Fecal material from animals and untreated sewage in waterways pose serious risks to human health. Methods that detect and determine sources of aquatic fecal pollution are a part of microbial source tracking (MST). Currently, one of the most widely-used MST methods is susceptible to sensitivity and specificity issues and exhibits variable geographic results which obstruct its widespread use in regulatory capacities. These drawbacks have led to the continued development and evaluation of new MST methodologies. One such new methodology is community-based MST, which uses high-throughput DNA sequencing (HTS) to construct taxonomic profiles of potential fecal sources and environmental samples. SourceTracker, a Bayesian classifier designed to determine sources of contamination in HTS data, is the most common community-based MST tool. Due to the relative novelty of SourceTracker, few studies have evaluated its limitations as a MST tool. This thesis explores the use and limitations of community-based MST with SourceTracker for identifying sources of fecal bacteria in waterways. HTS analysis of microbiota from a diverse collection of fecal samples and environmental samples revealed that the community compositions of freshwater and feces were significantly different, allowing for determination of the presence of fecal inputs. Moreover, the differences in community composition between multiple fecal sources were also statistically significant suggesting that differentiation between fecal sources was possible. When SourceTracker was challenged to identify fecal sources in a freshwater lake with the most diverse and extensive fecal source library used in a community-based MST field study, SourceTracker was able to identify wastewater effluent from a nearby wastewater treatment plant. SourceTracker also predicted the presence of geese and gull wastes which is in agreement with previous MST studies in that research location. To examine the limitations of community-based MST using SourceTracker, SourceTracker was challenged with identifying known fecal sources in in situ mesocosms using different fecal source library configurations. Across nearly all fecal source library configurations, SourceTracker was able to accurately predict most sources in the in situ mesocosms. These results were most reliable when the fecal source library contained only the known sources. When fecal sources were missing from the fecal source library, erroneous classifications not seen when all known sources were present became common. Results of this chapter also indicated that the ideal SourceTracker source profile has low-intragroup variability and shares few taxa with other sources. To understand how SourceTracker predictions are influenced by the concentration of feces, geographical variance, diet, and age of source animals, SourceTracker was challenged to identify spiked cow feces from a farm in St. Paul, MN, USA using cow feces from other farms across North America as sources. Analysis of the cow fecal microbiome revealed statistically significant differences in cow fecal taxa at the OTU level when evaluating sample relationships by farm, state, age groups, and diet. Most of the OTU variance was attributed to the individual farms the cows came from, and not to age and diet differences. Source samples from all locations yielded high SourceTracker predictions. On average, higher predictions in source attribution were associated with more concentrated samples and with cow feces from animals closer to the spiked fecal source location. While SourceTracker was able to detect fecal contamination with source animals that were not from the original location, animals sourced closer to the contamination site provided the most accurate predictions. All of these data demonstrate the ability of SourceTracker to accurately identify sources, making this program a powerful tool for community-based MST.
University of Minnesota Ph.D. dissertation. 2018. Major: Biochemistry, Molecular Bio, and Biophysics. Advisors: Michael Sadowsky, Daniel Bond. 1 computer file (PDF); ix, 120 pages.
Evaluation of the Use and Limitations of a Community-Based Microbial Source Tracking Method.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.