Modeling Enterococcus densities measured by quantitative polymerase chain reaction and membrane filtration using environmental conditions at four Great Lakes beaches 5 Justin W. Telech a, *, Kristen P. Brenner b , Rich Haugland b , Elizabeth Sams a , Alfred P. Dufour b , Larry Wymer b , Timothy J. Wade a a National Health and Environmental Effects Research Laboratory, Human Studies Division, U.S. Environmental Protection Agency, MD 58C, Research Triangle Park, NC 27711, USA b National Exposure Research Laboratory, U.S. Environmental Protection Agency, 26 W. Martin Luther King Dr., Cincinnati, OH 45268, USA article info Article history: Received 28 February 2009 Received in revised form 22 June 2009 Accepted 5 July 2009 Published online 8 July 2009 Keywords: QPCR Membrane filtration Enterococcus Great Lakes Liner regression abstract Data collected by the US Environmental Protection Agency (EPA) during the summer months of 2003 and 2004 at four US Great Lakes beaches were analyzed using linear regression analysis to identify relationships between meteorological, physical water characteristics, and beach characteristics data and the fecal indicator bacteria, Enterococcus. Water samples were analyzed for Enterococcus densities by quantitative polymerase chain reaction (qPCR) and membrane filtration (MF). This paper investigates the ability of regression models to accurately predict Enterococcus densities above or below a threshold value, using environmental data on a beach-by-beach basis for both methods. The ability to create statistical models for real-time water quality analysis would allow beach managers to make more accurate decisions regarding beach safety. Results from linear regression models indicate that environmental factors explain more of the variability in Enterococcus densities measured by MF than Enterococcus densities measured by qPCR. Results also show that models for both methods did not perform well at predicting occurrences in which water quality levels exceeded a threshold. ª 2009 Elsevier Ltd. All rights reserved. 1. Introduction The use of statistical models to predict fecal indicator bacteria (FIB) densities is one method which scientists and beach managers are attempting to use to provide information about microbial fecal contamination in water in real time (Nevers and Whitman, 2005; Olyphant, 2005; Francy and Darner, 2006). It has been shown that FIB densities, measured using membrane filter (MF) analysis, correlate well with certain environmental characteristics, such as wind direction, water temperature, pH and rainfall (Nevers and Whitman, 2005; Olyphant, 2005; Francy and Darner, 2006). Statistical models can express this relationship by incorporating FIB measure- ments with environmental measurements using multiple 5 The information in this document has been funded by the United States Environmental Protection Agency. It has been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents reflect the views of the Agency. * Corresponding author. Tel.: þ1 919 767 7208. E-mail address: telech.justin@epamail.epa.gov (J.W. Telech). Available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/watres 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.002 water research 43 (2009) 4947–4955