Freshwater ecosystems are characterized by both concentrated biodiversity and high extinction rates, with a disproportionate level of exposure and vulnerability to anthropogenic activities. Presently, land use poses the largest threat to freshwater ecosystem functioning, with climate change emerging as a considerable threat with potential to interact with local land-use stressors. Mean temperature increase is usually manipulated in isolation in climate- change experiments. However, heatwaves, elevated atmospheric CO2 and hydrological variability can be regarded as the three key climate-change variables affecting freshwaters. These factors are increasingly considered in freshwater research, yet all three have never been manipulated in combination in a realistic experiment. Further, future climate-change effects on a given freshwater system are likely to depend on the level of local land-use impact. There is urgent need to understand how these two broad stressor categories may interact in future, yet experimental evidence at the community- and ecosystem-level is rare. High fine sediment loads are a key land-use stressor known to cause strong shifts in invertebrate or periphyton community composition and reduce sensitive taxa. As a localized stressor, in-stream sediment loading can be managed at a local scale, thus any knowledge of its interactions with global climate-change stressors may allow mitigation of harmful effects.

My PhD research aimed to address the above knowledge gaps by carrying out the first outdoor freshwater mesocosm experiment worldwide to simultaneously manipulate these four factors. Using 128 experimental stream channels, we manipulated dissolved CO2 (ambient, enriched), fine sediment (no sediment, added), temperature (ambient, 2 week-long heatwaves) and flow velocity (constant, variable). To assess effects between trophic levels, one fish (upland bully, Gobiomorphus breviceps) was added to each channel. The entire benthic invertebrate communities were sampled at the end of the 7-week experiment and identified by morphology, with resulting abundance, diversity, community composition and individual taxa responses presented in Chapter 2. During the experiment, invertebrate drift and insect emergence responses were each sampled four times, presented in Chapter 3. Fish were sampled and measured for growth, survivorship and gut contents, presented in Chapter 4. Additionally, DNA metabarcoding was carried out on the benthic invertebrate community. From this, presence/absence data were produced for 403 operational taxonomic units (OTUs), presented in Chapter 5. Read abundances (number of reads of each OTU sequence) were also produced; however, as the robustness of such data is still being widely discussed in the literature, these were explored in Chapter 6 (Synthesis). Here, I compare the conclusions drawn from analysis of the benthic community in Chapters 2 (morphological identification) and 5 (metabarcoding), and also assess the viability of using DNA metabarcoding read abundance data in future experiments to detect stressor effects at high taxonomic resolution.

All four experimental factors individually changed benthic invertebrate community composition, with the strongest effect caused by sediment, followed by heatwaves, CO2, and flow velocity variability. Notably, this ranking was the same in Chapters 2 and 5, despite reflecting different drivers of community change. In Chapter 2, these changes were driven by shifts in relative abundances of 13 common taxa at mixed taxonomic resolution. In Chapter 5, they were driven by changes in the presence of less-common OTUs, suggesting that morphological ID and DNA metabarcoding can each provide unique insights into how these stressors modify ecosystems. In Chapter 2, CO2*sediment interactions were most common, a pattern not found in Chapter 5. In Chapter 3, a large increase in emerging Chironomidae during the first heatwave likely led to the reduced benthic abundance of two chironomid subfamilies caused by heatwaves seen in the benthos. Conversely, enriched CO2 lowered Chironomidae emergence across most dates, yet also reduced Orthocladiinae and Chironominae abundance in the benthos, suggesting these changes could not be explained by increased emergence rates and supporting the hypothesis that CO2 enrichment may reduce chironomid growth or survival rates via indirect effects of reduced algal food quality. In bullies, strong heatwave-induced shifts in abundances of key invertebrate prey led to changes in gut contents, but these did not translate to reduced fish growth or survival. CO2 was the only factor to lower fish growth, condition and survival. Reductions of growth and condition occurred regardless of sediment level, but survival was only decreased by CO2 in channels without sediment. These results suggest that CO2 imposed direct effects on bullies, disrupting their acid-base balance leading to a chronic energy deficit, which may have been exacerbated by reduced bottom-up energy transfer in stony-bottom channels without sediment.

Overall, my findings suggest a high potential for fine sediment loading to interact with global climate-change stressors, particularly elevated CO2, in unexpected ways. More research in this area is needed to further untangle the mechanisms behind these patterns, particularly with respect to potential changes in bottom-up energy transfer caused by raised CO2 in stony- bottom versus sediment-impacted streams. Finally, DNA metabarcoding has potential to increase the efficiency of experiments during an urgent time in climate-change research. However, inconsistencies between the two methods in detecting important stressor interactions mean that currently, DNA metabarcoding cannot replace identification based on morphology.