Anke Schneeweiß

• The massive use of chemicals by humans is increasing pollution of the world’s ecosystems. Yet, knowledge about exposure and effects of chemicals in real-world ecosystems remains limited. Prediction of chemical effects in the context of ecotoxicological research and chemical regulation continues to focus on organism- or population-level responses established under simplified conditions while aiming to protect the functioning of ecosystems. A unified, comprehensive framework for the prediction of chemical effects in real-world ecosystems is still lacking. A major limitation of ecotoxicological studies considered in predictive modelling is that they rarely consider spatial dynamics (e.g. gene flow or species dispersal) as relevant processes influencing the trajectory of populations or communities, respectively. For instance, the spatial propagation of pesticide effects from polluted to least impacted sites has been predicted in several modelling studies but has not yet been characterised in the field. The thesis starts in Chapter 1 with a brief introduction to chemical pollution in ecosystems, chemical effect prediction in ecotoxicology, and pesticides in freshwater ecosystems, then outlines the main objectives of the thesis. Subsequently, Chapter 2 presents a conceptual study about the current prediction of chemical effects in ecotoxicology and potential future avenues to improve ecological relevance of effect predictions by addressing the integration of different levels of biological organisation (termed biological levels). The study shows that approaches and tools that currently contribute to the prediction of chemical effects can be attributed to three idealised perspectives: the suborganismal, organismal and ecological perspective. The perspectives focus on different biological levels and are associated with distinct scientific concepts and communities. They complement each other so theoretical and empirical links between them may enhance prediction by capturing the entire phenomenon of chemical effects, from chemical uptake to ecosystem effects. Complex experimental studies accounting for eco-evolutionary dynamics are needed to cross barriers between biological levels as well as spatiotemporal scales. Overall, the conclusions of Chapter 2 may help to develop overarching frameworks for predicting chemical effects in ecosystems, including for untested species. Chapters 3 and 4 present a field study combined with laboratory analyses on the potential propagation of pesticides and their effects from agricultural stream sections to the edge of least impacted upstream sections, that can serve as refuges for many species. The study examines exposure and effects for different biological levels at three site types, the pesticide-polluted agricultural sites (termed agriculture), least impacted upstream sites (termed refuge) and transitional sites (termed edge) in six small streams of south-west Germany. The results in Chapter 3 show that regional transport of pesticides can lead to ecologically relevant pesticide exposure in forested sections within a few kilometres upstream of agricultural areas (i.e. at both edge and refuge sites). As further demonstrated in Chapter 3, the tested indicators of community responses (Jaccard Index, taxonomic richness, total abundance, SPEARpesticides) together suggest a species turnover from upstream refuge to downstream agricultural sites and a potential influence of adjacent agriculture on the edge sites. In contrast, Chapter 4 does not identify any particular edge effect that distinguish edge organisms and populations in edge sites from those in more upstream refuge sites. Gammarus fossarum populations at edges show equal levels of imidacloprid tolerance, energy reserves (i.e. lipid content) and genetic diversity to populations further upstream. Gammarus spp. from agricultural sites exhibit a lower imidacloprid tolerance compared to edge and refuge, potentially due to energy trade-offs in a multiple stressor environment, but related effects do not propagate to the edges (Chapter 4). Notwithstanding, the results of Chapter 4 indicate bidirectional gene flow between site types, supporting the hypothesis that adapted genotypes – if present at locally polluted sites – could spread to populations at least impacted sites. Taken together, Chapters 3 and 4, illustrate that pesticides and their effects can potentially propagate to least impacted upstream sections, empirically novel findings to our knowledge. These results of this thesis can help in predicting or explaining population and community dynamics in least impacted habitats and can ultimately inform pesticide management as well as freshwater restoration and protection of biodiversity.