The expanding urbanisation, growing population, and industrial development are threatening global surface water quality. With increasing concern about surface-water quality, it is crucial to deeply understand the evolution of surface-water quality problems and comprehensively de-termine its fundamental driving forces. In this Dissertation, systematic work on the mechanisms of water pollution with trace elements has been carried out in three steps: i) to identify the sources contributing to surface water pollution by receptor-based models, ii) to determine the factors dominating the pollution risk transmission from sources to surface water by a source-based model, and iii) to capture the primary driving forces to the spatiotemporal variation in surface water pollution by Bayesian-based approaches. The following specific topics were ad-dressed based on five publications: a) The temporal trends of trace metal pollution in the surface water were characterised by the Mann-Kendall test and the Generalised Additive Model. b) The primary source contributors to the long-term trace metal pollution in a river system were determined by the Self-organised Map, Positive Matrix Factorization receptor model, and Bayesian multivariate receptor model. The distributions of the source contributions to trace metal pollution were estimated. c) The risk transmission of trace pollutants in the surface water was estimated by a source-based dynamic model. The sensitivities of the risk to human activities, characteristics of wastewater treatment plants, and river flow regimes were evaluated. d) The contributions of hydro-chemical factors, climate impact, and sampling methods to water pollution and data uncertainty were analysed by the Wavelet Analysis and Bayesian Net-work. Both the models’ accuracy and robustness were evaluated by statistical analysis. The methods and results provided herein could improve the standard of statistical rigour and support the authorities’ decision-making.

Riverbank filtration (RBF) is defined as the infiltration of river water into the aquifer. It can either be induced by pumping or as a result of a losing river stretch. It is used worldwide as a pre-treatment technique for surface water, which allows for lower abstraction of groundwater and usually an increase in the infiltrated water (bank filtrate) quality. In Vietnam, RBF offers a potential alternative to reduce groundwater overexploitation and an effective pre-treatment for the high turbidity characteristic of Vietnam’s rivers. However, a few water quality problems may arise during RBF, mainly caused by the development of reducing conditions during infiltration, which may trigger in particular conditions the mobilization of manganese, iron, arsenic and ammonium. Arsenic is typically a regional problem in groundwater/bank filtrate and therefore not an issue in the majority of RBF sites. However, the Red River Delta (RRD) in north Vietnam is one of the regions affected by geogenic arsenic contamination. Arsenic alongside with ammonium, are two of the biggest water quality issues in the RRD and are therefore the main points of investigation of this thesis. Ammonium is an undesirable substance in the abstracted water of RBF schemes, due mainly to the complications it causes during post-treatment. A literature review of case studies from 40 RBF sites around the world was done. Typical concentrations of ammonium in the bank filtrate are between 0.1 and 1.7 mg/l. The most common source of ammonium in the bank filtrate was found to be the mineralization of organic nitrogen occurring in the riverbed, while the most common sink of ammonium was nitrification in the riverbed. Ammonium surface water concentrations do not directly translate to abstracted concentrations. Transformations in the riverbed play a critical role in determining ammonium concentrations, whereby riverbeds with high amounts of organic material in the riverbed lead to higher mineralization rates and have more electron donor competitors, thus limiting ammonium attenuation via nitrification. Column experiments with riverbed sediments and river water from the Elbe were performed to evaluate the controls on ammonium concentrations during riverbed infiltration. Concentrations of ammonium went from <0.1 mg/l in the feed water up to 1 mg/l in the columns effluent. Higher temperatures and lower infiltration rates led to increased ammonium concentrations in the effluent. This shows higher susceptibility to ammonium increases of RBF settings in warmer climates, such as in Vietnam. In the later phases of the experiments, after the columns have been flushed their pore volumes several times, ammonium concentrations continually decreased. This behavior was attributed to the partial consumption of easily degradable organic material in the sediments, leading to lower mineralization rates. Based on operation with varied nitrate concentrations and 15N isotopic measurements, dissimilatory nitrate reduction to ammonium (DNRA) was not shown to be relevant in the formation of ammonium. Anaerobic ammonium oxidation (anammox), however, was hypothesized to be an important sink of ammonium inside the columns, which indicates that rivers with high nitrate concentrations, such as the Elbe, may have a buffer of protection against ammonium formation during RBF. Geogenic arsenic contamination of groundwater poses a health threat to millions of people worldwide, particularly in Asia. A study in Hanoi was made, where the water quality of 48 RBF wells from 5 large well fields located in the Pleistocene aquifer along the Red River was analyzed. More than 80% of the wells had arsenic concentrations above the WHO limit of 10 µg/l. The riverbed sediment and riverbed pore-water from 23 sites along a stretch of 30 km of the Red River near the well fields was also analyzed. Muddy riverbeds were found to be a hotspot for arsenic release. Already at a 30 cm depth from the riverbed sediment surface, the pore-water at many sites had high concentrations of arsenic (>100 µg/l). Arsenic concentrations in the pore-water of sites where mud layers were present in the riverbed were significantly higher compared to sites with sandy riverbeds. At well fields along stretches of the Red River where riverbed was mostly muddy, higher arsenic concentrations were found than at well fields where the riverbed was mostly sandy. This indicates that river muds deposition and river morphology can influence arsenic concentrations in the aquifer in Hanoi and potentially other RBF sites in regions with geogenic arsenic contamination. In summary, processes in the hyporheic zone were assessed to be of key importance to the behavior of both ammonium and arsenic during RBF. Further investigations in two different regions of Vietnam were conducted, in the Red River Delta (RRD) and in the Binh Dinh province (central Vietnam). The results from the geological and water quality investigations are presented. The deltaic depositional environment of the RRD presents hydrogeological challenges to the application of RBF. Thick loamy layers overlaying the topmost aquifer can prevent hydraulic connection between river and aquifer, which is the principal pre-condition for RBF. Furthermore, aquitards separating the aquifers make RBF by abstraction in the deeper aquifer(s) not possible. In Binh Dinh, on the other hand, there are rarely clay layers separating the aquifers and often the topmost aquifer is close to the surface and has high hydraulic conductivity, which makes bank filtration application in Binh Dinh very cost efficient. Water quality of the bank filtrate in Binh Dinh was also generally high, especially in comparison with the RRD. / Unter Uferfiltration (UF) wird die Infiltration von Flusswasser in einen Grundwasserleiter verstanden. Die UF kann durch eine Wasserentnahme durch Brunnen in Gewässernähe induziert werden oder natürlich an einem infiltrierenden Flussabschnitt erfolgen. Die UF wird weltweit als Voraufbereitung für Oberflächenwasser eingesetzt, zur Verringerung von Grundwasserentnahmen und normalerweise zur Verbesserung der Qualität des infiltrierten Wassers (Uferfiltrats). In Vietnam ist die UF eine potenzielle Alternative zur Verringerung der Grundwasserübernutzung und als effektive Voraufbereitung für das für vietnamesische Flüsse charakteristische stark trübstoffhaltige Wasser. Allerdings kann es bei der UF zu einigen Probleme mit der Wasserqualität kommen, hauptsächlich durch die Entwicklung reduzierender Bedingungen während der Infiltration, die unter bestimmten Bedingungen zur Mobilisierung von Mangan, Eisen, Arsen und Ammonium führen können. Arsen ist in der Regel nur regional im Grundwasser/Uferfiltrat zu finden und daher an den meisten UF-Standorten kein Problem. Das Red River Delta (RRD) in Nordvietnam ist jedoch eine der Regionen mit geogen bedingter Arsenkontamination. Hohe Arsen- und Ammoniumkonzentrationen sind zwei der größten Probleme für die Wasserqualität im RRD und deshalb die Schwerpunkte dieser Arbeit. Ammonium ist eine unerwünschte Substanz im Rohwasser von UF-Anlagen, vor allem wegen der Schwierigkeiten bei der Trinkwasseraufbereitung. Im Rahmen einer Literaturrecherche wurden Fallstudien von 40 UF-Standorten in der ganzen Welt ausgewertet. Im Uferfiltrat sind Ammoniumkonzentrationen zwischen 0,1 und 1,7 mg/l typisch. Als häufigste Quelle für Ammonium im Uferfiltrat wurde die Mineralisierung von organischem Stickstoff im Flussbett ermittelt und als häufigste Senke die Nitrifikation in der Flusssohle. Die Ammoniumkonzentrationen im Oberflächenwasser können nicht direkt mit denen im Rohwasser gleichgesetzt werden. Stoffumwandlungsprozesse in der Flusssohle spielen eine entscheidende Rolle bei der Veränderung der Ammoniumkonzentrationen, wobei Flusssohlsedimente mit einem hohen Anteil an organischem Kohlenstoff zu höheren Mineralisierungsraten führen und mehr Elektronendonatoren als Konkurrenten für die Verringerung der Ammoniumkonzentration durch Nitrifikation aufweisen. Um die die Ammoniumkonzentration bei der Infiltration bestimmenden Prozesse zu analysieren, wurden Säulenversuche mit Flusssohlsedimenten aus Buhnenfeldern an der Elbe und Elbewasser durchgeführt. Die Ammoniumkonzentrationen reichten von <0,1 mg/l im Zufluss bis zu 1 mg/l im Ablauf der Säulen. Höhere Temperaturen und niedrigere Infiltrationsraten führten zu höheren Ammoniumkonzentrationen im Ablauf. Dies zeigt, dass in wärmeren Klimazonen, wie z. B. in Vietnam, die Zunahme der Ammoniumkonzentration bei der UF wahrscheinlicher ist. In der Endphase der Versuche nahm die Ammoniumkonzentration kontinuierlich ab, nachdem das Porenvolumen der Säulen mehrfach ausgetauscht war. Dieses Verhalten wurde auf den teilweisen Verbrauch von leicht abbaubarem organischem Material im Sediment zurückgeführt, der dann zu geringeren Mineralisierungsraten führte. Durch die Beschickung der Säulen mit Wasser unterschiedlicher Nitratkonzentrationen und Messungen von 15N-Isotopen wurde gezeigt, dass die dissimilatorische Nitratreduktion zu Ammonium (DNRA) für die Bildung von Ammonium nicht relevant ist. Es wurde jedoch die Hypothese formuliert, dass die anaerobe Ammoniumoxidation (Anammox) eine wichtige Senke für Ammonium in den Säulen ist, so dass es in Flüssen mit erhöhten Nitratkonzentrationen, wie z. B. der Elbe, zu einer Begrenzung der Ammoniumfreisetzung bei der UF kommt. Die geogene Arsenkontamination des Grundwassers stellt eine gesundheitliche Bedrohung für Millionen von Menschen weltweit dar, insbesondere in Asien. In Hanoi wurde eine Studie durchgeführt, bei der die Wasserqualität von 48 UF-Brunnen aus 5 großen Brunnenfeldern im pleistozänen Aquifer entlang des Red River analysiert wurde. Bei mehr als 80 % der Brunnen lag die Arsenkonzentration über dem WHO-Grenzwert von 10 µg/l. Außerdem wurden das Flusssohlsediment und das Porenwasser in der Flusssohle an 23 Standorten entlang eines 30 km langen Abschnitts des Red River in der Nähe der Brunnenfelder untersucht. Es wurde festgestellt, dass Flussabschnitte mit Schlammablagerungen ein Hotspot für die Freisetzung von Arsen sind. Bereits in 30 cm unter der Flusssohle wies das Porenwasser an vielen Stellen hohe Arsenkonzentrationen auf (>100 µg/l). Die Arsenkonzentrationen im Porenwasser war an Standorten mit Schlammablagerungen bzw. Schlammschichten in der Flusssohle deutlich höher als an Standorten mit sandigen Flusssohlsedimenten. In Brunnenfeldern entlang von Abschnitten des Red River, in denen die Flusssohle mehr organisches Feinmaterial enthielt, wurden höhere Arsenkonzentrationen festgestellt als in Brunnenfeldern, die Uferfiltrat aus Abschnitten mit sandiger Flusssohle gewinnen. Dies zeigt, dass Schlammablagerungen und die Flussmorphologie die Arsenkonzentration im oberen Grundwasserleiter in Hanoi und möglicherweise an anderen UF-Standorten in Regionen mit geogener Arsenkontamination beeinflussen. Zusammenfassend wurde festgestellt, dass Prozesse in der hyporheischen Zone für das Verhalten von Ammonium und Arsen bei der UF von entscheidender Bedeutung sind. Weitere hydrogeologische und hydrochemische Untersuchungen zur UF wurden in zwei verschiedenen Regionen Vietnams durchgeführt, im RRD und in der Provinz Binh Dinh (Zentralvietnam). Die Ablagerungsbedingungen im RRD bestimmen die mögliche Nutzung der UF. Mächtige lehmige Deckschichten, die den obersten Grundwasserleiter überlagern, können eine hydraulische Verbindung zwischen dem Fluss und dem Grundwasserleiter verhindern, welche Hauptvoraussetzung für die UF ist. Darüber hinaus machen geringdurchlässige Schichten (Aquitards) eine UF durch Wasserentnahmen aus tieferen Grundwasserleitern unmöglich. Dagegen gibt es in Binh Dinh kaum Aquitards, die die Grundwasserleiter voneinander trennen, oft hat der oberflächenahe, oberste Grundwasserleiter eine hohe hydraulische Durchlässigkeit, was die Nutzung der UF in Binh Dinh kosteneffizient macht. Auch die Qualität des Uferfiltrats war in Binh Dinh generell höher, insbesondere im Vergleich zum RRD.

Urban regeneration plays a significant role in the creation of climate-friendly urban areas. Urban regeneration strategies such as (1) the creation of open spaces and green infrastructure, (2) the effective use of inner-city land, (3) changes in land-use structure and elements through regeneration of inner-city areas, and (4) changes in building stock through retrofitting existing buildings and the construction of green buildings can all help with climate mitigation and adaptation efforts. Many cities throughout the world have enacted urban regeneration policies, plans, and programmes, highlighting the significance of implementing climate measures into urban regeneration policies and initiatives at the supranational, national, regional, and municipal levels. The government of the Republic of Korea has made urban regeneration a primary priority, investing a significant amount of public funds in the process. Korea's urban regeneration initiative has made tremendous progress, with projects in numerous cities around the country currently underway. However, Korea's urban regeneration rarely takes up the opportunities to incorporate climate measures into its policy and plans. As a result of these fragmented policy approaches, both urban regeneration and climate change policies may become inefficient in terms of inefficient budget execution and staff management. Although recent Korean studies have emphasised the significance of incorporating climate change measures into urban regeneration areas, these studies do not provide empirical evidence of specific institutional hurdles and only provide a limited amount of recommendations for successful climate policy integration in urban regeneration policies and plans. According to the literature on environmental policy integration and climate mainstreaming in worldwide academic discussion, there is a lack of empirical information on the factors that influence policy implementation phases. Identification of variables influencing the integration of climate change policies into urban regeneration—at both the policy development and implementation stages—is crucial to understanding effective climate policy integration in this policy domain. The goals of this dissertation are to (1) investigate the current urban regeneration process and how it incorporates climate measures, (2) identify the relevant internal and external factors that influence the integration of climate measures into urban regeneration policy during the policy development and implementation stages, and (3) identify implementation gaps in the integration of climate measures into urban regeneration projects during the policy development and implementation stages. This study develops a conceptual framework based on earlier studies that examine (1) the assessment of climate policy integration levels and (2) the factors that influence climate policy integration across policy domains. This dissertation takes a qualitative case study method, integrating content analysis, process tracing, and document analysis with data from semi-structured interviews with relevant stakeholder representatives, as well as a collection of relevant documents. Seoul was chosen as the case study area because it is a densely populated city with active urban regeneration and climate change policies. The case study provides empirical evidence of relevant factors for the climate policy integration in urban regeneration for cities that are pursuing successful integration of climate measures into urban regeneration policy. Throughout the policy cycle, this study gives a detailed list of internal and external factors that influence the incorporation of climate measures into urban regeneration. Political factors, organisational factors, and resources are critical factors in both policy development and policy implementation, as previous studies have shown. However, in the policy development and policy implementation stages, this dissertation studies more detailed aspects within these categories and examines them closely by categorising them into internal and external factors. According to existing research in this field, implementation gaps are caused by a lack of sustained political support and cooperation among key stakeholders, rather than a lack of knowledge or financial resources. Other factors related to implementation gaps, such as (1) a lack/absence of information about climate measures, (2) residents' pecuniary focus, (3) public awareness and support, and (4) spatial issues such as lack of facilities/spaces for adopting climate measures, are all significant impacts in the case study of Seoul. This study suggests that sharing information and research about the effectiveness of climate measures is a good place to start when it comes to improving the level of climate policy integration in urban regeneration policies; urban regeneration stakeholders should have enough information on potential climate measure strategies that can be integrated into urban regeneration projects, as well as the benefits of doing so for the neighbourhood (e.g. climate-related businesses which support community cooperation and profit generation, community-based climate activities, and more diverse ways of measuring the success of climate measures in order to educate the public better).:1. Introduction 1.1. Research background 1.2. Research questions and design 1.3. Structure 2. Theoretical/conceptual background 2.1. Responses to climate change 2.2. Concept of urban regeneration 2.3. Background of environmental/climate policy integration and mainstreaming 2.4. International trends and academic discussions on the opportunities for climate policy integration in urban regeneration 2.5. Factors for the climate policy integration in urban regeneration in the policy cycle 2.5.1. Structure of the climate policy integration 2.5.2. Detailed factors of the climate policy integration 2.5.3. The level of climate policy integration in different stages of the policy cycle 2.6. Developing an analytical concept 2.7. Research gaps regarding factors affecting climate policy integration along the policy cycle 2.8. Formulating research questions 3. Research design and methods 3.1. Research design 3.2. Selection of study areas 3.3. Data collection and analysis methods 4. Exploring the context: Seoul and its neighbourhoods 4.1. Seoul 4.1.1. Features of Seoul and evolution of urban regeneration in Seoul 4.1.2. Climate change issues 4.1.3. Characteristics of residential areas in Seoul 4.2. Features of study areas in Seoul 4.2.1. Jangwi-dong 4.2.2. Sangdo 4-dong 4.2.3. Amsa-dong 4.2.4. Garibong-dong 5. Urban regeneration and climate change policies in Korea 5.1. Urban regeneration policy and plan at national, city, and community levels 5.1.1. Urban regeneration policy in Korea 5.1.2. Urban regeneration policy and plan in Seoul 5.1.3. Neighbourhood urban regeneration plans 5.1.4. Process and stakeholders of urban regeneration at national, city, and community levels 5.2. Climate change policy at different levels in Korea 5.2.1. Climate change policy at national level 5.2.2. Climate change policy at city level 5.2.3. Climate change plan at community level 5.3. Integration of climate measures in the process of urban regeneration 5.3.1. Level of the integration of climate measures in urban regeneration policy at national level 5.3.2. Level of the integration of climate measures in urban regeneration policy at city level 5.3.3. Level of the integration of climate measures in urban regeneration plan at community level 5.4. Summary 6. Policy development: Internal and external factors to integrate climate measures into urban regeneration policy 6.1. Internal factors 6.1.1. Political factors 6.1.2. Organisational factors 6.1.3. Resources 6.1.4. Cognitive factors 6.1.5. Characterisation of the problem at hand 6.2. External factors 6.2.1. Public awareness and support 6.2.2. Lack of private sector support 6.3. Relevant factors and gaps regarding factors in the conceptual framework 6.4. Discussion 7. Policy implementation: internal and external factors to integrate climate measures into urban regeneration projects 7.1. Internal factors 7.1.1. Political factors 7.1.2. Organisational factors 7.1.3. Resources 7.1.4. Cognitive factors 7.1.5. Characterisation of the problems/opportunities at hand 7.2. External factors 7.2.1. Residents’ support 7.2.2. Characterisation of the problem at hand 7.2.3. Lack of private sector support 7.2.4. Cognitive factors 7.3. Relevant factors and gaps regarding factors in the conceptual framework 7.4. Discussion and implementation gaps between policy development and policy implementation 8. Conclusions: recommendations to enhance the level of integration of climate measures into urban regeneration 8.1. Synthesis of the dissertation 8.2. Limitations of the research and further research 8.3. Implications for academic discussion and practices 8.4. Recommendations 8.5. Overall conclusion References Appendix

The number of regions undergoing water scarcity, where the quantity of available water is not enough to meet human demand, is expected to increase in the future. Water reuse measures have been widely implemented to face these situations as a means of increasing the supply of water resources. Thus, ‘water scarcity – water reuse’ (WS-WR) situations will likely become more common. In these cases, water resources management to secure enough water supply is key. Risk and sustainability concepts have been consolidated as guiding discourses that also support the management of water resources. In particular, in the case of WS-WR situations, they can guide decision-makers towards reducing the risk of water scarcity and striving for the implementation of sustainable water reuse measures. In particular, the use of risk and sustainability assessments helps to deal with various social, economic, and environmental requirements and constraints. However, there is still the call for a more comprehensive and integrated assessments. This dissertation aims at providing new ideas for the integration of risk and sustainability in the case of WS-WR situations. Three objectives guide this research: (A) to develop a conceptual assessment framework to support decision-making concerning sustainable water reuse in regions facing risk of water scarcity; (B) to advance the conceptual framework interrelating existing risk and sustainability assessment methodologies and indicators in the context of decision support; and (C) to test the conceptual and methodological framework using a case study in Latin America. Each objective is associated with a research question: (RQ1) How is decision-making regarding water reuse understood and supported towards reducing the risk of water scarcity sustainably – and how can it be represented in a conceptual assessment framework?; (RQ2) How can a conceptual framework for assessing water reuse as sustainable water scarcity risk reduction measures be operationalised through a methodological framework?; and (RQ3) What are the findings from testing the framework in a case study – and what can be incorporated into the framework? Each objective and its respective research question was addressed as a separate step of the research approach, comprising the development of an integrated Risk and Sustainability Assessment (RSA) Framework for WS-WR situations, its operationalisation and testing. The research approach followed a deductive to inductive rationale relying on qualitative and quantitative methods. The outputs of this research are three scientific publications that build this cumulative dissertation (two published and one submitted for revision). The development of the conceptual framework followed three steps: (i) defining the concepts of ‘water scarcity’, ‘water reuse’, ‘risk’ and ‘risk assessment’, ‘sustainability’ and ‘sustainability assessment’, and ‘decision-making’; (ii) integrating these concepts by interpreting water scarcity from a risk perspective and water reuse from a sustainability perspective, and relating assessments with decision-making; and (iii) structuring the RSA Framework, following a risk assessment and framing it by the social, economic, and environmental dimensions of sustainability. Results allowed defining decision-making in WS-WR situations as a four-step cyclic process that can be supported by an integrated RSA that comprises an analysis (descriptive and objective) and evaluation (subjective). The methodological aspects for the operationalisation of the RSA conceptual framework focused mainly on developing an analytical concept to support an adequate derivation of the information required in an integrated RSA for WS-WR situations. The resulting concept is based on (i) understanding the WS-WR situation as a Coupled Human and Natural System (CHANS) and identifying the main biophysical elements (endpoints); (ii) translating the CHANS endpoints into an information system via a Multi-Layer (ML) approach using generic descriptors and specific indicators; and (iii) identifying and characterising interlinkages between the indicators via a Lane-Based (LB) approach. Additional methodological aspects related to the evaluation include the use of indicator-based multi-criteria decision-making methods that include the weighting and aggregation of these indicators, as well as the selection of threshold values as evaluation criteria. The testing of the integrated RSA Framework was carried out in Cerrillos de Tamaya, Chile. It involved an ex-post RSA of a water reuse measure implemented in 2018 to face the local water scarcity situation. The testing included (i) describing the case study location and adapting the RSA Framework to fit the local context; (ii) translating the case study’s CHANS via the ML approach and identifying and characterising interlinkages via the LB approach; and (iii) evaluating the degree of risk of water scarcity and sustainability of water reuse via the distance-based method TOPSIS. The results of the testing provided feedback for the RSA Framework. These mainly referred to the influence of the conceptualisation behind the indicators and their use, and the methodological challenges for integrating risk and sustainability evaluation. Further recommendations to the RSA framework are: the inclusion of interlinkage directionality; the use of existing system dynamics modelling approaches (e.g., CLD, SFD); the development of an established database of indicators; the automation of the interlinkages analysis (LB approach); and advance the use of scenarios for sustainability evaluation for better coupling with risk evaluation methods. Overall this research provides evidence of (a) the conceptual integration of risk and sustainability discourses under one decision support framework for the case of WS-WR situations; (b) the use of a system thinking approach for interpreting the WS-WR situation; (c) the relevance of indicators as a means of representing the situation; (d) the interlinkage of social, economic, environmental information; (e) the benefits of the use of conceptual maps; (f) gaps in the process of measuring the effect of water reuse on water scarcity levels via indicators; (g) the gap between a simulation-based risk assessment and a snapshot-focused sustainability assessment that hinders an operational integration; (h) the possibility of the RSA framework to bridge a system thinking view with a traditional assessment-based decision-making view.:Acknowledgements Abstract Contents List of Figures List of Tables Acronyms and Abbreviations Symbols Chapter 1 - Introduction 1.1 Background and problem statement 1.1.1 Water resources for water security 1.1.2 Risk and sustainability discourses for water-related decision-making 1.1.3 Problem statement and research focus 1.2 Objectives and research questions 1.3 Research approach and structure of the document 1.3.1 Research approach 1.3.2 Structure of the document 1.4 Chapter references Chapter 2 - Conceptual Framework 2.1 Introduction 2.2 Developing the conceptual framework 2.2.1 Definition and interpretation of the subject at stake 2.2.2 Identification and definition of key concepts 2.2.3 Construction of the conceptual framework 2.3 Results and discussion 2.3.1 Defining and interpreting the ‘water scarcity – water reuse’ situation 2.3.2 Identifying and defining key concepts 2.3.3 Construction of the integrated RSA Framework 2.4 Conclusions and outlook 2.5 Acknowledgements 2.6 Chapter references Chapter 3 - Methodological Aspects 3.1 Introduction 3.2 RSA Framework for a WS-WR situation 3.3 Systems thinking in a WS-WR situation 3.3.1 Identifying elements of a WS-WR situation and its interpretation as a system 3.3.2 Translation of the CHANS into an information system 3.4 Characterisation and interlinkage of indicators 3.4.1 Type and number of indicators 3.4.2 Type and number of interlinkages 3.4.3 Indicator connectivity 3.4.4 Structuring via a lane-based approach 3.5 RSA analytical concept and exemplification 3.5.1 RSA analytical concept 3.5.2 Exemplification of the analyitical concept 3.6 Discussion 3.6.1 Translating the CHANS into an information system 3.6.2 Supporting decision-making via the analytical concept 3.7 Conclusions 3.8 Acknowledgements 3.9 Chapter references Chapter 4 - Framework Testing 4.1 Introduction 4.2 Approach 4.2.1 RSA Framework 4.2.2 Case study site 4.3 Results 4.3.1 Analysis 4.3.2 Evaluation 4.3.3 General results for the case 4.4 Discussion 4.4.1 Analysis 4.4.2 Evaluation 4.4.3 Overall discussion on the testing of the RSA Framework 4.5 Conclusions 4.6 Acknowledgements 4.7 Chapter References Chapter 5 - Synthesis 5.1 Conceptual aspects 5.2 Methodological aspects 5.3 Testing aspects 5.4 Placing the RSA Framework in a broader context 5.5 Chapter References Chapter 6 - Conclusions and Outlook Annexes Annex A - Literature review: Found records Annex B - Example list of endpoints, descriptors, indicators, and attributes Annex C - Technique for Order Preference by Similarly to Ideal Solution (TOPSIS) Annex D - Translation into the Information System (from endpoints to attributes) Annex E - Interlinkages Identification Matrix Annex F - List of Most Interlinked Indicators (MII) Annex G - List of indicators, scores, and thresholds

The global increase in demand for wood products, calls for a more sustainable management of forests to optimize both the production of wood and the conservation of forest biodiversity. In this paper, we evaluate the status and future trends of forest birds in Central European forests, assuming different forest management scenarios that to a varying degree respond to the demand for wood production. To this end, we use niche models (Boosted Regression Trees and Generalized Linear Models) to model the responses of 15 forest bird species to predictors related to forest stand (e.g., stand volume of specific tree species) and landscape structure (e.g., percentage cover), and to climate (bioclimatic variables). We then define five distinct forest management scenarios, ranging from set-aside to productivity-driven scenarios, project them 100 years into the future, and apply our niche models into these scenarios to assess the birds’ responses to different forest management alternatives. Our models show that the species’ responses to management vary reflecting differences in their ecological niches, and consequently, no single management practice can benefit all species if applied across the whole landscape. Thus, we conclude that in order to promote the overall forest bird species richness in the study region, it is necessary to manage the forests in a multi-functional way, e.g., by spatially optimizing the management practices in the landscape.

Nitrogen-, sulfur-, and oxygen-substituted heterocyclic polyaromatic hydrocarbons (heterocyclic PAHs) are ubiquitous in all environmental compartments, often co-occurring with their unsubstituted carbocyclic analogues. However, they have received little attention in the hazard assessment of contaminated sites and little is known about their fate and impacts. Due to their higher polarity resulting from the presence of electronegative atoms possessing lone electron pairs, most heterocycles are expected to be less toxic and bioaccumulative than their homocyclic counterparts, which is perhaps why research on their hazard and fate is much more limited. If this assumption is not true (e.g., because factors other than polarity/hydrophobicity drive their hazard), this may lead to an underestimation of the risks posed by contaminated sites or emissions. Therefore, the main objectives of this thesis are (1) to improve the understanding of the physicochemical properties, biodegradability, ecotoxicity, bioaccumulation, and mobility of these compounds in environmental contexts, (2) to evaluate the effectiveness of different modeling approaches in accurately predicting their properties, and (3) to tackle experimental barriers by designing robust techniques that can produce reliable and reproducible data output in the long run. From the large family of heterocycles, structurally comparable representative compounds with log KOW values (as a measure of hydrophobicity) ranging over four orders of magnitude were selected as model structures. First, the most environmentally relevant physicochemical properties of heterocyclic PAHs were investigated, including water solubility (SW), n-octanol-water partition coefficient (KOW), and organic carbon-water partition coefficient (KOC). The effect of molecular size, type and number of heteroatoms on the solubility and partition coefficients was investigated. These properties were then used to test the performance of in silico models, including Quantitative Structure-Property Relationship (QSPR), Polyparameter Linear Free Energy Relationship (pp-LFER), and Conductor-like Screening Model for Realistic Solvation (COSMO-RS), that predict these properties. A screening-level environmental hazard assessment was performed by integrating the data and evidence from multiple sources. The screening results revealed (i) high hazard for 4- and 5-ring compounds mainly due to high persistence and bioaccumulation potential, (ii) limited availability of experimental data, and (iii) notable uncertainty due to the limited applicability domain of the models, stressing the urgent need for a detailed hazard assessment for this group of compounds. The persistence of heterocyclic PAHs in the environment was assessed experimentally through a series of biodegradation tests under aerobic conditions. These included ultimate biodegradation, primary biodegradation and inoculum toxicity tests at different exposure concentrations. None of the test substances (≥3 rings) were found to be readily biodegradable. Therefore, to stimulate the biodegradation process, the microbial community from a wastewater treatment plant was adapted to a mixture of heterocyclic PAHs and biodegradation tests were performed with either adapted or non-adapted bacteria. Surprisingly, there were trade-offs in bacterial adaptation: a positive outcome (degradation of carbazole) is accompanied by a less desirable outcome (increased sensitivity to toxic effects of benzo[c]carbazole). Four of the compounds were identified as primarily biodegradable, while five compounds showed no evidence of biodegradation in any of the tests, indicating a high potential for environmental persistence. Predictions from mathematical models were also compared with measured results and it was found that the models were only partially successful in predicting the degradation timeframe of the heterocycles tested. Furthermore, the short-term acute aquatic toxicity of heterocyclic PAHs was investigated using three surrogate species from different trophic levels of the aquatic food web, including bacteria (Aliivibrio fischeri), unicellular algae (Raphidocelis subcapitata), and water fleas (Daphnia magna). In order to maintain a stable exposure and to avoid loss of concentration caused mainly due to sorption, the passive dosing method was adapted to routine test protocols. Most of the compounds were identified as highly toxic to the tested organisms, and non-polar narcosis (baseline toxicity) was found to be the most likely mode of toxic action for the tested chemicals. Membrane-water partition coefficients (KMW) of heterocyclic PAHs were also determined by solid-supported lipid membrane binding experiments and modeling. The measured KMW values were closely aligned with the KOW values, suggesting that both may be good descriptors for predicting the baseline toxicity and bioaccumulation potential of the tested heterocyclic PAHs. Lastly, the long-term chronic reproductive toxicity and bioaccumulation of large heterocyclic PAHs were investigated using Daphnia magna as an aquatic model organism. The passive dosing method ensured constant exposure even at very low concentration levels (70 ng L-1 to 68 µg L-1) in large and complex experimental systems. Given the enormous time and effort required to conduct long-term toxicity tests with minimal data output, a method was developed to assess chronic toxicity and bioaccumulation potential in a single test. All four substances tested were highly toxic and bioaccumulative in Daphnia magna. However, daphnids’ fecundity recovered rapidly from the toxic effects of the heterocycles during the depuration period in the absence of chemicals. The Predicted No Effect Concentrations (PNECs) of the test compounds were also estimated from the chronic toxicity data and ranged from 1 to 150 ng L-1 in a freshwater environment. Due to the lack of regulation and attention to these compounds, no systematic environmental monitoring data are available for their concentrations in freshwater. However, environmental concentrations of heterocyclic PAHs appear to be higher than the PNECs, particularly at contaminated sites. Thus, the tested heterocycles are of high concern in terms of risks to ecosystems and human health. Overall, the results of this research not only contribute to a comprehensive understanding and progress in the environmental hazard assessment of heterocyclic PAHs, but also present several methodological advances that can be applied to future testing of other challenging substances commonly referred to as 'difficult-to-test'. / Stickstoff-, schwefel- und sauerstoffsubstituierte heterocyclische polyaromatische Kohlenwasserstoffe (heterocyclische PAK) sind in allen Umweltkompartimenten gegenwärtig und treten häufig zusammen mit ihren homocyclischen Analoga auf. Bei der Gefährdungsbeurteilung kontaminierter Standorte wurde ihnen bisher verhältnismäßig wenig Aufmerksamkeit geschenkt, zudem ist über ihr Verhalten und ihre Ökotoxizität nur wenig bekannt. Aufgrund ihrer höheren Polarität, die auf das Vorhandensein elektronegativer Atome zurückzuführen ist, ging man bislang davon aus, dass die meisten heterocyclischen PAK mobiler und weniger toxisch sind als ihre homocyclischen Pendants, weshalb sie wahrscheinlich bislang vergleichsweise wenig studiert wurden. Sollte die Annahme einer geringeren Toxizität nicht zutreffen, weil beispielsweise andere Mechanismen als die Polarität/Hydrophobie für ihre Gefährdung verantwortlich sind, könnte dies zu einer Unterschätzung der Risiken, die von kontaminierten Standorten oder anderen Emissionen ausgehen, führen. Die Hauptziele dieser Arbeit bestehen daher darin, (1) ein umfassenderes Verständnis der physikalisch-chemischen Eigenschaften, der biologischen Abbaubarkeit, der Ökotoxizität, der Bioakkumulation und der Mobilität dieser Verbindungen in der Umwelt zu erlangen; (2) eine Bewertung verschiedener Modellierungsansätze zur Vorhersage dieser Eigenschaften durchzuführen und (3) experimentelle Hindernisse zu überwinden, indem robustere Techniken entwickelt werden, die langfristig zuverlässige und reproduzierbare Daten liefern können. Aus der sehr breiten Gruppe der heterocyclischen Verbindungen wurden strukturell vergleichbare repräsentative Verbindungen mit unterschiedlichen log KOW-Werten (als Maß für die Hydrophobie), die vier Größenordnungen abdecken, als Modellstrukturen ausgewählt. Zunächst erfolgte eine Untersuchung umweltrelevanter physikalisch-chemischer Eigenschaften dieser heterocyclischen PAK, wie die Wasserlöslichkeit (SW), der n-Octanol-Wasser-Verteilungskoeffizient (KOW) und der Verteilungskoeffizient bezogen auf den organischen Kohlenstoff und Wasser (KOC). Dabei wurden die Auswirkungen von Molekülgröße, Art und Anzahl der Heteroatome auf die Löslichkeit und die Verteilungskoeffizienten systematisch untersucht. Diese experimentell ermittelten Parameter wurden dann verwendet, um die Leistung von In-silico-Modellen zu testen, einschließlich der quantitativen Struktur-Eigenschafts-Beziehung (QSPR), der linearen Polyparameter-Energie-Beziehung (pp-LFER) und des Conductor-like Screening Model for Realistic Solvation (COSMO-RS), die diese Eigenschaften vorhersagen. Zudem wurde eine Umweltverträglichkeitsprüfung auf Screening-Ebene durchgeführt, bei der Daten aus verschiedenen Quellen berücksichtigt wurden. Den Screening-Ergebnissen zufolge ist davon auszugehen, dass kleine Verbindungen mit zwei oder drei Ringen in der Umwelt mobil sind, während größere heterocyclische PAK im Allgemeinen als bioakkumulierbare Stoffe identifiziert wurden, was auf ein erhebliches Gefahrenpotenzial hinweist. Die Persistenz von heterocyclischen PAK wurde experimentell durch eine Reihe von Tests zum biologischen Abbau unter aeroben Bedingungen untersucht. Dazu gehörten Tests zur Mineralisierung, zum biologischen Primärabbau und zur Inokulumtoxizität bei unterschiedlichen Expositionskonzentrationen. Keine der Prüfsubstanzen mit drei oder mehr Kohlenstoffringen erwies sich als biologisch gut/schnell abbaubar. Um den biologischen Abbauprozess zu stimulieren, wurde daher die mikrobielle Gemeinschaft aus einer Kläranlage an das Gemisch heterocyclischer PAK angepasst, wobei die Tests zum biologischen Abbau entweder mit angepassten oder nicht angepassten Bakterien durchgeführt wurden. Überraschenderweise gab es bei der bakteriellen Anpassung Widersprüche: positive Effekte (wie der bessere Abbau von Carbazol) gehen mit weniger wünschenswerten Ergebnissen (erhöhte Toxizität gegenüber Benzo[c]carbazol) einher. Für vier der betrachteten Verbindungen wurde ein biologischer Primärabbau festgestellt, während fünf Substanzen in keinem der Tests Anzeichen für eine mikrobielle Umsetzung zeigten, was auf ein hohes Potenzial für eine Persistenz in der Umwelt hindeutet. Die Vorhersagen softwaregestützter Modelle wurden ebenfalls mit den Messergebnissen verglichen, und es zeigte sich, dass die Modelle bei der Vorhersage der Abbaukinetik der getesteten Heterocyclen nur teilweise übereinstimmende Ergebnisse lieferten. Darüber hinaus wurde die akute Kurzzeittoxizität heterocyclischer PAK anhand von drei Arten aus verschiedenen trophischen Ebenen des aquatischen Nahrungsnetzes untersucht: Bakterien (Aliivibrio fischeri), einzellige Algen (Raphidocelis subcapitata) und Wasserflöhe (Daphnia magna). Um eine stabile Exposition aufrechtzuerhalten und Verluste, zum Beispiel durch Sorption an Gefäßwandungen, zu vermeiden, wurde die Methode des „passive dosing“ an Routinetestprotokolle angepasst. Die meisten Verbindungen wurden in diesen Tests als hochgiftig für die getesteten Organismen identifiziert, wobei festgestellt werden konnte, dass unpolare Narkose (Basistoxizität) die wahrscheinlichste Art der toxischen Wirkung der getesteten Chemikalien ist. Darüber hinaus wurden die Membran-Wasser-Verteilungskoeffizienten (KMW) der heterocyclischen PAK durch Lipidmembran-Bindungsexperimente auf festem Trägermaterial sowie mit Hilfe von Vorausberechnungen (Modellierung) bestimmt. Die gemessenen KMW-Werte stimmten eng mit den KOW-Werten überein, was darauf hindeutet, dass beide Werte gute Deskriptoren für die Vorhersage der Basistoxizität bzw. des Bioakkumulationspotenzials der getesteten heterocyclischen PAK sein können. Schließlich wurden die langfristige chronische Reproduktionstoxizität und die Bioakkumulation großer heterocyclischer PAK mit Daphnia magna als aquatischer Modellorganismus untersucht. Die „passive dosing method“ gewährleistete eine konstante Exposition selbst bei sehr niedrigen Konzentrationen (70 ng L-1 bis 68 µg L-1) in groß angelegten und komplexen Versuchssystemen. Angesichts des enormen Zeit- und Arbeitsaufwands, der für die Durchführung von Langzeit-Toxizitätstests erforderlich ist, wurde eine Methode entwickelt, um die chronische Toxizität sowie das Bioakkumulationspotenzial in einem einzigen Test zu bewerten. Alle vier getesteten Substanzen waren gegenüber Daphnia magna hochtoxisch und stark bioakkumulierend. Unerwarteterweise erholte sich die Reproduktionsfähigkeit der Daphnien jedoch schnell von den toxischen Wirkungen der Heterocyclen, wenn die Chemikalienexposition endete. Die PNECs (Predicted No Effect Concentrations) der Testsubstanzen wurden ebenfalls anhand der Daten zur chronischen Toxizität verglichen. Die Werte lagen in einer Süßwasserumgebung zwischen 1 und 150 ng L-1 (PNEC). Für die meisten Testverbindungen gibt es keine Informationen zu Monitoringdaten in der Umwelt. Die Umweltkonzentrationen heterocyclischer PAK scheinen jedoch höher zu sein als die PNEC-Werte, insbesondere an kontaminierten Standorten, was eine Gefahr für die Ökosysteme und die menschliche Gesundheit darstellt. Insgesamt tragen die Ergebnisse dieser Forschungsarbeit nicht nur zu einem umfassenderen Verständnis und zu Fortschritten bei der Bewertung der von heterocyclischen PAK ausgehenden Umweltgefahren bei, sondern liefern auch mehrere methodische Verbesserungen, die in Zukunft bei der Prüfung anderer Stoffe, die gemeinhin als 'schwierig zu testen' bezeichnet werden, angewendet werden können.

There have been growing concerns over the decline of healthy river ecosystems and the severe consequences this decline could have on biodiversity, ecosystem services, and human well-being. These concerns have led to increased efforts in river restoration around the globe, which aim to improve the ecological health and functioning of rivers. The restoration is usually done by implementing strategies such as hydromorphological adaptation and flow management. These measures, nevertheless, do not guarantee the recovery of river ecosystems. This is because there are multiple factors contributing to the success of restoration projects, which can vary depending on the specific characteristics of each river system. Habitat modeling, one of the most widely used ecological quality assessment tools for rivers, has been applied in the evaluation of restoration projects. An aquatic ecosystem is complex, and its dynamic nature requires a comprehensive understanding of the interconnections between biotic and abiotic components. These components also have a high degree of spatial and temporal variability. Therefore, it is crucial that approaches and modeling techniques be tailored to capture this dynamic. In the assessment of river restoration, for instance, habitat modeling needs to account for the changes in flow patterns, sediment transport, water quality, and habitat availability/quality for the key indicator species that result from the restoration efforts. This study addresses the need for developing an integrated approach to habitat modeling, particularly for macrozoobenthos, an important indicator of river health that plays a crucial role in the functioning of aquatic ecosystems. The primary research objective is to improve the existing modeling framework (TRiMM) by focusing on three key aspects: 1) expanding the prediction factors of physical habitat that influence habitat suitability for macrozoobenthos; 2) integrating fuzzy algorithms in the suitability assignment process; 3) incorporating species' (re-)colonization capacity and habitat temporal variability into habitat connectivity assessment. The model adopts the fuzzy logic method in the habitat module to account for the interactions between various factors described in the habitat template (Poff & Ward, 1990). Moreover, the model considers both spatial and temporal changes in habitat parameters by running a transient simulation over a specific time period relevant to the life cycle requirements of the target species. This allows for a more accurate representation of the dynamic nature of river habitats and provides valuable insights into how they may change over time. Additionally, the model incorporates species' (re-)colonization potentials into habitat connectivity analysis by considering their dispersal capabilities. This helps in understanding how changes in habitat parameters can affect the overall connectivity of river habitats, which is crucial for assessing the resilience and sustainability of the systems. The proposed transient habitat modeling (TRiMM 2.0) is applied to two case studies of low-order rivers in Germany. The first case study focuses on a river that has been restored after a period of degradation. The habitat model was tested with sampling data, and the results reveal that the model improved when additional variables related to habitat were included. The second case study was a simulation of habitat suitability and connectivity in a hypothetical river reach. Hydraulic and morphological factors (water depth, velocity, temperature, and sediment) are simulated over a period of four years using SRH-2D. The simulation results showed that hydraulic and morphological factors had a significant impact on sediment characteristics, which in turn influenced habitat suitability and connectivity. This study also highlights the importance of considering multiple variables and their interactions when assessing river habitats. Additionally, the use of transient modeling provides information about long-term changes in habitat quality and connectivity.:Abstract Kurzfassung Contents List of figures List of tables Nomencature Acknowledgement List of publications 1. General introduction 1.1. Research motivation 1.2. Statement of research objectives 1.3. Structure of the dissertation 2. Macrozoobenthos and stream’s ecology 2.1. Macrozoobenthos and their habitat 2.2. Factors influencing the distribution of macrozoobenthos 2.2.1. Food sources 2.2.2. Water quality 2.2.3. Physical habitat 2.2.4. Colonization process 2.2.5. Presence of other species 2.3. Spatial scale and temporal variability 2.4. Conclusion 3. State of the art in river habitat modeling 3.1. Habitat modeling and river ecology assessment 3.2. Habitat modeling principles 3.2.1. Habitat suitability curves method 3.2.2. Fuzzy logic method 3.2.3. Generalized additive models 3.3. Existing benthos habitat modeling 3.3.1. PHABSIM 3.3.2. RHYHABSIM 3.3.3. BITHABSIM 3.3.4. CASiMiR 3.3.5. HABFUZZ 3.4. TRiMM and further development 3.5. Conclusion 4. Basis for the modeling concept and methodological framework 4.1. Physical habitat template 4.1.1. Streamflow regime 4.1.2. Substrate regime 4.1.3. Thermal regime 4.2. Habitat connectivity 4.3. Species colonization and habitat connectivity 4.4. Analysis scales 4.5. Conclusion 5. Transient river habitat modeling for macrozoobenthos – TRiMM 2.0 5.1. Habitat model description 5.2. Input data preparation 5.2.1. Field survey 5.2.2. Hydro-morphodynamic models 5.3. Habitat suitability calculation 5.4. Patch-building and patch dynamics analysis 5.5. Habitat connectivity calculation 5.6. Conclusion 6. Model applications 6.1. Case study 1: Simulation of habitat suitability for macrozoobenthos in a small restored stream (Saxony, Germany) Abstract 6.1.1. Introduction 6.1.2. Material and Method 6.1.3. Results 6.1.4. Discussion 6.1.5. Conclusion 6.2. Case study 2: Application of TRiMM 2.0 to simulate benthic habitat quality in a hypothetical reach of Zschopau river 6.2.1. Introduction 6.2.2. Methodology 6.2.3. Results 6.2.4. Discussion 6.2.5. Conclusion 7. Summary and future outlook 8. References

Invasive species are distributed all over the world and defined as organisms that have been introduced to a specific geographic area where they were not originally native. Such species have the ability to change the invaded ecosystem by for example competing with native organisms for limited food resources or occupying habitats with resulting environmental impacts. Therefore, knowledge about the characteristics of these invasive species as well as of the affected habitat types are very relevant to understand arising changes for endemic biodiversity and ultimately to protect it. For several large rivers in Germany, the round goby Neogobius melanostomus (Pallas 1814), a benthic fish species of Ponto-Caspian origin, represents a highly invasive fish species. In this thesis, I traced the invasion stage by conducting a literature research on the distribution pathways and documenting occurrence spots of N. melanostomus within two large rivers in Germany, the Elbe and the Oder. Within the Elbe River I determined the invasion front by applying electrofishing and environmental DNA (eDNA) analysis. This kind of knowledge enabled the results on diet composition gained from gut content and trophic niche width gained from stable isotopes (δ13C and δ15N) to be linked with the ongoing invasion stage. I found a bimodal distribution for N. melanostomus, from upstream and downstream. On examination of the downstream-directed invasion gradient, differences in N. melanostomus food composition as well as macroinvertebrate occurrence could be identified, with habitat structure and abundance being key determinants. Furthermore, N. melanostomus from sites invaded earlier exhibited the widest trophic niche size and individuals from the most recently invaded sites showed smaller niche sizes, most probably due to stronger competition for resources. Within the Oder River I focused on the competitive feeding interaction of N. melanostomus with the native fish species burbot (Lota lota). The main findings revealed that both fish species consumed preferred the same prey taxa (Crustacea, mainly Gammaridae) indicating potentially high competitive interactions for invertebrates. Whereas isotope ratios revealed a potential shift in the L. lota food web at the invaded site, represented by an enrichment of ẟ15N in muscle tissue and ẟ13C in liver tissue. This is indicative for a niche differentiation among both fish species.This thesis provides new insights into intra- and interspecies-specific impacts of N. melanostomus on riverine ecosystems under natural conditions, considering traceable invasion stages.:Table of contents 1 Abstract 3 Zusammenfassung 5 General introduction 7 Biological invasions: their introduction, dispersal and establishment 7 Application of environmental DNA to detect aquatic invasive species 10 The effects of invasive species on river food webs 10 Study Species: Round goby (Neogobius melanostomus) 12 Place of origin and spread 12 Round goby characteristics 13 Round goby feeding behaviour and effects on food webs 14 Round goby impacts on native fish species 15 Thesis Objectives 16 CHAPTER I 18 Tracing the colonization process of non-native gobies into a large river: the relevance of different dispersal modes 18 CHAPTER II 33 Diet compositions and trophic niche differentiation of Neogobius melanostomus along an invasion gradient in a large lowland river 33 CHAPTER III 54 Round goby (Neogobius melanostomus) impacts on diet composition and isotopic niche of native burbot (Lota lota) within the large Lowland River Oder, Germany. 54 General discussion 71 Factors influencing the invasion success of round goby 71 The importance of early invasion patterns 75 Invasion-induced alteration of riverine food webs 77 Challenges in the study of riverine invasions 80 Suggestions for research actions 82 Conclusion 85 References 86 Appendix 106 Chapter I 106 Chapter II 109 Danksagung 113 Selbstständigkeitserklärug 115

The discovery of antibiotics has been considered as one of the most remarkable scientific accomplishments of the last century. However, the extensive usage of antibiotics has led to the rapid emergence of antibiotic resistance genes (ARGs) and antibiotic resistant bacteria (ARB), which have been recognized as one of the biggest threats to human and environmental health. While ARGs and ARB are ubiquitous in diverse environments, aquatic environments play a particularly crucial role in their prevalence and dissemination. Furthermore, the microbial complexity and various pollutants persisting in aquatic environments significantly contribute to the evolution and spread of ARGs and ARB. However, the knowledge regarding the distribution pattern of ARGs on a large scale, as well as the interaction between microbial community, specific pollutants, and ARGs and ARB, is currently limited. In this study, I conducted systematic work at multiple scales, to fill crucial knowledge gaps that could support the future management of the spread of antimicrobial resistance. In particular, on the ecosystems level I depict the biogeographical patterns of ARGs in freshwater reservoirs, on the community level I explored the selection patterns of combinations of antibiotics on multidrug resistant strains in complex community context, and finally, on the population level I investigated the impact of cigarette smoke, and waste products on the dissemination of ARGs. In my first study, samples were collected from 24 freshwater reservoirs across southeast China and the biogeographical patterns of bacterial communities and ARG profile were characterized using 16S rRNA gene high-throughput sequencing and high-throughput-quantitative PCR. A distance-decay pattern for both, microbial communities and ARG profiles, were observed. However, larger differences between reservoir ARG profiles than microbial community compositions were detected. Further, I found that the biogeographical patterns of bacterial communities were simultaneously driven by stochastic and deterministic processes, while ARG profiles were not explained by stochastic processes, and the relationship between bacterial communities and ARG profiles was weak. In summary, this study indicated a decoupling of bacterial community composition and ARG profiles in inland waters under relatively low-human-impact at a large scale. In a second study, the selection dynamics for multidrug resistance between isogenic pairs of E. coli strains under exposure to multiple selective agents in the absence and presence of the microbial community were investigated using microcosm experiments. The presence of the community significantly decreased the selection for multidrug resistant strain under exposure to a single antibiotic. While pressure through the second antibiotic significantly decreased the activity and diversity of the community, its ability to reduce selection was consistently maintained at levels comparable to those recorded in the absence of the second antibiotic. This indicates that the observed effects of community context on selection dynamics are rather based on competitive or protective effects between the focal strains and a small proportion of bacteria within the community, than on general competition for nutrients. Last but not least, the effect of cigarette-derived pollutants on the proliferation of ARGs was explored using multifaced approaches. Cigarette smoke condensate in an artificial lung sputum medium significantly elevated the transfer rates of a multi-drug-resistance encoding plasmid between Pseudomonas strains. The overproduction of reactive oxygen species (ROS) as part of the bacterial stress response was directly connected to the increasing transfer rates. Similarly, cigarette ash leachate in an environmental medium significantly increased the plasmid transfer rates, and overproduction of ROS was equally detected. Furthermore, used cigarette filters with entrapped toxicants were submerged in a wastewater stream and colonized by distinct microbial communities compared to those colonizing unused control filters. The microbial communities colonizing used cigarette filters were significantly enriched in AMR, potential pathogenic bacteria and mobile genetic elements. Overall, the insights gained within this thesis into the spread of AMR at multiple scales constitute a valuable contribution to support future management and monitoring of ARGs in diverse ecosystems.

Research in grassland ecosystems worldwide has highlighted the crucial role many burrowing herbivore mammals provide as so-called “ecosystem engineers”. Many of these examples come from North America and have mostly been focused on Prairie Dogs (Cynomys sp.), which, due to their burrowing and grazing activities, are considered as a species of high ecological importance. Multiple studies have shown that their loss reduces grassland ecosystem functions and that their activities increase plant, arthropod, bird and mammal biodiversity, grassland heterogeneity, provide benefits to cattle, among many other key ecological roles. However, prairie dog species are distributed throughout multiple landscapes in North America, and most of the studies have been on mixed prairie grasslands. It is therefore not surprising that, when comparing the effects of different prairie dog species across ecoregions, the role of prairie dogs is not as clear and results vary, depending on spatial context, type of grasslands, climate and environmental conditions. Yet, grasslands are considered as some of the most highly threatened ecosystems, having high rates of conversion, desertification and biodiversity loss. Therefore, it is imperative to understand the effects prairie dog species have on the ecosystem. In this thesis, I focus on the Mexican prairie dog (C. mexicanus), a species in drastic decline due to habitat loss, caused by agriculture, overgrazing, human settlement, fragmentation, diseases (e.g. plague) and past eradication programs. C. mexicanus is considered endangered by the IUCN red list of threatened species and The Convention of International Trade in Endangered Species of Wild Fauna and Flora. It is also endemic to the Grassland Priority Conservation Area (GPCA) of El Tokio, located in Northeast Mexico within the Chihuahuan desert ecoregion and designated as a GPCA by the Commission for Environmental Cooperation (a cooperation between Canada, U.S. and Mexico) in 2009. Grasslands within GPCA El Tokio are highly fragmented as a result of many years of intense agricultural practices and other anthropogenic activities and around 90% of the original grasslands have been lost. However, it is considered an area of ecological importance due to its unique vegetation and edaphic associations, as well as being a key habitat for migratory birds. Up until now, very few studies have explored the role C. mexicanus has on plant taxonomical diversity and soil properties. Moreover, no studies focusing on prairie dogs have looked into the effect the species has on plant functional diversity or soil properties. Furthermore, GPCA El Tokio is a data-scarce area. Therefore, at this point, basic information and system understanding are urgently needed in order to identify which conservation efforts are most promising and which specific sites these efforts should focus on. Understanding the influence of prairie dog disturbance on vegetation parameters (taxonomic and functional), soil properties and their interaction with environmental conditions through time, will increase knowledge on the risks and vulnerability of grasslands, promoting solutions that can support grassland management. Furthermore, there is a clear gap in the literature about the varying effects of prairie dogs depending on different grassland types and environmental conditions. To study such conditions, I first classified and controlled for factors that could influence prairie dog ecosystem engineering effects addressing the following two general questions: (1) Which grassland types are present in GPCA El Tokio? (2) Do the different grassland types present in GPCA El Tokio influence prairie dog colony sizes and burrow density? When influencing factors were confirmed, I then assess the effects of prairie dog disturbance on vegetation and soil to answer the following questions: (3) Are there clear distinctions across GPCA El Tokio grassland environmental variables? (4) Do the different grassland types present in GPCA El Tokio influence prairie dog colony sizes and burrow density? To do so I selected a representative sample of sites covering the varying environmental conditions present in GPCA El Tokio grasslands, a data-driven clustering approach was used. Once clusters were defined, average burrow density was estimated from remote sensing imagery. To study vegetation measures I first, sampled plant species (92 species and 28 families and 6 traits) to obtain taxonomical and functional diversity measures, and obtained variables related to nutrient availability, carbon and climate regulation, and water regulation and purification. To study the effects on soil a total of 11 soil properties were obtained (382 samples were analyzed). Field work was done during the wet season and repeated during the dry season to obtain season variations of all measures. Soil measures were further obtained at depths of 0-10cm and 10 to 20cm. Results from the study classify GPCA El Tokio into 8 different grassland types with 4 clusters being the most dominant, named Agriculture, Arid, Calcareous, Mountain. I found agricultural grasslands had significantly lower colony sizes compared to calcareous grasslands when outliers were considered, but no differences when outliers were removed and burrow density varied depending on the grassland type, with mountain grasslands having a significantly lower burrow density compared to agricultural and calcareous grasslands. Regarding the effects of prairie dogs on vegetation and soil my findings suggest that functional metrics and community weighted mean (CWM) analyses responded to interactions between prairie dog disturbance, grassland type and season, whilst species diversity and cover measures were less sensitive to the role of prairie dog disturbance. I found weak evidence that prairie dog disturbance has a negative effect on vegetation structure, except for minimal effects on C4 and graminoid cover, but which depended mainly on season. Grassland type and season explained most of the effects on plant functional and taxonomic diversity as well as CWM traits. In the case of soil properties WOP grasslands tended to have higher nutrient availability than WP grasslands. However, mounds played a key role within WP grasslands. Mounds reduce compaction and increase nutrient levels of soil organic carbon, nitrogen, potassium, magnesium, and phosphorus. Such an effect was particularly present in calcareous and arid grasslands, and in many cases the effect was enhanced under wet conditions. In general, the effect of disturbance on soil property changes was dependent on environmental conditions. Overall, my study shows evidence that grassland type and season have a stronger effect than prairie dog disturbance on the vegetation of this short-grass, water-restricted grassland ecosystem.:3 TABLE OF CONTENTS Declaration of conformity 2 1 Acknowledgments 3 2 Extended Summary 5 4 List of Figures 11 5 List of Tables 13 1 Introduction 1 1.1 Importance of grasslands 1 1.2 Grassland conservation and ecosystem engineers 2 1.3 Prairie dogs: ecosystem engineers of grasslands in North America 3 1.4 Distribution of prairie dogs across North American grasslands 6 1.5 Current threats and conservation of the Mexican prairie dog C. mexicanus 7 1.6 Research questions, aims and objectives 9 2 Scientific background 13 2.1 Differences in grassland definitions and classifications 13 2.2 Dryland biases in grassland research 13 2.3 Research gaps on C. mexicanus 15 2.4 Functional diversity and disturbance 16 2.5 Soil complexity and limited research 17 3 Study Area: Grassland Priority Conservation Area of El Tokio 19 3.1 General overview 19 3.2 Bioregions, geology, topography and soils 20 3.3 Climate 23 3.4 Ecoregions, Land cover and Biodiversity 24 3.5 Anthropogenic activities and environmental issues 28 4 Assessing and classifying factors that could influence prairie dog ecosystem engineering effects 31 4.1 Data-driven identification of grassland types 33 4.1.1 Introduction 33 4.1.2 Methods 33 4.1.3 Results 36 4.1.4 Discussion and conclusion 39 4.2 C. mexicanus colony trends and differences across grassland types 41 4.2.1 Introduction 41 4.2.2 Methods 42 4.2.3 Results 48 4.2.4 Discussion and conclusion 52 5 Prairie dog disturbance effects on vegetation and soil across different grassland types and seasons 55 5.1 General methodology 57 5.1.1 Selection of grassland locations 57 5.1.2 Defining study sites and plots 58 5.1.3 Vegetation and soil sampling design 59 5.2 The effect of C. mexicanus on taxonomical and functional diversity across different seasons and grassland types 61 5.2.1 Introduction 61 5.2.2 Methods 63 5.2.3 Results 65 5.2.4 Discussion and conclusion 72 5.3 Soil property responses to the presence of C. mexicanus and its mounds across different grassland types, seasons and depths 77 5.3.1 Introduction 77 5.3.2 Methods 78 5.3.3 Results 81 5.3.4 Discussion and conclusion 88 6 Synthesis 95 6.1 Key findings 95 6.2 Management and Conservation opportunities 97 6.2.1 Designing management plans to suit existing environmental conditions 97 6.2.2 Functional diversity: an opportunity to identify patterns of vegetation across environmental conditions and scales. 98 6.3 Future research 99 6.3.1 Exploration of spatial and temporal effects 99 6.3.2 Mechanistic understanding of disturbance dynamics 100 6.3.3 A need for plant and soil ecological data 101 7 References 103 8 Appendix 131 8.1 Appendix 1. Prairie dog research reference list 132 8.2 Appendix 2. List of individual colonies with areas and other relevant information 140 8.3 Appendix 3. Historical and present colony data. 142 8.4 Appendix 4. Correlation results 144 8.5 Appendix 5. Top 3 models 146 8.6 Appendix 6. Species list per grassland type. 151 8.7 Appendix 7. Correspondence analysis (CA) for each grassland 154 8.8 Appendix 8. Post-hoc results 155 8.9 Appendix 9. Top 3 dominant species. 157 8.10 Appendix 10. HPD credible interval and ROPE. 159