Article pubs.acs.org/est
Quantifying and Reducing Uncertainty in Estimated Microcystin Concentrations from the ELISA Method Song S. Qian,*,† Justin D. Chaffin,‡ Mark R. DuFour,† Jessica J. Sherman,† Phoenix C. Golnick,† Christopher D. Collier,† Stephanie A. Nummer,† and Michaela G. Margida† †
Department of Environmental Sciences, The University of Toledo, Toledo, Ohio 43606, United States Franz Theodore Stone Laboratory, The Ohio State University, Put-in-Bay, Ohio 43456, United States
‡
S Supporting Information *
ABSTRACT: We discuss the uncertainty associated with a commonly used method for measuring the concentration of microcystin, a group of toxins associated with cyanobacterial blooms. Such uncertainty is rarely reported and accounted for in important drinking water management decisions. Using monitoring data from Ohio Environmental Protection Agency and from City of Toledo, we document the sources of measurement uncertainty and recommend a Bayesian hierarchical modeling approach for reducing the measurement uncertainty. Our analysis suggests that (1) much of the uncertainty is a result of the highly uncertain “standard curve” developed during each test and (2) the uncertainty can be reduced by pooling raw test data from multiple tests. Based on these results, we suggest that estimation uncertainty can be effectively reduced through the effort of either (1) regional regulatory agencies by sharing and combining raw test data from regularly scheduled microcystin monitoring program or (2) the manufacturer of the testing kit by conducting additional tests as part of an effort to improve the testing kit.
■
INTRODUCTION Cyanobacterial blooms have become a global problem due to excessive anthropogenic nutrient inputs and a warming climate.1,2 Aside from the negative ecological impacts associated with blooms,3 these blooms also pose a threat to human health because they produce toxic compounds that impair the nervous system, liver, and skin.4 Because toxic cyanobacterial blooms often occur in waters that are sources for drinking water, frequent and accurate quantification of cyanboacterial toxins in treatment-finished drinking water is paramount in protecting the public. Microcystins (MCs) are a group of liver toxins commonly produced by many genera of cyanobacteria worldwide.5 These toxins have been associated with liver cancers and human fatalities; particularly in people with poor liver function already being treated through dialysis.4,6 Although many authors have discussed the need for controlling harmful algal blooms (through reducing N and P loadings to waters) to reduce the risk of MC exposure, the problem of MC measurement uncertainty is not adequately discussed in the literature. Information about measurement uncertainty is important to local government agencies responsible for providing safe drinking water. The Toledo water crisis of August 2014 started when one MC concentration measurement from a treated water sample in Collins Park Water Treatment Plant of City of Toledo, Ohio, exceeded the Ohio drinking water standard of 1 μg/L. (The Ohio standard is based on the World Health Organization drinking water quality criterion for microcystin.7) After the initial detection, three additional tests, each with the usual two replicates, were carried out on the same day using the same water sample © XXXX American Chemical Society
and each time at least one replicate showed a concentration above the criterion. These results prompted the City of Toledo to issue a “Do not drink” advisory on the morning of August 2, 2014, affecting about a half million residents. Additional tests were conducted on drinking water from the water treatment plant and throughout the distribution system until all samples consistently showed microcystin concentration below “detectable” levels (
