Surface waters are used for recreation, irrigation water, and as source water for drinking water treatment plants. These uses can be associated with human health risks when fecal contamination from point and non-point sources introduces pathogens into surface waters capable of causing waterborne disease. The quality of surface waters is typically monitored with indicator organisms, and it has commonly been assumed in literature, quantitative microbial risk assessments (QMRA), and surface water management decision-making that indicator and pathogen persistence are similar, and that indicators and pathogens decay at a constant rate in the environment. To address these assumptions, this dissertation presents i) a systematic literature review that collated, compared, and analyzed the available persistence data for indicators and pathogens in surface waters; ii) a meta-analysis that fit exponential decay and alternative models to the database of over 600 experiments, identified a model that best fit the data most frequently, and statistically evaluated the relationships between frequently documented water quality factors and observed persistence dynamics; iii) a general model developed with Bayesian hierarchical modeling that quantifies the uncertainty between indicator and pathogen persistence; and iv) a QMRA case study that evaluates the impact of persistence knowledge for decision-making pertaining to a recreational waterbody impacted by a sewage spill event. The systematic literature review (Chapter 2) found that the 61 selected studies predominantly evaluated FIB, freshwater matrices, and culture-based methods of detection. Comparing the methods and results across the studies qualitatively suggested potential interactions between sunlight, water type, and method of detection, and between predation, water type, and temperature. Within the subsequent meta-analysis (Chapter 3), the Juneja and Marks 2 (JM2) model, based on the logistic probability distribution, provided the best fit to the data most frequently. First-order decay kinetics provided the best fit to less than 20% of the analyzed data. Random forest methods identified temperature, water type, and predation as the most important factors influencing persistence, and the protozoa target type differed the most from FIB. A general model was developed using the comprehensive database of persistence experiments, the JM2 model, temperature, predation, and water type data, and Bayesian hierarchical modeling techniques (Chapter 4). A varying-intercept model with target-specific intercepts and population-level coefficients for temperature, predation, and water type was the optimal evaluated model form. The general model indicated that protozoa persistence more commonly has initial periods of minimal decay and virus decay typically tapers off the most quickly over time. Median uncertainty factors quantified with the general model ranged from 1 to 3.4 for bacteria, bacteriophage, virus, and protozoa persistence behaviors compared to FIB. The application of the uncertainty factors was demonstrated within a QMRA case study in which the JM2 model was fit to culturable enterococci (cENT), enterohemorrhagic Escherichia coli (EHEC), and adenovirus (HAV) data to characterize the persistence of the targets after the containment of a sewage spill (Chapter 5). Applying temperature-specific uncertainty factors to the cENT data ensured the risk of illness associated with EHEC and HAV ingestion fell below the Recreational Water Quality Criteria limit of 36 in 1,000 swimmers. The work presented herein indicates that broadly applying first-order decay kinetics to persistence data may lead to erroneous decision making in the fields of water management and protection, and that a general model for persistence can add value to the indicator-pathogen paradigm.
Read
