Visualizing The Invisible: Using Microscopy to Observe Microbial Communities Over Time

Microorganisms cycle carbon and other nutrients to produce key compounds necessary for all life; thus, it is important to understand how these processes will be affected by a changing environment. While molecular techniques have made critical insights into microbial ecology and biogeochemistry, they often examine only components of the ecosystem, such as certain size fractions or domains. Microscopy-based approaches offer a chance to examine the entire microbial community holistically and infer trophic interactions but are often laborious and low-throughput. We are developing high-throughput microscopy methods to systematically characterize and quantify all levels of the microbial ecosystem (viruses, bacteria, phytoplankton, zooplankton) with the goal of creating automated microscopy workflows which are accurate and comparable to other high-throughput plankton visualization technologies, such as Imaging FlowCytobot and FlowCam. These microscopy approaches were applied to understand how different sources of organic matter structure microbial community composition in the Pamlico Sound Estuary, NC, USA in September 2021. Water was collected and incubated under one of three light regimes (visible wavelengths, UV-only, and dark) for 30 days to examine microbial responses to different sources of organic matter (autochthonous, high organic matter photochemical degradation, and allochthonous, respectively). SYBR-stained and autofluorescent cells were imaged in z-stacks at multiple magnifications and in both DIC and fluorescent channels. Abundances and cell properties were determined using automated artificial intelligence techniques with supervised training and compared to manually determined counts and properties. Automated methods were moderately successful in reproducing trends in counts for single cells and chains (>1 µm) but were an order of magnitude fewer in counts than manual estimations. Methods were less successful at counting cells within aggregates and cells with low levels of fluorescence, especially as tanks accumulated debris over the course of the experiment. Using these automated methods, we were able to show that UV and especially visible light treatments slowed cell population decline compared to dark treatments. By combing these high-throughput microscopy methods with nutrient analyses, physiology rate measurements, and molecular techniques, this research is establishing a holistic understanding of microbial community structure and biogeochemical cycling in the estuarine environment.Bachelor of Scienc