Navigating complexity: exploring the untapped potential of mesocosm experiments for multiple-stressor research

Freshwater ecosystems worldwide are exposed to multiple anthropogenic stressors, resulting in pronounced declines in freshwater biodiversity. To effectively direct mitigation measures to prevent further biodiversity declines, we have to be able to make quantitative predictions for the effects of these stressors. This is precisely the goal of multiple-stressor research. Here, mesocosm experiments play a central role, because they offer strong experimental control, high replication as well as ecologically realistic conditions. Despite their widely recognized importance, much of their potential in improving our mechanistic understanding of multiple-stressor effects remains untapped. In this thesis, I demonstrate how mesocosm experiments can be more effectively designed and analyzed to seize key opportunities, which have so far been underexplored. Mesocosm experiments can simulate entire ecosystems, however, they are often analyzed focusing on single target groups. In Chapter 1, I perform a holistic assessment of the joint effects of salinization and reduced flow velocity in an urban stream ecosystem across multiple organism groups (macroinvertebrates, eukaryotic algae, hyphomycetes and parasites) and ecosystem functions (organic-matter decomposition, primary production, microbial respiration). While community composition was strongly restructured by changes in flow velocity across most organism groups, there was no or only minor evidence that salinization affected communities or ecosystem functions. The comprehensiveness of this assessment allows for the development of novel hypotheses concerning complex causal effect pathways through the food chain and ecosystem-wide legacy effects from historic salt pollution in the study area. Research on global warming effects in flow-through mesocosms is logistically challenging, because it requires precise temperature control against fluctuating ambient temperatures for large quantities of water. Thus, most stream mesocosm experiments address heatwaves instead of chronic warming or require substantial manual effort. In Chapter 2, I present a modular, automatic heating system for flow-through mesocosm systems. In a field application, I show that the heating module accurately maintains temperature differences between the ambient water temperature and a warming treatment for multiple days. The heating module makes it now possible to address underexplored aspects in global warming research. Therefore, Chapter 3 demonstrates two case studies focusing on how biotic interactions modulate the effects of and recovery from warming. The first mesocosm experiment investigates the role of predator-prey dynamics, warming and salinization in shaping insect emergence patterns. Warming stimulated insect emergence, but only in the presence of insectivorous fish, salinization or both, stressing that biotic interactions can be as impactful as additional abiotic stressors in shaping biological responses to global warming. The second mesocosm experiment focused on recovery of periphyton communities after release from warming, salinization and reduced flow velocity. While periphyton biomass rapidly rebound to levels before stressor exposure, the recovery of community composition was slowed down or impeded, probably by priority effects from tolerant resident taxa. Therefore, community recovery after release from warming or other stressors has to be explicitly addressed in mesocosms, since the altered biotic context can result in an asymmetry between community degradation and recovery pathways. Due to their strong experimental control, mesocosms are an important tool to test hypotheses concerning stressor interactions. This involves contrasting experimental observations for joint stressor exposure with null-model predictions, which calculate the net effect of two or more stressors from their single-stressor effects. Null-model selection can be guided by mechanistic assumptions of co-tolerance and similarity in the stressors’ modes of action, however, commonly used statistical models often impose a specific null model and therefore limit the researcher’s choice. In Chapter 4, I introduce an analytical framework that enables researchers to flexibly test any a priori defined null model for multiple stressors with the best-fitting statistical model. By decoupling null-model choice from statistical model fitting, this analytical framework untaps the potential for hypothesis-driven multiple-stressor research and enhances the interpretability of experimental outcomes. In conclusion, this doctoral thesis illustrates how mesocosms can be used more effectively to investigate the processes of ecosystem degradation and recovery from multiple stressors. Developing such a mechanistic understanding opens new opportunities to advance freshwater biodiversity conservation and restoration in a rapidly changing world.