The overall goal of this project was to produce an efficient, broadly applicable set of protocols for the use of Environmental DNA (eDNA) techniques for monitoring sensitive aquatic vertebrate species and their invasive threats at Department of Defense (DoD) installations. Specifically, the research team sought to develop and validate eDNA protocols for a variety of aquatic target species (both species of concern and invasive species that threaten their persistence) across a range of environmental conditions. This project was a field-scale demonstration of eDNA techniques.
Performance objectives for this project included demonstration of an eDNA sampling and analysis protocol that 1) has a high probability of detecting species when they are present, 2) has a higher probability of detecting rare species than field surveys, 3) can detect amphibian disease when present, 4) minimizes the probability of contamination, 5) is cost-effective, 6) is easy to use by technician-level workers, 7) is applicable to different aquatic species, and 8) reduces permitting requirements for surveys of protected species. An additional performance objective was to demonstrate the ability of an empirical model to predict success of eDNA methods based on correlation of environmental factors with probability of detecting eDNA.
Environmental DNA methods use the trace DNA that all organisms leave in their environments to draw inference to the presence of target species. For aquatic species, a water sample can be collected from waterbodies in which the target species may occur. DNA is then extracted from the sample and analyzed using quantitative polymerase chain reaction (qPCR) to determine whether and how much of the target species’ DNA is present in the water sample.
Detection of target species involves two phases: eDNA test development and eDNA test application. Test development requires several steps: 1) identifying the target species for the test; 2) collecting sequence data from either databases or DNA sequencing; and 3) creating the test and verifying its specificity and sensitivity. Environmental DNA tests for target species are species-specific, and test development occurs once for every new species to be monitored.
Once the test has been developed, the application phase involves the collection and analysis of water samples. Replicate samples are collected from each monitoring location and analyzed with the validated eDNA qPCR test to detect target species. The resulting data are the basis for a survey or monitoring program. Replicate samples are used to estimate detection probabilities, which are used as a measure of uncertainty in an occupancy modeling framework. This framework allows researchers to detect changes in occupancy over time, as well as providing detailed data on the probability that the target species was present but not detected at each location. Additionally, covariates of the probability of occupancy and detection can be included to provide more accurate estimates of occupancy. If detection probabilities are not high enough to provide necessary levels of certainty, the results of these models can also be used to adjust sampling design spatially and temporally to maximize detection probability. This latter approach was taken during this demonstration project to improve initial occupancy estimates. However, many of the findings of this demonstration project have been incorporated in sampling recommendations that can be applied across systems, potentially negating the need for this optimization step in many systems.
The research team demonstrated environmental DNA methods for detecting aquatic vertebrates and pathogens at three DoD installations that host an array of at-risk aquatic species and invasive threats: Fort Huachuca, Arizona; Eglin Air Force Base, Florida; and Yakima Training Center, Washington. Each site hosts aquatic species of immediate management concern to site managers and represents a different type of aquatic system. This demonstration showed that eDNA can be a sensitive and cost-effective technology for monitoring aquatic species under a range of conditions that included factors expected to limit eDNA detection (e.g., low pH). The research team fully or partially met seven of the eight quantitative and qualitative performance objectives.
Target species detection with eDNA protocols
In the oak woodland and desert grassland wetlands at Fort Huachuca, researchers developed and applied eDNA protocols for detecting the federally endangered Sonora tiger salamander (Ambystoma tigrinum stebbinsi) and federally threatened Chiricahua leopard frog (Lithobates [Rana] chiricahuensis). They also developed and applied eDNA tests for threats to these species, including the invasive American bullfrog (L. catesbeianus), non-native barred tiger salamander (A. t. mavortium), amphibian pathogenic chytrid fungus (Batrachochytrium dendrobatidis; Bd) and salamander-specific iridovirus (Ambystoma tigrinum virus, ATV). They collected water samples for eDNA analysis concurrently with ongoing Arizona Game and Fish Department field surveys for Sonora tiger salamanders and Chiricahua leopard frogs. In the first year of sampling they identified different limiting factors for detection of each species (sample volume for Sonora tiger salamanders, wetland area for Chiricahua leopard frogs, and water temperature for American bullfrogs). In the second year they adjusted sampling protocols to increase sample volume for salamanders and increase spatial sampling at larger wetlands for leopard frogs. With improved sampling designs, they detected target species at almost all sites where field crews detected them, plus 19 additional detections with eDNA methods only.
The eDNA tests for ephemeral wetlands at Eglin Air Force Base included federally endangered reticulated flatwoods salamanders (Ambystoma bishopi), the amphibian pathogen (Batrachochytrium dendrobatidis; Bd), and ornate chorus frogs (Pseudacris ornata), a Florida Species of Greatest Conservation Need. The project team joined with researchers from Virginia Tech to conduct concurrent eDNA water sample collection and field surveys for flatwoods salamanders and ornate chorus frogs. Initial results indicated that pH was the limiting factor for detecting flatwoods salamanders and ornate chorus frogs in Eglin’s acidic, spatially complex wetlands. They adjusted the sampling protocol for these species by increasing sample volume and, at sites with pH < 5, sampled more locations within the wetland. With the adjusted protocol they detected flatwoods salamanders and ornate chorus frogs at almost all sites with field detections and two additional sites where field crews did not detect them, but failed to detect target species at two sites in which field crews found them at very low densities.
In streams on and near Yakima Training Center, the eDNA tests included the federally threatened bull trout (Salvelinus confluentus), its non-native threat brook trout (Salvelinus fontinalis), and Chinook salmon (Oncorhynchus tshawytcha). For samples in which Chinook salmon were detected, the research team applied additional tests targeting single nucleotide polymorphisms to determine the probability that the Chinook salmon present were from the federally endangered Upper Columbia Spring Chinook salmon evolutionarily significant unit (ESU), rather than the non-listed Upper Columbia Summer-/Fall-run Chinook Salmon ESU. They detected bull trout in all sites at which field crews detected the species. The eDNA tests were successful for detecting brook trout in five sites, but because no field surveys were conducted during the demonstration, they were unable to compare detection probabilities for this species. The research team detected spring-run Chinook salmon in all but one location where they were thought to be located.
Field sample collection protocol
The research team developed, tested, and iteratively revised a field protocol for collecting water samples for eDNA analysis. They evaluated the ability of field staff to correctly follow the field protocol using a qualitative survey of experienced and technician-level biologists working in a range of lentic and lotic settings in Idaho, Washington, Florida, Arizona, and New Mexico. The response scores for the final version of the protocol were 4.5 or greater on a 5-point Likert scale for ease of use and versatility, demonstrating that the protocol is easy to follow in field settings and applicable across different types of aquatic systems.
The research team successfully demonstrated the cost-effectiveness of eDNA surveys for Sonora tiger salamanders at Fort Huachuca and flatwoods salamanders at Eglin Air Force Base (AFB), the only species for which both field costs and field detection probabilities were available. eDNA sampling includes front-end costs for developing and validating qPCR assays for target species in the initial year of sampling, and ongoing costs of collecting and analyzing samples in subsequent sampling periods.
For Sonora tiger salamanders, costs for the initial season of eDNA sampling, including all front end costs, was $13,629 for a single survey of 20 sites, compared with $5,582 per survey for current seining surveys. One eDNA survey or two seining surveys would be needed each year to achieve a detection probability of ≥0.95. Ongoing surveys for tiger salamanders after the initial year would cost $5,459 per year for eDNA sampling or $10,714 per year for seining surveys.
For flatwoods salamanders, the initial year of sampling cost $16,520 for one eDNA survey and $4,460 for one survey using current dipnetting methods. One eDNA survey or four dipnet surveys would be needed each year to for detection probabilities ≥0.90. Costs for ongoing flatwoods salamander surveys after the initial year would be about $3,698 per year for eDNA sampling or $4,460 per year for dipnet surveys.
The following technology implementation issues were encountered during this demonstration and may be anticipated to affect future applications of eDNA methods:
Regulatory issues – Environmental DNA sampling requires only collection of water samples similar to water sampling for other types of monitoring such as water quality parameters. The research team expected that permits under the Endangered Species Act would not be required for an eDNA-based monitoring program for listed species. However, permitting requirements are species-specific and may vary across regions. Although U.S. Fish and Wildlife Service staff in several states concurred that permits would not be required for eDNA water sampling, they were not able to receive consensus that eDNA surveys would have lower permitting requirements than conventional field surveys for aquatic species.
Spatial issues – The presence of a species’ eDNA at a location may not indicate the presence of the species. There are two main examples of this: 1) in streams, eDNA may flow into an installation from an upstream source; and 2) eDNA may be deposited from an allochthonous source such as animal movements. eDNA samples should always be collected and analyzed as replicates to be able to understand the strength of evidence for presence of a species at a site. The detection of a large amount of eDNA in a sample is stronger evidence for species presence than one sample with a small amount of eDNA detected of the target species. For some applications (e.g., the detection of aquatic invasive species), several strongly positive eDNA samples may need to be found to trigger management actions if field surveys cannot locate the species.
Temporal issues – eDNA degrades quickly in water and the research team demonstrated that degradation rates are strongly influenced by temperature, ultraviolet (UV) radiation, and acidity. The period of time eDNA remains detectable likely ranges from a few days in warmer, sunnier systems to a few weeks in sites with colder water and less UV exposure. The sampling design should target the period in which the species is most likely to be present at the highest density. Because eDNA in sediment can persist up to thousands of years, sediment samples should only be analyzed if the question of interest is whether the target species has ever been present at the site.
Procurement issues – Midway through the demonstration, the filter funnel the research team used for water collection was discontinued by the manufacturer. They tested other available filters with different combinations of filter pore size and material to see what types of filters would be adequate for capturing eDNA from freshwater samples. In the test, filters made of mixed cellulose ester, cellulose nitrate, or polyethersulfone materials had similar effectiveness at capturing and efficiently filtering eDNA, while polycarbonate track-etched filters had lower eDNA capture and efficiency. Filter pore size did not influence the amount of eDNA captured, but smaller pores were generally associated with longer filtration times, and filters with larger pores are likely to be more efficient in systems where smaller pore size filters are likely to become clogged.
An additional major issue remains for installations wishing to implement targeted eDNA sampling of species. There is still a very limited number of labs offering analysis services for these samples, and no certification program to ensure the quality of the analysis. To address this concern, we developed guidelines for practitioners interested in partnering with laboratories to analyze eDNA samples. Additionally, we worked with eDNA researchers from around the world to develop a set of best practices for eDNA laboratory analyses.
Sampling considerations - The research team found that spatial sampling design, sample volume, and filter characteristics can have a strong effect on detection probabilities. Practitioners who are implementing eDNA sampling for a novel species or system should consider these factors carefully in determining an appropriate sampling approach. For situations in which eDNA sampling approaches have been successfully demonstrated for similar species and aquatic systems, practitioners can use the sampling design from existing applications to inform sampling, which will likely result in similar detection probabilities. For applications in new types of systems, a pilot survey can help practitioners evaluate whether the eDNA sampling strategy detects the target species with sufficient accuracy and sensitivity to meet survey objectives. Concurrent field sampling can be useful but is not necessary for initiating an eDNA sampling program. This comparison allows managers to select the most efficient and reliable survey method, whether that may be field surveys, eDNA surveys, or an integration of the two methods.