Article révisé par les pairs
|Assessing the presence of human pathogenic Cryptosporidium oocysts in surface water remains a significant water treatment and public health challenge. Most drinking water suppliers rely on fecal indicators, such as the well-established Escherichia coli (E. coli), to avoid costly Cryptosporidium assays. However, the use of E. coli has significant limitations in predicting the concentration, the removal and the transport of Cryptosporidium. This study presents a meta-analysis of E. coli to Cryptosporidium concentration paired ratios to compare their complex relationships in eight municipal wastewater sources, five agricultural fecal pollution sources and at 13 drinking water intakes (DWI) to a risk threshold based on US Environmental Protection Agency (USEPA) regulations. Ratios lower than the USEPA risk threshold suggested higher concentrations of oocysts in relation to E. coli concentrations, revealing an underestimed risk for Cryptosporidium based on E. coli measurements. In raw sewage (RS), high ratios proved E. coli (or fecal coliforms) concentrations were a conservative indicator of Cryptosporidium concentrations, which was also typically true for secondary treated wastewater (TWW). Removals of fecal indicator bacteria (FIB) and parasites were quantified in WWTPs and their differences are put forward as a plausible explanation of the sporadic ratio shift. Ratios measured from agricultural runoff surface water were typically lower than the USEPA risk threshold and within the range of risk misinterpretation. Indeed, heavy precipitation events in the agricultural watershed led to high oocyst concentrations but not to E. coli or enterococci concentrations. More importantly, ratios established in variously impacted DWI from 13 Canadian drinking water plants were found to be related to dominant fecal pollution sources, namely municipal sewage. In most cases, when DWIs were mainly influenced by municipal sewage, E. coli or fecal coliforms concentrations agreed with Cryptosporidium concentrations as estimated by the meta-analysis, but when DWIs were influenced by agricultural runoff or wildlife, there was a poor relationship. Average recovery values were available for 6 out of 22 Cryptosporidium concentration data sets and concomitant analysis demonstrated no changes in trends, with and without correction. Nevertheless, recovery assays performed along with every oocyst count would have enhanced the precision of this work. Based on our findings, the use of annual averages of E. coli concentrations as a surrogate for Cryptosporidium concentrations can result in an inaccurate estimate of the Cryptosporidium risk for agriculture impacted drinking water intakes or for intakes with more distant wastewater sources. Studies of upstream fecal pollution sources are recommended for drinking water suppliers to improve their interpretation of source water quality data. © 2014 Elsevier Ltd.