung hochauflösender Artbäume. Die gewonnenen Erkenntnisse klären die evolutionären Beziehungen innerhalb der streptophyten Linien und offenbaren genomische Anpassungen, die die Terrestrialisierung begünstigten. So zeigt die genomische Architektur der Klebsormidiophyceae Anpassungen an extreme Austrocknungstoleranz, darunter reversible Zellschrumpfung und widerstandsfähige Zellwände, während Analysen von Signalnetzwerken in den Zygnematophyceae die evolutionären Ursprünge von Stressantwort-Signalwegen aufdecken, einschließlich Homologen der Abscisinsäure-(ABA-)Signalkaskade.
Die Integration phylogenomischer und funktionell-genomischer Ansätze verdeutlicht die Kontinuität zwischen den ursprünglichen Merkmalen streptophyter Algen und denjenigen der Embryophyten. Genexpressionsanalysen zeigen die Aktivierung evolutionär alter genetischer Werkzeuge unter Umweltstress und illustrieren so die evolutionäre Entwicklung von hydro-terrestrischen Algen hin zu Landpflanzen. Horizontaler Gentransfer, Genduplikationen und selektive Drücke erweiterten dieses genetische Repertoire weiter und versetzten diese Organismen in die Lage, komplexe Umweltbedingungen zu bewältigen.
Diese Arbeit trägt zur Klärung langjähriger Fragen nach den evolutionären Ursprüngen der Landpflanzen bei und vertieft zugleich unser Verständnis der genetischen und molekularen Signalwege, die ihre Terrestrialisierung ermöglichten. Durch die Aufklärung der Strategien, die die Pflanzenevolution geprägt haben, erweitert diese Forschung nicht nur unser Wissen über frühe terrestrische Ökosysteme, sondern liefert auch wertvolle Impulse für Anwendungen in Landwirtschaft, Naturschutz und Biotechnologie. Die Ergebnisse unterstreichen die Widerstandsfähigkeit und Anpassungsfähigkeit streptophyter Algen und zeichnen ein eindrucksvolles Bild der Innovationskraft der Evolution unter restriktiven Bedingungen.The transition of plants from aquatic to terrestrial habitats represents a pivotal milestone in Earth’s biological history, catalyzing the development of the planet’s diverse ecosystems. Streptophyte algae, the closest relatives of land plants, played a fundamental role in this transformation, developing and refining adaptations that enabled life on land. This research explores the evolutionary journey of streptophyte algae, emphasizing their innovative strategies to overcome abiotic challenges, including ultraviolet radiation, desiccation, and temperature fluctuations. These adaptations include photoprotective systems, advanced cell wall structures, and molecular mechanisms for managing oxidative stress. For instance, the evolution of non-photochemical quenching mechanisms enabled these organisms to dissipate excess light energy efficiently, while their responses to desiccation stress involved structural reinforcements and biochemical pathways to counteract reactive oxygen species (ROS). These evolutionary innovations provided the foundational toolkit for the terrestrialization of plants.
Key streptophyte lineages such as the KCM and ZCC grades showcase a remarkable range of ecological adaptations. The Klebsormidiophyceae exhibit unique traits tailored for survival in extreme environments, from polar regions to arid deserts, while the Zygnematophyceae demonstrate evolutionary innovations in cellular organization and stress tolerance, identifying them as the closest algal relatives to embryophytes. A detailed exploration of their phylogenomics and functional genomics sheds light on the molecular and genetic mechanisms that underpinned their evolutionary success and laid the groundwork for land plant evolution.
Central to this study is the development and application of a robust phylogenomic framework, PhyloRSeq++, which addresses critical challenges in phylogenetic reconstruction. This pipeline resolves issues such as gene tree incongruences, dataset contamination, and ortholog identification, enabling the generation of high-resolution species trees. These insights clarify the evolutionary relationships within streptophyte lineages and reveal genomic adaptations that facilitated terrestrialization. For example, the genomic architecture of Klebsormidiophyceae reveals adaptations for extreme desiccation tolerance, such as reversible cell shrinkage and durable cell walls, while signaling network analyses in Zygnematophyceae highlight the evolutionary origins of stress response pathways, including homologs of the abscisic acid (ABA) signaling cascade.
The integration of phylogenomic and functional genomic approaches reveals the continuity between ancestral traits of streptophyte algae and those observed in embryophytes. Gene expression analyses underscore the activation of ancient genetic tools under environmental stress, illustrating the evolutionary trajectory from hydro-terrestrial algae to land plants. Horizontal gene transfer, gene duplication, and selective pressures further expanded the genetic toolkit, enabling these organisms to navigate complex environmental challenges.
This work contributes to resolving long-standing questions about the evolutionary origins of land plants while advancing our understanding of the genetic and molecular pathways that enabled their terrestrialization. By illuminating the strategies that shaped plant evolution, this research not only enhances our knowledge of early terrestrial ecosystems but also informs applications in agriculture, conservation, and biotechnology. The findings underscore the resilience and adaptability of streptophyte algae, offering a compelling narrative of evolution’s power to innovate under constraint.2026-02-1
Transaldolase from Thermoplasma acidophilum: new reactivities and applications in biocatalysi
No full textThis project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 956631In this PhD thesis, the enzyme transaldolase from the thermophilic organism Thermoplasma
acidophilum (TacTAL) was studied from two related perspectives: its reactivity toward 5-carbon
phosphosugars and its potential applications in biocatalysis.
The non-oxidative branch of the pentose phosphate pathway maintains a delicate equilibrium,
involving enzymatic reactions among sugars of various lengths. TacTAL catalyzes the transfer of a
dihydroxyacetone (DHA) unit from donor substrates such as fructose-6-phosphate (F6P) or
sedoheptulose-7-phosphate (S7P) (6 or 7 carbon atoms) to acceptor substrates like erythrose-4
phosphate (E4P) or glyceraldehyde-3-phosphate (G3P) (3 or 4 carbon atoms). To preserve this
equilibrium, it is essential that TacTAL avoids reacting with other similar compounds present in
the cell.
The enzyme’s reactivity toward the most important 5-carbon phosphosugars involved in the non
oxidative branch of the pentose phosphate pathway such as xylulose-5-phosphate (X5P),
arabinose-5-phosphate (A5P), and ribose-5-phosphate (R5P), was investigated using biochemical,
biophysical and analytical techniques including X-ray crystallography, HPLC, steady-state kinetics,
and mass spectrometry. The results demonstrate that the phosphate-binding site of the enzyme
plays a crucial role in preventing X5P from penetrating deep enough into the active site to react
as a donor. A5P is not utilized as an acceptor and can only act as a donor with low efficiency. The
lower efficiency was due to the absence of a terminal α-CH2-OH group, a feature present in the
physiological substrates F6P and S7P. Meanwhile, R5P can function as an acceptor, producing
octulose-8-phosphate, but reacts as a donor with very low efficiency because its non-ideal
stereochemistry at the α-carbon hinders proper alignment in the active site.
The TacTAL E60Q/F132Y variant (aldolase variant) catalyzes the reversible cleavage of F6P into
DHA and G3P. With its exceptional thermostability and catalytic efficiency, TacTAL represents a
promising tool for biocatalytic applications. Rational design and inspiration from previous studies
on related enzymes were employed to further engineer the aldolase variant, expanding its
substrate scope to include non-phosphorylated polar compounds and non-polar substrates. Using
HPLC and HR-MS, the variants TacTAL E60Q/F132Y/R135E, E60Q/F132Y/R135V, and E60Q/F132Y
were found capable of catalyzing aldol condensation reactions with DHA or, alternatively,
hydroxyacetone as donors. Key reactions included the formation of 8- and 9-carbon sugars with
acceptor substrates such as R5P, ribose, L-rhamnose, and galacturonic acid, as well as reactions
involving less polar substrates like glyoxal and pyruvaldehyde. The structure of TacTAL in complex
with the 9-carbon sugars derived from DHA and L-rhamnose or DHA and galacturonic acid was
resolved, and the products of the most efficient reactions, particularly those with galacturonic
acid, were characterized through NMR. This study provides valuable insights into the substrate
specificity of TacTAL and its applications in biocatalysis.2025-07-0
Monocentric Evaluation of the 2017 ACR/EULAR Myositis Classification
No full textThis retrospective, single-center study evaluates the applicability and diagnostic accuracy of the 2017 ACR/EULAR classification criteria for idiopathic inflammatory myopathies (IIM), comparing it to previous established classification systems. A total of 97 patients treated at the University Medical Center Göttingen between 2003 and 2020 were included, all with clinically and histologically confirmed IIM subtypes, including polymyositis, dermatomyositis, necrotizing myopathy, inclusion body myositis, and unclassified forms. The study assessed clinical, serological, imaging, and histopathological data. Statistical analysis revealed that while older classification systems had variable sensitivity, the ACR/EULAR criteria showed high reliability, practical applicability, and diagnostic performance, particularly in polymyositis and dermatomyositis. The study supports the implementation of the ACR/EULAR classification in clinical practice for improved diagnostic consistency.2025-08-0
Optical Creation and Control of Magnetic Spin Textures
No full textExploring the rich physics of magnetic spin textures, particularly topologically non-trivial magnetic skyrmions, opens a promising pathway for the development of novel spintronic and magnonic devices. This thesis investigates the laser-induced dynamics of ferrimagnetic Fe/Gd multilayers, which host a broad variety of dipolar-stabilized magnetic spin textures at room temperature. The formation of either stripe domains, a dense lattice of topologically trivial magnetic bubbles and nontrivial magnetic skyrmions (BSKs), or a single-domain state can be controlled by adjusting the strength of an out-of-plane magnetic field. Ultrafast Kerr spectroscopy reveals the laser-induced dynamics of these magnetic spin textures, which are characterized by unique collective breathing modes enabling the unambiguous identification of stripe domains and the BSK lattice, even without resorting to imaging techniques. This thesis demonstrates versatile control of these magnetic spin textures by tuning i) the optical excitation fluence, ii) the base temperature, and iii) an external in-plane magnetic field. These control parameters enable i) the nonequilibrium nucleation and annihilation of the BSK lattice, ii) the identification of pathway-dependent ground states of magnetic spin textures within the (H,T)-phase diagram unveiling a nonadiabatic nucleation mechanism for BSKs, and iii) the significant enhancement of the proportion of skyrmions within the BSK lattice. In addition to these static control strategies, this work also showcases the possibility to directly control the spin dynamics of the BSK lattice. Employing dual-pulse excitation with variable time delay allows to initiate and to tailor the breathing mode of BSKs in both amplitude and phase. In total, this thesis explores multiple pathways for controlling dipolar-stabilized magnetic spin textures and manipulating laser-induced breathing mode dynamics. These findings provide valuable insights into the fundamental mechanisms of these processes, thereby paving the way for future applications.2026-02-0
Contribution of Heart-Brain Interactions to Conscious Visual Perception
No full textThe objective of this thesis is to investigate the interplay between afferent cardiac signals and neural processing of exteroceptive inputs and how this dynamic contributes to conscious visual perception. This investigation is grounded in two EEG-ECG studies that explore how exercise-induced modulation of cardiac activity influences perceptual suppression, a phenomenon where a salient visual stimulus becomes subjectively invisible despite unchanged sensory input. Chapter 1 introduces the anatomical, physiological, and theoretical foundations of heart-brain interactions in conscious perception, framing the aims of the experimental studies. Chapter 2 presents the first experimental study, which investigates how light-intensity cycling exercise affects perceptual suppression, pre-stimulus alpha activity, and heartbeat-evoked potentials (HEPs) using the Generalized Flash Suppression (GFS) paradigm. Exercise reduced the likelihood of perceptual suppression and decreased alpha amplitudes, indicating increased cortical excitability. These results challenge the baroreceptor hypothesis, which predicts that increased heart rate should reduce cortical excitability and suppress sensory processing. Additionally, exercise-induced decreases in HEP amplitudes provide novel evidence of the influence of cardiac physiology on the neural processing of cardiac signals. However, HEPs did not predict perceptual suppression. Nonetheless, the concurrent observation of low alpha activity, reflecting increased cortical excitability, and low HEP amplitudes, suggesting decreased interoceptive processing, indirectly hints at a potential attentional trade-off between exteroceptive and interoceptive processing during exercise. Chapter 3 reports the second experimental study, which examines the impact of exercise-induced heart rate changes on perceptual suppression dynamics in the Motion-Induced Blindness (MIB) paradigm. Unlike the GFS findings, no significant effect of exercise was observed on perceptual suppression rate or duration, and no correlations were found between heart rate changes and suppression metrics. Stimulus-locked analyses of HEPs in a Replay condition revealed significantly lower amplitudes during physical target removal compared to target presence. However, response-locked analyses of HEPs in both MIB and Replay conditions showed no significant differences, indicating no difference in interoceptive processing concurrent with subjective perceptual changes. Taken together, these findings suggest that cardiac influences on perceptual suppression may depend on the cognitive and neural demands of the specific paradigm. While subtle indications of attentional trade-offs between interoceptive and exteroceptive processing were observed, these effects appear to operate independently of cortical excitability. The absence of a clear link between interoceptive signals and perceptual outcomes points to a modulatory, rather than deterministic, role for interoception. Furthermore, the results reinforce the idea that neural variability plays a more prominent role in shaping perceptual processes than cardiac influences. Chapter 4 discusses the broader implications of these findings on the role of heart-brain interactions in conscious perception, highlights the limitations of the present studies, and outlines potential directions for future research.2025-07-0
Axonale Schädigung und Makrophageninfiltration bei Entmarkungserkrankungen
No full textEven though the etiology of multiple sclerosis (MS) has been of scientific interest for a long time, it still remains unclear what exactly causes the axonal damage leading to neurological dysfunction.
Especially since pathological changes can also be found in normal appearing white matter (NAWM) of visually intact myelin.
Therefore, this study uses different immunohistochemical markers to point out the range of phagocytic and lymphocytic inflammation as well as the extent of axonal damage within demyelinated lesions, periplaque white matter (PPWM) and NAWM of MS patients and for comparative reasons of NMO and PML patients.
A cohort of 25 MS patients (3 autopsies, 22 biopsies), 8 NMO patients (5 autopsies, 3 biopsies), 15 PML patients (11 autopsies, 4 biopsies) and a control group (6 autopsies, 5 biopsies) have been analyzed for this purpose.
The inflammatory activity in terms of macrophage infiltration and microglia activation was mostly measurable in lesions and PPWM, although there has also been an increased number of inflammatory cells in the NAMW.
To analyze the extent of axonal damage two different markers have been examined.
Disruptions in axonal transport detected by the amyloid precursor protein (APP) have only been slightly increased in the NAWM whereas non-phosphorylated neurofilament (SMI32) was especially measurable in the PPWM and NAWM of MS and NMO patients.
In conclusion, those findings bring up the question if the axonal damage in MS only derives from the demyelinating process so that further investigations regarding pre-lesional axonal pathologies might be useful for diagnostic and therapeutic purposes.2025-09-1
Land-atmosphere interactions of agroforestry systems in Germany: quantifying the ecosystem-scale carbon uptake and potential wind erosion reduction
No full textAufgrund des Klimawandels kommt es weltweit immer häufiger zu extremen Wetterbedingungen. Die Gesellschaft und Landwirtschaft sind mit intensiveren, lang anhaltenden Warm- und Trockenzeiten und extremen Niederschlägen konfrontiert. Diese sich verschärfenden Wetterbedingungen erfordern eine nachhaltige Intensivierung der Landwirtschaft, die widerstandsfähiger gegen Dürren und Erosion ist, so dass die weltweite Nahrungsmittelproduktion weiterhin gewährleistet werden kann. Darüber hinaus kann die Abschwächung und Verringerung von Kohlendioxid- (CO2) und anderen Treibhausgasemissionen dazu beitragen, die Auswirkungen des Klimawandels zu minimieren. Die Agroforstwirtschaft könnte eine gute Alternative zu Monokulturen oder Grünland sein, indem sie eine widerstandsfähigere und an den Klimawandel angepasste Landwirtschaft und eine zusätzliche Minderung von CO2 ermöglicht. In der vorliegenden Arbeit liegt der Schwerpunkt speziell auf Agroforstsystemen mit Kurzumtrieb und Alleenanbau (AF), d. h. Acker- oder Grünlandflächen, auf denen in großen Abständen parallele Baumreihen gepflanzt werden, die natürliche Alleen bilden. Es ist bekannt, dass die heterogen Struktur von Agrarforstsystemen eine Reihe von Vorteilen bieten, wie z. B. ein verändertes Mikroklima, das widerstandsfähiger gegenüber Wetterextremen ist, eine geringere Bodenerosion, eine größere Artenvielfalt und eine verbesserte Kohlenstoffaufnahme. Obwohl davon ausgegangen wird, dass die Winderosion verringert und die Kohlenstoffaufnahme im Laufe der Zeit durch die Aufforstung verbessert wird, wurde die Verringerung der Winderosion bisher nicht quantifiziert und die Kohlenstoffaufnahme im Laufe der Zeit nur mit punktuellen Messungen des Bodens und der Bäume gemessen. Ziel dieser Arbeit ist es daher, (1) die kontinuierliche Netto-Kohlenstoffaufnahme von Kurzumtriebsplantagen und angrenzenden Monokulturen auf Ökosystemebene zu vergleichen und zu bewerten und (2) die potenzielle Verringerung der Winderosion durch Kurzumtriebsplantagen zu quantifizieren. Um die Kohlenstoffaufnahme im Ökosystem bei AF und MC messen zu können, war ein zusätzliches Ziel (3) die Charakterisierung und Evaluierung neuartiger kostengünstiger Eddy-Kovarianz-Systeme (LC-EC).
Um diese Ziele zu erreichen, wurden von Mitte 2019 bis 2021 in Norddeutschland kontinuierliche Eddy-Kovarianz-Messungen (EC) des Kohlenstoffaustauschs auf vier AF-Systemen und vier angrenzenden MC-Systemen ohne Bäume durchgeführt. Die EC-Messungen wurden mit neuartigen LC-EC-Systemen durchgeführt, die auch in der aktuellen Arbeit getestet und charakterisiert wurden. Darüber hinaus wurden unter Verwendung simulierter Windgeschwindigkeitsreduktionen aus Large-Eddy-Simulationen in Kombination mit lokalen Windgeschwindigkeitsmessungen Abschätzungen der potentiellen Winderosionsreduktion innerhalb eines gesamten Agroforstsystems mit Kurzumtrieb vorgenommen.
Die simulierten Schätzungen der potenziellen Winderosion zeigten, dass ein Agroforstsystem in einem gemäßigten Klima die potenzielle Winderosion um mehr als 80% reduzieren kann, wenn die Agroforstfläche gut geplant und bewirtschaftet wird. Eine optimale Winderosion kann erreicht werden, wenn die Baumstreifen senkrecht oder diagonal zur Hauptwindrichtung ausgerichtet sind, wodurch die potentielle Winderosion um 92% bzw. 86% reduziert wird. Außerdem wurde gezeigt, dass die Höhe der Baumstreifen mit dem Abstand zwischen den Baumstreifen im Gleichgewicht sein muss und dass die Verringerung der Winderosion mit AF an Standorten mit anfälligen Sandböden zu einem geringeren Verlust an organischem Kohlenstoff (SOC) im Vergleich zu MC führen. Schließlich bot ein Abstand von ≤48 m zwischen dichten Baumstreifen von nur 2 m Höhe bereits eine potenzielle Winderosionsminderung von 86%, wenn sie senkrecht oder diagonal zur vorherrschenden Windrichtung ausgerichtet waren.
Die getestete und charakterisierte LC-EC zeigte eine gute Übereinstimmung mit gleichzeitigen konventionellen Eddy-Kovarianz-Messungen (CON-EC). Die LC-EC-CO2-Flüsse waren 4-7% (R² = 0,91-0,95) niedriger als die CON-EC-CO2-Flüsse, und auch die Flüsse latenter Wärme (LE) stimmten 2020 gut überein, mit 1-5% (R² = 0,86-0,91) höheren LE-Flüssen vom LE-EC im Vergleich zu Flüssen mittels CON-EC gemessen. Im Jahr 2021 überschätzten die LC-EC LE-Flüsse die CON-EC LE-Flüsse um 23% (R² = 0,84), was jedoch wahrscheinlich auf eine Unterschätzung der CON-EC-Flüsse zurückzuführen ist. Trotz der erhöhten Unsicherheit bei LC-EC aufgrund der Verwendung langsamer reagierender Sensoren war die Leistung von LC-EC vergleichbar mit der Variation der EC-Flüsse zwischen verschiedenen CON-EC-Setups. Daher konnte gezeigt werden, dass die neuartigen LC-EC-Aufbauten eine kostengünstigere Alternative zu CON-EC sind, mit ungefähr 25% der Kosten eines konventionellen EC Aufbaus.
Die kontinuierlichen LC-EC-Messungen zeigten, dass die Netto-Ökosystemproduktion (NEP) - oder die Netto-Kohlenstoffaufnahme - der AF-Flächen im Vergleich zu den angrenzenden MC-Flächen deutlich höher war. Die NEP auf den AF-Anbauflächen war im Durchschnitt 16--190% höher als auf den MC-Anbauflächen, und auf dem AF-Grünland war die NEP im Durchschnitt 195% höher als auf dem MC-Grünland. Es wurde auch gezeigt, dass auf den Ackerflächen die Bruttoprimärproduktion (GPP) und die Ökosystematmung (Reco) zwischen 10--35% und 9--15% auf der AF im Vergleich zur MC höher waren, und auf dem Grünland waren die GPP und Reco 13% and 31% niedriger auf der AF im Vergleich zur angrenzenden MC. Auf den Grünlandstandorten war der Unterschied im NEP zwischen AF und MC in jedem Jahr relativ konstant, während die Ackerstandorte starke jährliche Schwankungen aufwiesen, die von den lokalen Wetterbedingungen sowie der Boden- und Kulturart beeinflusst wurden. Darüber hinaus wird hervorgehoben, dass langfristige EC-Messungen ein detaillierteres und solideres Verständnis der Unterschiede in der Kohlenstoffbilanz zwischen AF- und MC-Standorten liefern könnten, insbesondere weil die langfristige Kohlenstoffbindung eine Zunahme des langsam anfallenden SOC erfordert. Darüber hinaus wird gezeigt, dass insbesondere auf Ackerflächen die Kohlenstoffverteilung und der Kohlenstoffexport durch die Ernte überwacht werden müssen, um die langfristige SOC-Akkumulation oder -degradation vollständig zu verfolgen. Darüber hinaus wurde gezeigt, dass die Ernte der Baumstreifen die Kohlenstoffaufnahme von AF im darauffolgenden Jahr nicht systematisch beeinflusst, da die Baumstreifen nur einen relativ kleinen Teil des gesamten Ökosystems ausmachen.
Zusammenfassend konnte gezeigt werden, dass (1) die Netto-Kohlenstoffaufnahme auf Ökosystemebene in den Alleeanbausystemen mit Kurzumtrieb im Vergleich zu den angrenzenden Monokulturen deutlich höher war und dass (2) ein gut konzipiertes agroforstliches Alleeanbausystem im Kurzumtrieb die potenzielle Winderosion um mehr als 80% verringern könnte. Darüber hinaus wurde gezeigt, dass (3) die neuartigen, kostengünstigeren Eddy-Kovarianz-Anordnungen gut funktionieren und eine kostengünstigere Alternative zur teuren konventionellen Eddy-Kovarianz darstellen können. Schließlich sind weitere Verbesserungen der LC-EC-Leistung möglich, und die künftige Gestaltung und Wartung von AF bleibt ein Optimierungsprozess zwischen Vorteilen und Nachteilen auf individueller Standortebene, jedoch werden in der vorliegenden Arbeit einige konkrete Vorschläge unterbreitet.The world is more frequently experiencing extreme weather conditions due to climate change. Society and agriculture face more intense long-lasting warm and dry seasons, and extreme rainfall. These intensifying weather conditions ask for a sustainable intensification of agriculture, which is more resilient to droughts and erosion, so global food production can keep being provided. In addition, mitigating and reducing carbon-dioxide (CO2) and other greenhouse gas emissions can contribute to minimizing the effect of climate change. Agroforestry could be a good alternative for monocropping (MC) croplands or grasslands by providing more resilient and adapted agriculture for climate change and additional mitigation of CO2. In the current thesis, the focus is specifically on short rotation alley-cropping agroforestry systems (AF), which are croplands or grasslands in which widely-spaced parallel rows of trees are interleaved, creating natural alleyways. The more heterogeneous landscape of AF is known to provide a range of benefits such as, an altered microclimate which is more resilient for weather extremes, reduced soil erosion, improved biodiversity, and an improved carbon uptake. Despite that it is expected that wind erosion is reduced and the carbon uptake is improved over time with AF, wind erosion reduction has not been quantified and the carbon uptake over time is only measured with intermittent point measurements of the soil and partial harvest of trees. Therefore, the aims of the current thesis are to (1) compare and asses the continuous net ecosystem-scale carbon uptake of short rotation alley-cropping agroforestry and adjacent monocropping, and (2) quantify the potential wind erosion reduction by short rotation alley-cropping agroforestry. To be able to measure the ecosystem-scale carbon uptake at AF and MC, an additional aim was to (3) characterize and evaluate novel lower-cost eddy covariance (LC-EC) setups.
To achieve these aims, continuous eddy covariance (EC) measurements of the carbon exchange over four AF systems and four adjacent MC systems without trees were performed in northern Germany from mid 2019--2021. The EC measurements were performed using novel LC-EC setups, which were also tested and characterized in the current work. Furthermore, using simulated wind speed reduction output from large eddy simulations in combination with local wind speed measurements, potential wind erosion reduction estimates inside one entire short rotation alley-cropping agroforestry system were made.
In the first study of this thesis it was shown with simulated potential wind erosion estimates that an AF in a temperate climate could reduce the potential wind erosion by more than 80%, when the AF is well-designed and managed. Optimal wind erosion can be achieved when the tree strips are orientated perpendicular or diagonal to the main wind direction, reducing wind erosion with 92% and 86%, respectively. Besides, it was shown that the height of the tree strips needs to be in balance with the spacing in between the tree strips, and that wind erosion reduction with AF could result in reduced loss of soil organic carbon (SOC) compared to MC at locations with susceptible sandy soils. Finally, a distance of ≤48 m in between dense tree strips of only 2 m height, already provided a potential wind erosion reduction of 86% when orientated perpendicular or diagonal to the dominant wind direction.
In the second study of this thesis the tested and characterized LC-EC showed good agreement with simultaneous conventional eddy covariance (CON-EC) measurements. The LC-EC CO2 fluxes were 4–7% (R² = 0.91–0.95) lower compared to the CON-EC CO2 fluxes, and the latent heat (LE) fluxes also agreed well in 2020, as the LC-EC LE fluxes were 1–5% (R² = 0.86–0.91) higher than the CON-EC LE fluxes. In 2021, the LC-EC LE fluxes overestimated the CON-EC LE fluxes by 23% (R² = 0.84), however, likely due to underestimation of the CON-EC. Despite increased uncertainty with LC-EC due to using slower response sensors, the performance of LC-EC was comparable with the variation in EC fluxes between several CON-EC setups. Therefore, it was shown that the novel LC-EC setups are a cheaper alternative to CON-EC at approximately 25% of the costs.
In the third study of this thesis the continuous LC-EC measurements showed that the net ecosystem production (NEP) - or net carbon uptake - of the AF sites was significantly higher compared to the adjacent MC. The NEP at the AF croplands was on average 16--190% higher compared to MC croplands, and at the AF grassland the NEP was on average 195% higher compared to MC grassland. It was also shown that at the croplands the gross primary production (GPP) and ecosystem respiration (Reco) were between 10--35% and 9--15% higher at the AF compared to MC, respectively, and at the grassland the GPP and Reco were 13% and 31% lower at the AF compared to the adjacent MC, respectively. At the grassland site the difference in NEP between AF and MC was relatively constant for each year, however, the cropland sites showed strong yearly fluctuations, influenced by local weather conditions and soil and crop type. In addition, it is highlighted that long-term EC measurements could provide more detailed and solid understanding of the carbon balance differences between AF and MC sites, especially because long-term carbon sequestration requires an increase of the slowly accumulating SOC. Moreover, it was shown that especially at croplands carbon allocation and carbon export through harvest needs to be monitored to keep complete track on the long-term SOC accumulation or degradation. Furthermore, it was shown that the harvest of the tree strips did not systematically affect the carbon uptake of AF in the successive year, as the tree strips covered a relatively small part of the entire ecosystem.
The main conclusions of this thesis are that (1) the net ecosystem-scale carbon uptake of the short rotation alley-cropping systems was significantly higher compared to the adjacent monocropping sites, and that (2) a well-designed short rotation alley-cropping agroforestry systems can provide more than 80% reduction of potential wind erosion. Furthermore, it was shown that (3) the novel lower-cost eddy covariance setups performed well and can be a cheaper alternative to costly conventional eddy covariance setups. Finally, further improvements of the LC-EC performance are possible, and future design and maintenance of AF remains an optimization process between benefits and drawback at individual site level, however some concrete suggestions are provided in the current thesis.2025-08-1