Date Approved

12-2019

Graduate Degree Type

Thesis

Degree Name

Biology (M.S.)

Degree Program

Biology

First Advisor

Richard R. Rediske

Second Advisor

James N. McNair

Third Advisor

Charlyn Parttridge

Fourth Advisor

Daniel J. Ftobish

Academic Year

2019/2020

Abstract

Public beaches are routinely tested for potentially pathogenic bacteria to protect beachgoers from possible illness. An EPA approved method, Colilert™, used for testing E. coli in recreational water requires 18 – 22 hours before a result is reported but, recreators have already contacted unsafe water before the beach is closed. My study focused on a U.S. EPA proposed qPCR method (Draft Method C) to quantify E. coli in recreational waters that can provide same-day results. In Chapter 2, I examined the calibration procedure used to validate Draft Method C and compared standard curve intercept and slope estimates calculated with a Bayesian model to estimates generated from a simpler weighted linear regression (WLR) model to determine if it can replace the complicated Bayesian model for method implementation. A < 1% difference in the overall mean and median intercept and slope was observed between the two models, demonstrating that the WLR model results were comparable. I also analyzed inter-lab variability in intercept, slope, and R2 estimates produced by passing curves from the WLR model in a multi-lab 2018 data set. Significant pairwise differences were detected in 11% of the 36 inter-lab intercept comparisons; no differences were detected in the slope or R2 parameters. I concluded that the proposed standard curve acceptance criteria showed minimal variation between labs, thus ensuring reported results are accurate and reliable. In Chapter 3, I measured E. coli concentrations with Colilert™ and Draft Method C in water samples collected during the summer of 2018 from 14 inland lake and 6 Lake Michigan beaches in Muskegon County, MI. Kaplan-Meier distribution curves and log-rank trend tests were used to identify categorized environmental variables that significantly impacted E. coli concentrations based on the quantification method and the lake type. Bird count and birds present/absent significantly impacted E. coli levels at beaches of both lake types and quantification methods. The remaining variables significantly impacted E. coli concentrations depending on lake type and quantification method. Therefore, predictive models for beach water quality should consider lake type and quantification method to account for the influence of environmental variables.

Download

Share

COinS