Detecting and identifying Schistosoma infections in snails and aquatic habitats: A systematic review

Background We were tasked by the World Health Organization (WHO) to address the following question: What techniques should be used to diagnose Schistosoma infections in snails and in the water in potential transmission sites? Our goal was to review and evaluate the available literature and provide recommendations and insights for the development of WHO’s Guidelines Development Group for schistosomiasis control and elimination. Methodology We searched several databases using strings of search terms, searched bibliographies of pertinent papers, and contacted investigators who have made contributions to this field. Our search covered from 1970 to Sept 2020. All papers were considered in a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework, and retained papers were grouped by technique and subjected to our GRADE (Grading of Recommendations, Assessment, Development and Evaluations) evidence assessment profile determined in consultation with WHO. We also considered issues of sensitivity, specificity, coverage, cost, robustness, support needs, schistosome species discrimination, and relevant detection limits. Principal findings Our PRISMA process began with the perusal of 949 articles, of which 158 were retained for data extraction and evaluation. We identified 25 different techniques and for each applied a GRADE assessment considering limitations, inconsistency, imprecision, indirectness, and publication bias. We also provide advantages and disadvantages for each category of techniques. Conclusions Our GRADE analysis returned an assessment of moderate quality of evidence for environmental DNA (eDNA), qPCR and LAMP (Loop-mediated isothermal amplification). No single ideal diagnostic approach has yet been developed, but considerable recent progress has been made. We note a growing trend to use eDNA techniques to permit more efficient and replicable sampling. qPCR-based protocols for follow-up detection offer a versatile, mature, sensitive, and specific platform for diagnosis though centralized facilities will be required to favor standardization. Droplet digital PCR (ddPCR) can play a complementary role if inhibitors are a concern, or more sensitivity or quantification is needed. Snail collection, followed by shedding, is encouraged to provide specimens for sequence verifications of snails or schistosomes. LAMP or other isothermal detection techniques offer the prospect of less expensive and more distributed network of analysis but may face standardization and verification challenges related to actual sequences amplified. Ability to detect schistosome infections in snails or in the water is needed if control and elimination programs hope to succeed. Any diagnostic techniques used need to be regularly verified by the acquisition of DNA sequences to confirm that the detected targets are of the expected species. Further improvements may be necessary to identify the ideal schistosome or snail sequences to target for amplification. More field testing and standardization will be essential for long-term success.

haematobium. The eggs hatch in water and release swimming miracidia that may locate and penetrate an appropriate freshwater or amphibious snail species. Specificity is evident as S. mansoni successfully infects only Biomphalaria snails, S. haematobium develops in Bulinus snails, and S. japonicum in amphibious snails in the genus Oncomelania. Asexual development occurs in sporocysts in the snail host, culminating in the production of fork-tailed cercariae that leave the snail host and swim towards and penetrate unbroken human skin to initiate new infections. Infected snails can produce tens of thousands of cercariae over a period of several months [8].
Schistosomiasis is enabled by poor sanitation, allowing schistosome eggs in feces or urine to pass into snail-containing habitats, and by the widespread use of such habitats for fishing or other occupations, bathing, recreation, washing of clothes, and as a source of drinking water. The long-term persistence of cercariae-producing snail infections in the water renders control more difficult. Even if infected people are successfully treated (usually with praziquantel) to eliminate their adult worms, they may quickly reacquire infections.
The World Health Organization has championed the view that the elimination of schistosomiasis is achievable [9]. In their new roadmap for sustainable development (WHO, 2020), they set a target to validate the elimination of schistosomiasis as a public health problem (defined as <1% proportion of heavy intensity schistosomiasis infections) in all 78 affected countries by 2030. A further goal concerning elimination is to eventually verify and declare the interruption of transmission in a country-by-country manner [10].
Given the importance of development and application of improved diagnostic and surveillance methods for the elimination effort, we have been asked by the WHO to review the suitability of available techniques to answer the following PICO (Problem Intervention Comparison Outcome) question: What techniques should be used to diagnose Schistosoma infections in snails and in the water in potential transmission sites? In other words, what are the best approaches for determining the presence and identity of schistosomes in populations of vector snails (frequently known as xenomonitoring), or of snails, schistosome life stages (like miracidia or cercariae) or their DNA in the water of suspected transmission sites? We did not consider literature pertaining to using human fecal bacteria as surrogates for assessing the likelihood of schistosome contamination of surface waters [11].
Such methods provide a needed alternative view to diagnosis or surveillance of infections in the human host. Snail or water-oriented methods might uniquely detect transmission if, for instance, eggs from non-human reservoir hosts were responsible for infecting snails. Monitoring events in snails or in the water also helps determine if particular schistosome species have been introduced into new areas or re-introduced into former endemic areas, define transmission hot spots spatially and temporally, and gauge the success of intervention methods targeted at the human population [12,13]. The successful effort to control Schistosoma japonicum in China has been aided by monitoring infections in snails and in the water of suspected transmission sites [14]. Considerations important to our evaluation include the points listed in Table 1.
A detailed list of the general diagnostic procedures we considered is provided in the results section. For those who may not be familiar with the general approaches taken over the years to detect and identify schistosome infections in snails or in the water column, we provide an overview in Fig 1 of the available diagnostic approaches that have been undertaken relative to our PICO question.

PRISMA process
Inclusion guidelines. Following the PRISMA guidelines [15], we examined published papers and other materials primarily to do with human-infecting Schistosoma species wherever Table 1. The ideal diagnostic test to answer the stated PICO question should include the following characteristics.
• Be specific in the sense of correctly detecting the target schistosome or snail species in the specific country/geographic area under consideration • Be sensitive such that false negatives are avoided even in areas where low levels of infection are expected (particularly important for elimination considerations) • Take into consideration that infections in snails may be pre-patent at the time of sampling, meaning they have not yet developed sufficiently to produce cercariae • Be robust to using samples from different kinds of habitats or snail species • Take into consideration geographic (and possibly sequence) variability of the target species • Be scalable such that the degree of spatial and temporal coverage is sufficient (and possibly allow for pooling of samples if needed) • Be able to be checked by a method like sequencing to validate identifications • Be periodically subjected to independent quality control • If possible, return quantitative results as opposed to simple present/absent results • Allow for documentation of the sampling effort in the form of specimens that can be contributed to curated museum collections for confirmation of identities of snails or schistosomes examined and for future study • Be fast, not labor intensive nor demanding of high levels of technical skills • Employ standardized sets of reagents and instrumentation to the extent possible • Be robust to variability due to sampling and processing by different lab groups • Not require complex or expensive protocols or instrumentation • Not impose complex storage demands https://doi.org/10.1371/journal.pntd.0009175.t001 Shown are methods for detecting schistosome infections in snails or for detecting schistosome or snail signals in the water. Identification protocols rely on traditional morphological means or, increasingly, sequence-based identifications. With respect to finding infections in snails, the traditional technique of isolating and shedding snails for cercariae is widely practiced, but alternative methods, mostly relying on amplification of schistosome DNA sequences by a variety of means, is also widely practiced. Additional follow up techniques may be required to provide sufficient sequence information for identification. A variety of techniques has been devised to detect schistosomes in the water column ranging from using sentinel mice to collect cercariae from the water to collection of environmental DNA samples containing DNA from intact or disintegrated bodies of schistosomes or their snail hosts. Again, follow-up sequencing may be required to confirm species identifications.
https://doi.org/10.1371/journal.pntd.0009175.g001 they might occur, from 1970 to Sept 2020, from WHO's six official languages. Where relevant papers existed, as with the study of cercarial dermatitis or fascioliasis in snails, we have examined those papers as well.
Exclusion guidelines. We excluded from this analysis papers published before 1970. We also excluded papers that were essentially reviews that did not include new data about diagnostic techniques. We also excluded papers dealing with the diagnosis of schistosome infections in humans for which an extension of the technology to other contexts did not seem likely. We also excluded papers for which the full text was not available or insufficient data were present to evaluate.
Information sources identified. We searched PubMed, Web of Science, Google Scholar, China Academic Journals Full-text Database, Mendeley, and ResearchGate using the combination of search terms listed below. Particularly ResearchGate held papers that might not have been indexed by the major search engines. We also searched the reference lists provided in the articles or reports we found that were most germane to our PICO question. We contacted 11 experts and previous prominent contributors to this literature by email and asked for any unpublished or recent reports not yet available on search engines. We viewed relevant You-Tube videos bearing on our PICO question. By persistently pursuing the "literature cited" sections of the papers we examined and adding papers that had not yet come to light, we eventually did not find any new papers that had not already turned up multiple times in our various search processes.
Search strategy. Each query was executed independently. The search strings were designed using logical operators to encapsulate terms related to schistosomiasis and the various monitoring techniques. In addition, search terms included related parasite species to capture papers that hold methods suitable for schistosomiasis detection in snails or the environment. Trap AND Detection AND Schistosome Inclusiveness. We interrogated databases independent of the language content of the titles and abstracts. The team writing this report has members proficient in Chinese, French, Spanish, Arabic, and English. We examined all possible literature available on our PICO question in these different languages if it existed.
Reference database construction and management. Search results from each query were exported in RIS/BIB/MEDLINE format, merged, and imported into the open source reference manger JabRef [16]. The database was then de-duplicated, using the built-in de-duplication function in JabRef. Some missed duplicates were then manually removed. Records that included authors from the team preparing this report were noted and are highlighted with an asterisk.
Additional sources. References obtained by contacting authors directly were added to the database.
Inclusion/exclusion criteria. We examined the combined non-redundant database of all search outcomes. Despite the specificity of our search strategy, many articles dealt with the detection of schistosome infections in human subjects and were excluded unless some connection to snail or water-related research was evident. Articles dealing with social or economic aspects of schistosomiasis were deemed irrelevant to our question and excluded. Papers describing differentiation techniques of various schistosome species or, in a few cases, snail species were included in the analysis. Any articles presenting a method for the detection of schistosomes in snails or water bodies were included.
Data extraction. Using the predefined template shown in S1 Table, we set out to extract information from all articles included in the analysis by the criteria described above. The full text of each article was downloaded, and additional articles in press or in review were obtained from authors. Interlibrary loan was used to retrieve articles unavailable on the web. The results of the data extraction were agglomerated in a shared spreadsheet accessible to the whole team, and the comments provided reflect our collective input.
Full-text exclusion/inclusion criteria. During the process of data extraction from the full-text articles, some additional articles were marked for exclusion from our final PRISMA tally if it was determined they only dealt with human subjects, no new data were included, or we were unable to obtain the full text.
Determination of certainty of evidence following GRADE criteria. We were asked to determine the certainty of evidence or confidence in effect estimates using the GRADE approach used in clinical medicine and public health policy [17]. The studies we evaluated had different designs, employed a range of techniques, and targeted different schistosome species or sequences for amplification. Because these studies did not focus on a single effect, a formal meta-analysis was not undertaken. Given the lack of data in the randomized control trial format for the studies we reviewed, we consulted with WHO before initiating a modified GRADE evidence assessment using the following criteria: Limitations (risk of bias). These were first considered for each study, then evaluated for all studies grouped together using a specific technique. This involves assessing the use of an unbiased approach for measurement of outcomes, using adequate controls, and asking if there are apparent confounders. We included criteria relevant to our PICO question including an adequate number of samples examined to make a judgment; the adequate number of different conditions tested (water, reagents); quality of methods and equipment used to obtain results (molecular procedures, microscope, gel rigs, etc.); the amount of sample required; the amount of time to process samples; and use of proper controls. This also included questions like if gel bands were sufficiently sharp, bright and measurable, output was quantifiable and comparable between groups and correct identification of schistosomes down to the level desired (species level, hybrid level) was achieved.
Inconsistency. Were results variable across studies (do points of view vary widely between studies), or were there non-overlapping confidence intervals among studies? We also asked if there is day-to-day variability, variability between labs, or if the technique is "finicky" (overly dependent on specific reagents or successful only in a handful of reports).
Imprecision. This considers if data are sufficiently precise to adequately analyze or make a correct conclusion from it, and associated power of statistical analysis. Imprecision refers to the size of the evidence (adequate number of samples, sufficiently narrow confidence intervals).
Indirectness. For our consideration, if the method did not apply to real-world practice, it would be labeled as indirect. For example, if a particular method requires overly complex equipment or hard-to-get reagents. Furthermore, we upgraded the quality of GRADE evidence for observational study results if 1) dose-response gradient data were provided; 2) if the study persistently handicapped itself by the inclusion of the most severe cases needed to document an effect; and 3) if the magnitude, precision or consistency of the effect reported was deemed strong.
In addition to these five GRADE criteria, for our specific PICO question, we also considered some additional points deemed to be relevant, as follows: Sensitivity of detection. For our PICO question, the issue of "sensitivity" needs to be considered in a way different from how it might normally be conceived. For example, it is relevant to know if the method can detect early/immature infections in the snail or light infections. Failure to detect them would be considered false negatives. Molecular methods usually define sensitivity as the minimum amount of DNA needed for a positive result. Other methods that rely on recovering the actual parasites depend on the concentration of parasites per snail or specified water volume. Some methods combine both the collection of parasites and molecular detection limits, such that sensitivity is related to both abilities to concentrate parasite material and the minimum DNA amount possible for molecular detection. We have identified the following categories of "detection limits" as the best way to consider "sensitivity" relative to our task.
For snail-based techniques: High sensitivity-ability to detect snail with one parasite at 1-day post-infection, or the ability to detect prepatent infections two weeks or younger in duration; for pooled samples, one positive snail among a pool of 100 snails. Marginal sensitivity-the ability to detect prepatent infections >2 weeks old or patent infections. Inadequate sensitivity-inability to reliably detect snails with patent (shedding) infections.
For water samples. High sensitivity-ability to detect the equivalent of one cercaria or less in a 20-liter water sample. Marginal sensitivity-ability to detect the equivalent of 5 cercariae per 20-liter water sample. Inadequate sensitivity-inability to detect > 5 cercariae (or equivalent DNA amounts) in a 20-liter sample of water. This information was not reported in all the studies we evaluated. Therefore, we had to base our judgments about a particular technique on those studies that did specify such information or used other parameters as a proxy to judge the sensitivity, such as the amount of DNA, or amount of starting material needed for a positive signal.
Species differentiation. We considered if the technique is specific enough: to tell species within a species group apart (for example, S. mansoni from S. rodhaini, or S. haematobium from S. bovis); possibly to tell hybrids from non-hybrids, and to not provide false positives for snails with infections with non-schistosome trematodes. Each method can be judged as: Monospecific-verified to detect the target species in question only (e.g. only S. haematobium); Narrowly heterospecific-detects the target species but cannot differentiate close relative (e.g. haematobium, also detects other members of S. haematobium species group like S. bovis or S. mattheei); Heterospecific-records Schistosoma of multiple species groups (e.g. S. mansoni and S. haematobium); Nonspecific-not clear from the literature if the method can reliably detect Schistosoma species at the exclusion of other digeneans species or other organisms with ample confidence.
Coverage. How much labor is needed to permit the collection, preparation, and analysis of samples required for a specified technical approach? Can adequate coverage of habitats be achieved to facilitate sound judgment?
Cost. Cost involved for specialized training, reagents, computer analysis or processing time. In most cases, the cost was provided by the authors in the paper. When that was not discussed, we compared the prices of reagents required to current known prices. In complex methods, we considered the cost of equipment and reagents needed to execute the particular method. Support needs. Requirements for extensive lab space, rearing facilities, electrical service freezers, distilled water, specialized instrumentation, and service contracts were considered.
Pros and cons. As a more intuitive overview of our deliberations, we include a table that highlights the perceived pros and cons of each technique we considered.

PRISMA flow diagram
The Prisma diagram (Fig 2) provides an overview of how we arrived at the final 158 articles included in our systematic review process.
Papers reviewed, and our extraction summaries S1 Table provides the 158 articles along with our extraction summaries that were considered in our GRADE assessments. Three of the articles we examined were authored by one or more members of our team, so an asterisk has singled out these articles in the table.

Determination of certainty of evidence following GRADE criteria
We grouped the techniques covered in the 158 papers into 25 categories ( Table 2). Some papers discussed more than one technique. As noted in Fig 1, the techniques cover a wide range of approaches. Some are directed towards detecting schistosome infections in snails, and some involve the detection of schistosomes (or snails) in water samples. In some cases, like shedding of snails, identifications of schistosomes can be undertaken using morphological criteria or by submitting them to molecular techniques. For some methods like eDNA, the   Table. We used the numbers 0, -1, -2 to score every method, with 0 being adequate, -1, serious concern, -2 very serious concern. It is worth noting that publication bias is not part of this table, as we did not have any reason to suspect it has occurred with any of the methods. We defined species differentiation ability of each method as N.S = Nonspecific, H.S = Heterospecific, N.H.S = Narrowly Heterospecific, M.S = Monospecific. And relevant detection limits as Hig.S, High sensitivity, Mar.S = Marginal sensitivity, Ina.S = Inadequate sensitivity.

Method
Number of studies

Relevant detection limits
Direct Shedding 20

Water-based detection
Filtering then direct exam of filter process typically involves collecting a sample followed by extraction and the submission of the sample to some type of amplification protocol. The data supporting the scores provided in Table 2 are provided in S1 Table. The five criteria contributing to the overall certainty in the grade score provided were limitations, inconsistencies, imprecision, indirectness, and publication bias. As we did not detect the latter, it does not appear in the table. The remaining five columns (coverage, cost, support needs, species differentiation, and relevant detection limits) reflect our view of their importance, but they were not included in the GRADE scores. Note that by the GRADE criteria provided, many techniques were assessed as having very low or low overall certainty: in several cases, they reflect techniques reliant on older technology. Three techniques were scored as providing moderate certainty: LAMP, eDNA based techniques, and qPCR. One of the significant considerations diminishing confidence for several of the techniques was the lack of widescale testing and standardization considerations. In S2 Table, following GRADE criteria, we present determinations of sensitivity and specificity using the definitions often applied to diagnostic techniques in the medical literature (see table legend for definitions). Insofar as most of the papers we reviewed simply did not provide this information, its relevance to our discussion is lessened. Lastly, in Table 3, with an eye on practical guidance, we provide an overview of the pros and cons of each of the techniques we classified.

Discussion
Successful application of any diagnostic technique for the determination of schistosome infection in snails or in the water of natural snail habitats requires accurate knowledge of the species of schistosomes and their snail vectors historically or currently present. A single schistosome and snail species might be involved in some locations, but often multiple Schistosoma or potential snail host species are present. For instance, S. rodhaini might complicate the diagnosis of S. mansoni in some parts of Africa, but not in South America. S. haematobium may be especially hard to differentiate from closely related animal schistosomes or their hybrids in some sub-Saharan Africa areas. Multiple species of bulinid snails might be implicated to one extent or another as vectors. Some snail species not susceptible to human schistosome infections like Planorbella duryi can easily be mistaken as vector species upon superficial examination [18], potentially resulting in misdirected sampling effort. Familiarity with the literature for a specific area, consultation with parasitologists or malacologists, and examining available keys are all helpful. Submission of voucher specimens of schistosomes and snails to an appropriate museum both for subsequent identification and to provide historical documentation should be encouraged.
Another fundamental reality is that the scale of the task required of the diagnostic technique will vary dramatically. The problems posed for ascertaining if and where schistosomiasis transmission might be occurring on a Caribbean island or in a desert country with few transmission sites are quite different compared to some countries where schistosomiasis transmission potentially occurs across vast geographical areas in almost innumerable freshwater habitats. Clearly, such situations may require different sampling and diagnostic approaches.
Especially in the context of verifying elimination, an all-important consideration is the habitat sampling protocol to be used, a topic requiring much further collective discussion. Given that schistosome transmission is often focal, sampling effort should be guided by available information about past transmission foci and any new evidence (such as positive serology results from recent human surveys) providing relevant locality data. One general goal though, should be to generate many samples from different locations and time points to avoid the possibility of overlooking new or unknown foci of transmission, possibly including ones involving

ELISA/ Immunodetection
-Lends itself to a highthroughput format, which is established for many ELISA assays -Can be very sensitive and specific in particular contexts -The ability to distinguish among closely related schistosome species has not been rigorously tested -The needs for ELISA reader and specialized monoclonal antibodies would likely preclude use in most field labs -Not a lot of current technique development -Many improvements are theoretically possible, but it seems unlikely ELISA-based techniques will soon displace sequence-based detection methods with the ability of the latter to confirm species identities -It is possible particular antigens characteristic of schistosomes will prove to be abundant and stable in snails or water samples to warrant further ELISA-based detection

Biochemical Analyses
-Some approaches could be adapted to high-throughput -Lacks specificity needed to be sure of parasite species identification -Cumbersome and expensive for most field labs -General lack of any widespread field testing or current development

RAPD PCR
-Provides a distinctive profile of randomly amplified bands that might enable more specific detection of a certain species -Can be sensitive -Less likely to produce consistent band patterns, especially across broad geographical regions -Identity of amplified bands is unknown unless followed by explicit sequencing -Largely replaced by more repeatable primer specific driven amplification -It seems unlikely this approach will receive additional development to enable standardization and simplification -In light of more specific primer and sequence driven approaches, it seems unlikely RAPDs RCR will play a major role in future diagnostic efforts

Repeat Sequence PCR
-Sensitivity is high, at least in some cases as with S. haematobium group -Highly repetitive sequences have an inherent appeal because of their abundant representation in the schistosome genome -They have been used by more studies than any other PCR targets -Depends on the availability of thermocycler and specialized reagents -The same considerations that apply to conventional PCR -Specificity maybe not sufficient for distinguishing among closely related species, but workarounds are being devised -May also be a need to check for distantly related but nonetheless cross-reacting species -This approach has been used in real-life surveillance context, but some additional specificity testing in demanding field contexts is warranted -Compatibility with eDNA based sampling needs to be checked Nested PCR -Very sensitive because of the nested approach -Specific because of a nested approach -Depends on the availability of thermocycler and specialized reagents -Cost and time per sample higher than with conventional PCR because of the need for two successive runs -Contamination a concern -Still a need for periodic verification by sequencing -It seems unlikely nested protocols will be needed to achieve desired sensitivity or specificity, so further development is not expected, especially in view of probable further development of ddPCR   animal reservoirs. Consequently, whatever methods will be chosen to detect positive snails or water samples should be compatible with efficient and low-cost sampling methods that provide realistic coverage to support an accurate assessment of achievement of elimination. Spear et al. [19] noted, "the development of environmental diagnostics for the infective stage of parasites such as schistosomes is stunningly behind those of other parasites present in water." In Table 1, we identified basic features required of the ideal diagnostic test to answer the stated PICO question. Although the ideal technique does not exist, considerable strides have been made in recent years in devising sensitive and specific detection methods and providing more straightforward methods for extractions, amplifications, and documentation. We are making progress.

Filtration then molecular characterization
-See eDNA above as filtration is usually the collection method for this technique -Actual costs of filters and apparatus can be very low with low-tech gravity filter set-ups -Filtrates lend themselves to extraction and further molecular processing -Samples may be inconsistent in quality because of variation in water turbidity from locations to location -Is labor-intensive and cumbersome to collect water samples and to filter them -Coverage limited by time and labor -Has already become the first step in eDNA-based molecular protocols -More needed to decide on volumes of water to filter, and development of less laborintensive collection methods -See eDNA above https://doi.org/10.1371/journal.pntd.0009175.t003 As we move forward, we may learn that the application of a combination of diagnostic approaches is inescapable. For example, although frequently maligned, the snail shedding technique has some benefits that may be impossible to replace with even the most sophisticated molecular technique. For one thing, it is specimen-based, providing indisputable evidence about the presence of a particular snail vector species and possibly of the schistosome species it is transmitting. The specimens, or portions derived from them, can be used in follow-up molecular analyses and can be vouchered to become part of a more permanent record and facilitate their use in applications we cannot yet anticipate. The shedding method can also provide spatially and temporally explicit information from multiple locations about how common schistosomes might be. It could be argued that familiarity with this method should be part of the training of any person working in schistosome environmental diagnostics, simply to reinforce an understanding of the underlying biological realities of the system.
One of the most promising recent developments in schistosome environmental diagnostics is to use as starting material schistosome or snail eDNA as derived from water samples. Suppose eDNA samples are shown not to rapidly decay [20], not to be prone to inhibition and enable sensitive, specific, repeatable analyses across multiple aquatic habitat types.. In that case, it is easy to imagine eDNA as the preferred method of sample collection, perhaps supplemented with occasional snail specimen collection for validation purposes. One great appeal of eDNA samples is the potential for rapidly collecting many samples, each of which effectively integrates a signal over space and time, providing further coverage. Additional improvements to facilitate the collection of eDNA samples are sure to follow, further underscoring this approach's value.
Once a DNA sample is in hand, then most modern diagnostic approaches proceed to amplify DNA via an ever-growing variety of techniques. This field is moving fast, and what seems best in 2021 may soon be dated. As noted in Table 3, there are pros and cons to each. One notable development to go along with the increasing use of eDNA is qPCR-based technology to analyze extracted specimens [20][21][22][23][24]. A quenchable fluorescent probe designed to detect a particular sequence of the target organism of interest is typically included (FRET-qPCR). Significant general advantages for qPCR are that amplification and detection are combined in a single step, it is a mature technique with well-defined protocols and data analysis, it is relatively quantitative in allowing the amount of amplified product to be determined (but requires reference samples and development of a standard curve), costs are relatively low per sample, and it can be set up for high-throughput and multiplexing (perhaps simultaneously detecting schistosome and snail signals in the same sample). It is supported commercially by the active development of new products. Smaller, less expensive qPCR machines are commercially available but are less amenable to high-throughput sampling. Droplet digital PCR has some key advantages over qPCR should eDNA samples prove difficult to analyze: ddPCR is much less subject to inhibition, sensitive to detection of rare targets, and provides absolute quantification.
qPCR and ddPCR offer many advantages, but acquiring and maintaining the instrumentation, procuring reagents in a cost-effective manner, and standardizing protocols among separate labs would remain a challenge. To avoid these problems, and to take advantage of purchasing reagents in bulk and using high-throughput protocols, submission of relatively stable eDNA/DNA samples to one or a few central facilities charged with the responsibility of extracting and amplifying the samples and interpreting the data seems to us to be the way forward. A strong case could also be made for LAMP or other isothermal techniques, including the possibility of having a more widely distributed system of labs employing the technique.
No matter the means of sampling or the technique chosen for detection, some key general points remain. It would be highly desirable from a global elimination program's standpoint to ensure that the techniques employed have been widely tested on samples collected from many different habitat types on other continents. Standardization with respect to extraction methods, particular reagents or kits used, means to test for or prevent inhibition of detection reactions, amplification protocols, and data interpretation and presentation are all highly desirable.
In the case of amplification-based techniques such as PCR/qPCR strict precautions must be taken to prevent contamination of the lab environment with amplicons of the target species to avoid the recording of false positives. The inclusion of both positive and negative controls and standards in all protocols is essential. Aliquoting of key reagents is vital to minimize contamination risk. Physical separation of separate steps in preparation of amplification reactions, if possible, is also essential. If qPCR, ddPCR, or other complex equipment prone to noise-to-signal inaccuracies are to be involved, regular and careful calibration of the machines is critical.
With respect to testing the primers used for various techniques, to avoid misleading results, there is a need to be sure 1) the closest and most common relatives of the target species are tested, and 2) that spurious cross-reactions even with parasite species that are not close relatives of the target species are excluded as a possibility. Testing against local co-occurring snail and trematode species is preferred as these are the species likely to cause cross-reactions and false positives to occur [22]. Having additional genome sequences for the trematode species most likely to co-occur with targeted schistosome species, either in the same snail species or the same environment, would be helpful. It would help eliminate the possibility of designing primers based on schistosome genes or genomic repeats that quite unexpectedly share sequence similarity with other trematodes. Extraneous trematodes may be more of a diagnostic complication for detecting schistosomes in snails than presently envisioned because of the tendency of groups like echinostomes, xiphidiocercariae, and strigeids to encyst as metacercariae within field snails. Similar considerations apply to eDNA samples taken from environments supporting diverse trematode faunas, which is often the case.
Primer design often relies on in silico predictions to limit cross-reactivity, which may have pitfalls [25], underscoring the need for follow-up confirmation. Sanger or high-throughput amplicon sequencing with comparison to sequences in GenBank [26,27] will be particularly crucial during elimination and surveillance operations to ensure positive results that are recorded by a band, Ct value, or a color change in a tube definitively belong to the schistosome or snail species responsible for the positive sample. Cross-reactions could lead to erroneous conclusions about the persistence of transmission or lead to misidentification of hot spots, in turn leading to erroneous targeting of special control efforts and wasting limited resources.
It is also important to rule out the effect of inhibitors in diminishing or precluding amplification. Inhibitors can originate from the sample or can be introduced during extraction. The standard method to deal with inhibitors is through control reactions spiked with known amounts of a reliably amplifiable non-target sequence. By comparing the quantity of non-target detected to the originally spiked amount, the level of inhibition can be determined in any given sample. This helps to guard against obtaining false negatives with unknown samples. If the presence of inhibitors is deemed a persistent problem, then sample treatments may be required, or the use of ddPCR may then be preferred.
Regarding the molecular targets that have been used for schistosome detection (see [28] for one list), they are quite understandably typically based on sequences expected to be represented many times in a single schistosome cell, thereby increasing the ability to sensitively detect a signal. They include mitochondrial markers (such as 16S rRNA, Cox1, Cox3, Nad5 (ND5) [29]), or nuclear sequences including portions of the rRNA complex (18SrRNA, intergenic 28S-18S spacers) and repetitive sequences such as Sm1-7 and DraI [12]. Many of these sequences have also been used to detect schistosomes in human samples such as plasma [30], however snails and environmental samples pose added challenges due to presence of a multitude of potential competing sequences, that as noted above might have surprising homologies with schistosome target sequences. Certainly, there is additional scope for development of better target sequences that might avoid cross-reaction or resist degradation, but it is also important to more thoroughly test the primers we have. One recent comparative study found primer sets targeting ND5 or 28S to work better than primers targeting a repeat element [31], the latter potentially also favoring amplification of unknown products from other complex genomes also present in extracted samples.
Particularly needed for eDNA-water based approaches are studies comparing different primer sets under a diversity of realistic field conditions in which known quantities of schistosome specimens or DNA are present, preferably undertaken by the same research team. Such controlled comparative studies should include considerations of how eDNA moves and degrades over time. Included in the comparisons would be considerations of different extraction techniques and downstream amplification protocols such as LAMP and qPCR. An important consideration is tailoring primers to the amplification method: gel-based PCR detection can accommodate long amplicons which can be made highly specific, however qPCR and ddPCR have better performance using shorter amplicons in the 70-200 bp range, limiting the potential regions to amplify. Considering the plethora of genomic resources [32] and molecular targets already found for schistosomes, it is likely that the limiting factor to the identification of the best targets to use is the lack of comprehensive comparative studies rather than the lack of useful molecular targets.
It will be tempting to pool samples, especially when in a surveillance program, it becomes clear that most samples collected and analyzed prove to be negative. Pooling could allow expanded coverage and sharply lower expenses. For some snails like the tiny Oncomelania involved in S. japonicum transmission, this may prove more feasible, as some of the papers examined suggest (one cercaria can be detected per 1000 uninfected snails). For larger snails like Bulinus or Biomphalaria, pooling may prove more complicated, especially if rare and/or prepatent infections are represented among the pooled snails. Recognizing that a trade-off is involved in the pooling of samples, one way to check the ability of pooling to detect small schistosome signals is to spike some of these samples with various doses of parasites to see if they reliably detect positives. Similar considerations apply to pooled water samples.
For studies based on detection of eDNA from habitats, it is important to recognize that positive signals from filters could be derived from cercariae that were intact and alive at the time of sampling or from environmental DNA no longer associated with living cercariae or any other life cycle stage. Positive signals from both sources are valuable but tell different stories. If positive signals from living cercariae are being detected, then the techniques convey valuable information that infections are ongoing and where they are being acquired. If the signals originate from eDNA, they are more indicative of a signal integrated over time. The length of the persistence of eDNA signals is bound to be variable, from days to weeks [33], which will influence the extent to which signals may be distributed within a habitat. Developing techniques to detect eRNA, indicative of the presence of living specimens, may prove to be useful in some contexts.

Conclusions
Some important knowledge gaps going forward are: 1. We are not necessarily there yet with respect to identifying the best schistosome sequences to target for amplification. The optimum length of the target to be sequenced may well depend on the method of amplification. Further comparative studies to assess existing primers [31] and more testing to improve specificity with respect to the realistic identification challenges faced in a particular location are needed.

2.
To field-test the most promising sample collection, extraction methods, primers, and detection methods in multiple locations, possibly with standardized spiked samples. This would be undertaken preferably in a coordinated, comparative fashion, under centralized supervision such that quality control and standardization can be promoted. This would help identify ways to reduce costs and identify the most suitable combination of techniques for broader application.
3. Further emphasis on the development of eDNA based methodology is warranted to determine: temporal stability of amplifiable signals in the water column; how far eDNA might be dispersed and still be detected; if robust and consistent signals be retrieved from complex tropical aquatic ecosystems, and; if the samples be easily shipped elsewhere for analysis?
4. Ideally, our diagnostic tool kit would eventually include approaches that might allow differentiation among live or dead specimens and different life cycle stages of the same species, for instance, differentiating between miracidia (indicative of contamination) or cercariae (indicative of the potential for infection) and determining if infections in snails derive from humans or reservoir hosts.

5.
We must remain open-minded for dramatic new approaches that rely on robotics for sample collection/processing or for self-contained microfluidics devices that, if proven effective and made inexpensive, could solve many of the coverage and standardization problems.
6. Inducements to encourage more researchers to participate and communicate with one another to develop standardized new diagnostic techniques should be provided.
Supporting information S1 Table. Criteria used for the evaluation of each reference, and the extraction summaries for each reference evaluated-presented in Excel format.
(XLSX) S2 Table. Specificity here is defined as the ability of the method to correctly identify specimens without signal/infection (true negatives); Sensitivity, the true positive rate (specimens with infection/signal); Sensitivity to input: In most of the studies we looked at this was the proxy used for sensitivity as the minimum amount of sample required which in turn determines the ability of the method to detect true positives. This is reported as either DNA amount or number of cercariae. n = to the number of studies used to obtain the numbers reported. (DOCX)