|Home | About | Journals | Submit | Contact Us | Français|
Giardia species recovery by U.S. Environmental Protection Agency method 1623 appears significantly impacted by a wide size range (2 to 30 μm) of particles in water and organic matter. Cryptosporidium species recovery seems negatively correlated only with smaller (2 to 10 μm), presumably inorganic particles. Results suggest constituents and mechanisms interfering with method performance may differ by protozoan type.
Prevention of waterborne diseases of protozoan etiology has remained a public health priority since the early 1990s, with concern considerably heightened by the 1993 Milwaukee cryptosporidiosis outbreak, which sickened over 400,000 residents (10). Source water contamination by Cryptosporidium spp. and other protozoan parasites (Giardia spp. and Toxoplasma spp.) presents a particular challenge to water quality managers due to the ubiquity of protozoans in wastewater effluents (1), the widespread infection of domestic animals and wildlife (6), the resistance of protozoans, especially Cryptosporidium, to traditional disinfection methods (12), and the uncertain relationship between the presence of protozoans and fecal indicator bacteria typically used in water quality monitoring (3, 4, 9).
In order to address the unique threat to drinking water supplies posed by protozoans, the United States Environmental Protection Agency (EPA) promulgated the Information Collection Rule in 1992, which required source water monitoring for Cryptosporidium and Giardia by municipal water authorities (13). To facilitate effective and comparable monitoring, the EPA developed standardized method 1623 for simultaneous cyst and oocyst recovery and detection (14). The method involves concentration of (oo)cysts in water samples via filtration and centrifugation, purification via immunomagnetic separation, labeling via immunofluorescent stain, and epifluorescence microscope enumeration. As a performance-based method, quality assurance requires regular demonstration of minimum (oo)cyst recovery levels.
Although validated by the EPA and employed in numerous research studies and compliance monitoring, low (oo)cyst recovery, particularly from poor-quality waters, remains a serious problem impacting regulatory decisions. Many studies have noted declines in Cryptosporidium or Giardia recovery levels to near or below minimum acceptable levels in waters of high turbidity or high total suspended solids (5, 6, 7, 11). Because turbidity and the total suspended solids concentration provide only gross estimates of particle concentration and no information regarding size distribution, the size and type of suspended solids responsible for this interference remain unknown and unexplored.
To determine the sizes of particulates responsible for reductions in recovery by method 1623, data on (oo)cyst recovery as well as particle size distribution in 52 7.0-liter freshwater grab samples were analyzed for statistical associations. These data were collected as part of a field study investigating indicator organism and protozoan fate and transport in New York City's Kensico Reservoir watershed. Results for wild-type (oo)cyst presence and correlation with indicator organisms have been published previously (4), but there has been no analysis that considered water quality parameters possibly responsible for the observed variations in protozoan recovery.
All samples were analyzed via U.S. EPA method 1623 for wild-type Giardia and Cryptosporidium concentrations, with ColorSeed spikes (8, 15) for determination of matrix recovery. Detailed descriptions of sampling procedures and materials have been provided by Cizek et al. (4). A Coulter Multisizer I (Coulter Electronics Ltd., Luton, England) provided the number, volume, and surface area of particles in discrete 0.5-μm-diameter increments (range, 2 to 60 μm). Total organic carbon concentrations were determined via standard method 5310B (16) using a Shimadzu TOC-5000 combustion-infrared analyzer (Shimadzu Corp., Kyoto, Japan). Sample pH levels were consistently near neutral. As the data were not normally distributed (Shapiro-Wilk test; P < 0.01), the Spearman rank correlation coefficient (ρs) was used to identify correlations. Statistical analyses were conducted using SAS statistical software (SAS Inc., Cary, NC), with significance established at an α level of 0.05.
Recoveries of matrix spike Giardia cysts and Cryptosporidium oocysts were consistently low, with no differences between storm event and dry weather samples. Average Giardia recovery for all samples was 22% (range, 3 to 45%), and average Cryptosporidium recovery was 17% (range, 0 to 74%). While low, average recovery values did meet the minimum criteria established in method 1623 (11 to 100% recovery of Cryptosporidium and 14 to 100% recovery of Giardia).
As expected, the recovery of spiked (oo)cysts decreased significantly as the total number of suspended particles increased (P < 0.05). However, separation of particle number concentration into six diameter size classes revealed that the observed decreases in Cryptosporidium and Giardia recoveries were correlated with different particle sizes (Table (Table1).1). While the decrease in Giardia recovery was correlated with increasing particle concentration within a wide range of size classes (2 to 30 μm), Cryptosporidium recovery was significantly negatively correlated only with particles of diameters from 2 to 10 μm. Negative correlations between Giardia recovery and particle concentration were similarly significant for all particle size classes between 2 and 30 μm.
Recovery of Giardia cysts was also significantly correlated with total organic carbon concentration (ρs = 0.37; P < 0.01), although Cryptosporidium recovery was not (ρs = 0.09; P = 0.52). This suggests that while Cryptosporidium recovery is negatively impacted by smaller particles, which are more often inorganic (i.e., clays, silicates) (2), Giardia cyst recovery may be sensitive to a wider range of suspended particle types and sizes.
Percent recoveries of matrix spike (oo)cysts were used to correct raw counts of wild-type (oo)cysts for observed organism loss during the recovery and enumeration procedures. A comparison of Giardia matrix spike recovery and corrected wild-type Giardia concentrations revealed a significant negative correlation (ρs = −0.35; P = 0.01), i.e., as matrix spike recovery increased, recovery-corrected observations of Giardia concentration decreased (Fig. (Fig.1A).1A). Giardia presence was positively correlated with particle concentration, while cyst recovery was negatively associated with particle concentration. This suggests that higher recoveries are likely from waters of low turbidity, where Giardia is likely present at lower concentrations. However, this relationship may also reflect a tendency of the method to overestimate actual Giardia concentrations when matrix spike recovery is very low. For example, detection of a single wild-type cyst will be interpreted as 10 cysts in a sample with 10% recovery, or 100 cysts in a sample with 1% recovery. It is unclear whether this is an accurate interpretation of results, i.e., whether matrix spikes and wild-type Giardia recovery are sufficiently similar at very low values, as previous comparisons have only compared sample matrices with high recoveries (15). This potential source of systematic experimental error has yet to be addressed and could have major implications for regulatory programs prioritizing actions based on protozoan parasite contamination levels. A similar relationship between matrix spike recovery and corrected Cryptosporidium concentration was not observed (Fig. (Fig.1B1B).
Identified differences in the sizes of particles responsible for reduced recovery by method 1623 may indicate that different mechanisms are responsible for the observed interference and that these microbes may be best enumerated separately. The effects of reduced recovery of seeded (oo)cysts on actual wild-type (oo)cyst concentrations may also differ between Giardia and Cryptosporidium. This is an area of potential systematic experimental error that should be further considered in samples of differing water quality characteristics to ensure more reliable enumeration.
We gratefully acknowledge Adrienne R. Cizek, Jeffrey A. Hayes, Otto D. Simmons III, Steve DiLonardo, and Kerri A. Alderisiio for their various contributions to this project.
We acknowledge the New York City Department of Environmental Protection for funding this study.
Published ahead of print on 14 August 2009.