Use of qPCR-Based Cercariometry to Assess Swimmer’s Itch in Recreational Lakes

The change in cercariae concentrations relative to the time of day, and wind direction relative to the shoreline, was assessed using a generalized linear model with a log link function, based on the negative binomial distribution. The outcome variable was cercariae concentration. Explanatory variables were lake, week (standardized to account for different days during the same week across years), time of day (morning or afternoon), and wind direction (onshore, offshore, and alongshore). In the initial model, the interaction effect of time of day and wind direction was also tested, but this interaction was found nonsignificant and was removed to improve model fit. Additionally, site on the lake, year, and lake were modeled as random effect variables in the original model with independent correlation matrices, but were found to be insignificant and were removed from the final model. All statistical analyses were performed using SPSS (version 24, IBM, Armonk, North Castle, New York, USA). T tests and graphs were executed using PRISM (ver. 7.0, GraphPad).

Mitochondrial cox1 DNA sequences confirmed T. stagnicolae (99.4–100% matches to GenBank vouchers) to be the common schistosome emerging from S. emarginata among Higgins Lake, Crystal Lake, and Lake Leelanau. Miracidia from M. merganser (common merganser duck), Anas platyrhynchos (mallard duck), and Branta canadensis (Canada goose) were collected. As suspected, Trichobilharzia sp. miracidia were found in the feces of mergansers and mallards. A miracidia, which most closely matched to a member of the Schistosomatidae family, was found in Canadian geese. The avian trematode Dendritobilharzia sp. miracidia were also found in the feces of mallard ducks (Table 1) (Brant et al. 2011).

The common merganser has long been assumed to contribute the majority of avian schistosomes to the lakes in northern Michigan (Blankespoor and Reimink 1991), and our survey confirms this observation. We also report Dendritobilharzia spp. miracidia, another schistosome species capable of causing swimmer’s itch in the feces of mallard ducks (Table 1). The qPCR assay in its initial validation by Narayanan et al. (2015) was designed to detect all species of the family Schistosomatidae, which includes Trichobilharzia and Dendritobilharzia spp.

qPCR has become a well-vetted and well-established method for recreational water monitoring for enteric bacteria (USEPA 2012); however, it is still an emerging methodology in parasitology. Molecular methods for the detection of the environmental stages of parasites have become more common in recent years and can be applied to fill crucial knowledge gaps related to the environmental transmission stages. As a review by Bass et al. (2015) points out, environmental sampling for parasites coupled with molecular methods can reduce sampling bias incurred by specimen-based sampling. Additionally, molecular parasitological methods have been demonstrated to be more cost-effective, less labor intensive than specimen-based detection (Huver et al. 2015). However, there are a number of limitations, notably sample inhibition, DNA losses, and difficulties in quantifying to the organismal level, and the inability to distinguish between live and dead organisms. Eukaryotic targets in qPCR can be more challenging to validate because they are multi-cellular, and the 18S rRNA gene target is multi-copy within the genome (Weiss et al. 2017; Bik et al. 2013). Ribosomal gene copy numbers are known to vary by orders of magnitude in bacteria and fungi, and apparent copy numbers quantified by qPCR may also be influenced by downstream methods such as losses due to DNA extraction (Cankar et al. 2006; Adamska et al. 2010; Green and Field 2012). Given this, quantitation to an organismal level for a eukaryotic target must include a conversion between gene copy number and organismal number, which is predicated on an assessment of DNA losses due to extraction. The majority of the Michigan lakes studies were relatively pristine, oligotrophic lakes. Silt and sandy debris in the water samples occasionally presented a challenge, but could be overcome by decanting off the liquid portion of a sample, leaving the debris behind before extraction. Procedures to the DNA extraction protocols may need to be modified if a similar study was replicated in a mesotrophic or eutrophic environment, or in the presence of cyanobacteria or algal blooms.

The average 18S rRNA gene copy number in a single cercaria is 55,851 copies; however, our proposed conversion between gene copy number and cercariae number utilizes the conversion equation and accounts for losses due to DNA extraction. We opted to use a standard curve to predict the number of cercariae rather than using a single number to estimate, as interpolating based on a standard curve better accounts for uncertainty and variability in the target measurement compared to using a single-point measurement. Extraction efficiencies between 0.5 and 4% have been reported for bacteria (Hwang et al. 2012). Therefore, it is not surprising that DNA extraction efficiencies ranged between 1 and 7% (Table 3). These results demonstrate how imperative it is to quantify losses through the extraction procedure in order to avoid underestimations in target concentrations. Other studies that have used microscopic cercariometry to quantify cercariae abundance have found as many as between 200 and 300 cercariae/L in still-water environments (Theron et al. 1978), while typical values in a more dynamic system range typically average 10 cercariae/L (Aoki et al. 2003).

Accuracy of the qPCR methods was tested in a single blinded study. Using the conversion equation and an estimation extraction efficiency of 4.2%, the actual number of cercariae was estimated correctly in each vial, suggesting that under ideal circumstances (i.e., purified cercariae in a clean matrix), this method of estimating cercariae concentrations from qPCR copy numbers is accurate and robust (Fig. 1B). Precision was tested in the field by collecting 3 consecutive 25-L water samples at ten beaches from Crystal Lake in 2016. Variation between samples in this dynamic natural environment is low, with the highest samples exhibiting variation of between 8 and 12 cercariae in between replicate samples. These results demonstrate that the assay provides a precise estimate of cercariae concentration over time in a natural environment (Fig. 1C), but that multiple samples will result in more robust estimation of cercariae concentrations at a particular location.

Laboratory studies have demonstrated that cercariae of Trichobilharzia are positively phototactic and negatively geotactic, and the parasites are typically shed in the morning (Feiler and Haas 1988; Haas et al. 2008). Given these biological determinants of cercariae behavior, it could be expected that avian schistosome cercariae will be more likely to be found during the morning and in the topmost column of the water (Feiler and Haas 1988). Results of the water depth study demonstrated that the majority of cercariae reside in the top 15–20 cm of the water (Fig. 1D), which confirms observations made by others in the laboratory (Feiler and Haas 1988). The time of day experiment shows that on average, more cercariae are found in the water earlier in the day, between 8 a.m. and 12 p.m., and gradually decline throughout the day (Fig. 1E). However, the majority of samples collected on this day were collected during onshore or alongshore winds; therefore, it remained unclear if wind, or the time of sample collection may be responsible for the higher concentrations of cercariae observed between 8 a.m. and 12 p.m.

To better elucidate the biological and environmental drivers of cercariae abundance, a generalized linear model was performed on the entire dataset from 2016 to 2017. Results of the generalized linear model demonstrated the importance of onshore, and alongshore wind in explaining higher cercariae concentrations. Time of sampling (morning or afternoon) did not explain higher cercariae concentrations. There were 2 weeks, corresponding to early July and early August, which also predicted higher numbers of cercariae that could be due to seasonal trends in patent infections among snails (Table 4). Wind conditions likely blow cercariae closer to shore, as such it is intuitive that on- and alongshore winds would increase concentrations of cercariae along the shore. This is intuitive given that our study has also confirmed laboratory experiments that cercariae position themselves in the topmost portion of the water body, and therefore, their movement in the water is likely influenced greatly by surface winds, akin to the movement of surface cyanobacteria scums (Kahru et al. 2007; Kanoshina et al. 2003). Wind direction has also been previously implicated in moving Trichobilharzia cercariae (Leighton et al. 2000). Our model suggests that time of day does not have a strong relationship with cercariae abundance. However, as avian trematodes do leave their snail hosts in the morning, early morning bathers may still be at greater risk of encountering high concentrations of cercariae. There are numerous reports in the literation of strong seasonal trends in snails shedding cercariae, and this likely explains the higher overall cercariae concentrations found in weeks 4 and 8. Numerous studies have reported seasonal trends in patent infections in snails (Crews and Esch 1986; Brown et al. 1988; Gordy et al. 2016). It must be acknowledged that we tested a few factors which might affect cercariae concentrations—however, there are a number of potentially important variables which might influence cercariae concentrations, including water temperature, unpredictable cercariae shedding from snails, and shoreline bird abundance (Lo and Lee 1996; Abrous et al. 1999; Byers et al. 2008; Soldánová et al. 2016). Furthermore, qPCR as a DNA-based molecular method will detect both live and dead cercariae in the water column and therefore may overestimate infection risk in some instances (Bass et al. 2015).

The continued development of molecular testing for environmental stages of parasite development is essential to fully elucidating and understanding transmission and infection risks in parasitic lifecycles (Bass et al. 2015). We hope that our contributions to the avian schistosomes qPCR cercariometry method enable others to continue to elucidate environmental and biological drivers of infection in a variety of host parasite systems. Here we confirm a few previously hypothesized drivers of infection risk, but with this method, a variety of other hypotheses can be tested in the future.

This study demonstrates the utility of these molecular methods more broadly. For instance, analogous methods could be applied to tracking the spatial and temporal changes in Schistosoma mansoni, or Schistosoma japonicum cercariae which cause human schistosomiasis. While qPCR tests have been developed for both of these species, and have been trialed in ambient waters, to our knowledge no comprehensive assessment of cercariae concentrations and conditions that may increase or decrease their concentrations has been undertaken (Hertel et al. 2004; Hung and Remais 2008). Such an assessment is a natural next step for both DNA-based parasitological testing, but also could aid in monitoring the effectiveness of control efforts (Melo et al. 2006).