Repurposing environmental DNA samples—detecting the western pearlshell (Margaritifera falcata) as a proof of concept

Abstract Information on the distribution of multiple species in a common landscape is fundamental to effective conservation and management. However, distribution data are expensive to obtain and often limited to high‐profile species in a system. A recently developed technique, environmental DNA (eDNA) sampling, has been shown to be more sensitive than traditional detection methods for many aquatic species. A second and perhaps underappreciated benefit of eDNA sampling is that a sample originally collected to determine the presence of one species can be re‐analyzed to detect additional taxa without additional field effort. We developed an eDNA assay for the western pearlshell mussel (Margaritifera falcata) and evaluated its effectiveness by analyzing previously collected eDNA samples that were annotated with information including sample location and deposited in a central repository. The eDNA samples were initially collected to determine habitat occupancy by nonbenthic fish species at sites that were in the vicinity of locations recently occupied by western pearlshell. These repurposed eDNA samples produced results congruent with historical western pearlshell surveys and permitted a more precise delineation of the extent of local populations. That a sampling protocol designed to detect fish was also successful for detecting a freshwater mussel suggests that rapidly accumulating collections of eDNA samples can be repurposed to enhance the efficiency and cost‐effectiveness of aquatic biodiversity monitoring.


2011), salamanders
, mollusks (Goldberg et al., 2013;Stoeckle, Kuehn, & Geist, 2015), crustaceans , mammals (Padgett-Stewart et al., 2015), lamprey , and bony fishes (Brandl et al., 2015;Mahon et al., 2013;. Although quantitative (qPCR)-based eDNA sampling typically targets single species, each sample has the potential to provide multispecies occupancy data because a single sample potentially contains DNA of all animals present at or upstream from a location (Thomsen et al., 2012;Valentini et al., 2016). Given the dramatic and recent increase in eDNA surveys for single invasive or rare native species, eDNA sample collections are rapidly accumulating. Using eDNA to detect a single species typically only requires a portion of the total sample. Thus, if samples are properly preserved, archived, and annotated, these eDNA collections represent a trove of biodiversity data accessible at relatively low cost that can be repurposed for additional species.
One rapidly increasing collection of eDNA samples is associated with a range-wide survey for bull trout (Salvelinus confluentus ;Young et al., 2017). In this survey, environmental DNA samples were collected by dozens of collaborators throughout the Pacific Northwest and sent to the National Genomics Center for Wildlife and Fish Conservation, at the USDA Rocky Mountain Research Station, Missoula, MT. Here, the samples were extracted, analyzed for bull trout, archived, annotated, and stored in a central repository along with all metadata including sampling dates and locations. The bull trout detection results were then uploaded onto a publicly accessible database (https://www. fs.fed.us/rm/boise/AWAE/projects/BullTrout_eDNA/SurveyStatus. html) that has a user-friendly interface allowing users to extract data associated with specific sampling locations. While this archive provides sample coverage across an extensive area, and therefore the potential to determine the occurrence of many species, the efficacy of repurposing eDNA samples is largely unknown, particularly when sampling strategies were initially designed to maximize detection rates for organisms with very different life histories.
Freshwater mussels are a diverse group of organisms with roughly 300 species native to North America and represent one of the most highly endangered and rapidly declining fauna on Earth (Haag, 2012).
These extinctions and declines are often attributed to anthropogenic activities that impact water quality and foster the establishment of non-native species (Bogan, 1993;Williams, Warren, Cummings, Harris, & Neves, 1993). In addition, freshwater mussels require native fish hosts for reproduction and dispersal, and so, freshwater mussel declines may also be tied to declining or changing fish communities. As declines continue, sensitive and reliable methods to assess distributions of freshwater mussel species are critical for focusing conservation efforts. Traditional mussel surveys are time-intensive and require specialized expertise to provide reliable results, which discourages their application at broad scales. Thus, alternate approaches providing more rapid assessments with equal or greater sensitivity would be a significant contribution to conservation of freshwater mussels.
The western pearlshell mussel (Margaritifera falcata) is a freshwater bivalve native to western North America from California to southern Alaska and east to the headwaters of the Missouri River in Montana, the Snake River in Wyoming, and the Great Basin in Utah and Nevada (Nedeau, Smith, Stone, & Jepsen, 2009). Western pearlshell mussels are benthic organisms that are patchily distributed in low-gradient habitats in clear, cold streams, with densities varying from locally abundant to very rare (Limm & Power, 2011;Stone, Barndt, & Gangloff, 2004). This species has been extirpated or is declining in many portions of its historical range (Nedeau et al., 2009), and intensive surveys to evaluate the distribution of this species have been recommended (Xerces Society, https://xerces.org/western-pearlshell/). Western pearlshell have been shown to emit significant quantities of organic matter likely to contain DNA in the form of feces (Limm & Power, 2011), and other possible sources of DNA may be released during molting, reproduction, mucus production, or decomposition posthumously (Deiner & Altermatt, 2014). However, their benthic habit, localized distribution, and low relative biomass could reduce detection probability with eDNA methods (Stoeckle et al., 2015).
Nonetheless, eDNA analysis has proven to be an invaluable tool for detecting a variety of taxa and has resulted in per site detection efficiencies exceeding traditional methods severalfold (Valentini et al., 2016;Wilcox et al., 2016). Reliable, whole-basin eDNA sampling designs have been developed for other taxa  resulting in an accumulation of archived eDNA. While these samples were collected targeting the DNA of fishes, they likely captured DNA shed by other organisms, including western pearlshell, and could provide all or a significant portion of a western pearlshell survey minimizing the need for additional field effort.
Our primary goal was to design and validate an eDNA assay specific to the western pearlshell. Our secondary goal was to determine whether archived eDNA samples collected for detection of species with very different life histories (in this case, fish) could be repurposed to confirm the presence of western pearlshell at locations of historical occurrence. Ultimately, we demonstrate the utility of maintaining a well-annotated archive of eDNA samples accessible from a central repository.

| METHODS
We designed an environmental DNA assay specific to western pearlshell in three phases: in silico, in vitro, and in vivo. First, we compiled sequences of the cytochrome oxidase subunit I (COI) mitochondrial gene of this species (Mock, Brim Box, Chong, Furnish, & Howard, 2013; Table 1) and 10 other mollusk species (Table 1). We screened the sequences in silico using the DECIPHER package (Wright et al., 2014) in R v. 3.2.3 (R Core Development Team (2015)) and obtained candidate primers unique to western pearlshell. We aligned the primers with sequence data in MEGA 6.0 (Tamura, Peterson, Peterson, Filipski, & Kumar, 2013) and adjusted primer lengths and position in Primer Express 3.0.1 (Life Technologies) to optimize annealing temperatures (Table 2). In addition, we compared the primers to additional western pearlshell sequence data (GenBank accessions AY579126-579128 and DQ272374-272383) and identified a T A B L E 1 Species, sample size (n), and GenBank accession number for DNA sequences used for in silico eDNA marker development. Also included is the minimum number of base pair differences between each component of the eDNA marker and the nontarget sequences  (Kwok, Chang, Sninsky, & Wang, 1994;Wilcox et al., 2015) at this position in the forward primer. The resulting primers amplify a 106-nucleotide fragment of the COI gene.
Within this fragment, we visually identified an area unique to western pearlshell and designed a FAM-labeled, minor-groove-binding, nonfluorescent quencher (MGB-NFQ) probe (Table 2) to maximize nucleotide differences with nontarget sequences. We assessed the annealing temperature of the probe in Primer Express 3.0.1 (Life Technologies; Table 2) and examined potential secondary structure formation of the primer-probe set using IDT OligoAnalyzer (https:// www.idtdna.com/calc/analyzer). To confirm the specificity of the western pearlshell assay in silico, we performed BLAST searches on each primer and the probe.
We also evaluated in silico the potential for cross-amplification of common fish associates of the western pearlshell to confirm that these species did not pose the risk of primer competition, potentially limiting the efficacy of the assay. Thus, we compared the primers to genetic sequence data of fish species suggested in Nedeau et al. To test the specificity of the assay in vitro, we performed qPCR analysis on DNA extracted from tissue of 23 western pearlshell specimens from 12 locations, as well as 23 nontarget mussel and fish species with which they co-occur (  . Concentrations resulting in the earliest C t value and a high endpoint fluorescence relative to the most concentrated level tested were selected for all subsequent analyses (Table 2).
Using these optimized primer concentrations, we then performed a standard curve analysis to examine the sensitivity of the assay.
To validate the western pearlshell assay in vivo, we compiled western pearlshell occurrence data from the Middle and North Forks of the John Day River in Oregon (Brim Box et al., 2003, 2006 and from 16 streams in Montana and one in eastern Idaho (historical surveys; Stagliano, 2010Stagliano, , 2015. Historical surveys were conducted using traditional techniques such as snorkeling, SCUBA, aquascopes, and direct observation in Oregon in 2003 (Brim Box et al., 2003), and Montana and Idaho between 2007 and 2014 (Stagliano, 2010(Stagliano, , 2015. We mapped these historical surveys onto our archive of eDNA sampling surveys to look for adjacency or overlap among survey types. We found eDNA surveys conducted in 2015 and 2016 targeting bull trout, smallmouth bass Micropterus dolomieu, and Arctic grayling Thymallus arcticus were near historical western pearlshell surveys (Table 4).
Where sites from both surveys were overlapping or adjacent, we se-  Table 4).
Overlap among survey types was also found in the West Fork Rock Creek, Montana (Figure 1). Here, extensive surveys based on traditional techniques for western pearlshell (2014) and on eDNA techniques for bull trout (2016) were independently conducted, and precise location data were available for both surveys (Stagliano, 2015;Young et al., 2017). In this area, we directly compared the results of these basin-level surveys.
All eDNA samples were collected following methods outlined in Carim, Dysthe, Young, McKelvey, and Schwartz (2016). Briefly, 5 l of water was pumped through a glass microfiber filter (pore size 1.5 μm) using a peristaltic pump, and the filter was stored in silica desiccant until extraction. DNA was extracted from each filter using the DNeasy T A B L E 4 Collection information for in vivo testing of the western pearlshell assay. All samples were collected during surveys for other taxa (see text). Expectation of western pearlshell presence was based on proximity to historical locations (Brim Box et al., 2003, 2006Stagliano, 2010Stagliano, , 2015. In the West Fork Rock Creek, all eDNA samples collected within the basin were analyzed for western pearlshell mussels to provide a formal comparison with basin-wide historical mussel surveys (Stagliano, 2010(Stagliano, , 2015; Figure 1). Site ID for the West Fork Rock Creek samples corresponds to sampling locations shown in Figure 1 Site ID Waterbody ( Stagliano, 2010Stagliano, , 2015 and bull trout eDNA surveys (Triangles; Young et al., 2017) were conducted. Black symbols represent sites where surveys failed to detect western pearlshell mussels, magenta disks represent positive results in the western pearlshell mussel surveys, and green triangles represent positive results for western pearlshell mussels obtained by repurposing the collected eDNA samples. Repurposed eDNA samples labeled in this figure are shown in Table 4 # Tissue and Blood Kit (Qiagen, Inc) following a modified protocol (Carim, Dysthe, et al., 2016). All eDNA was extracted in a room dedicated solely to this practice, and extracts were stored at −20°C until analyzed. Each sample was analyzed in triplicate 15μl reactions containing 7.5 μl Environmental Master Mix 2.0 (Life Technologies), optimized primer concentrations (Table 2), 250 nM probe, a TaqMan Exogenous Internal Positive Control (Life Technologies) including 1.5 μl of 10× IPC assay and 0.30 μl of 50× IPC DNA, and the remainder with deionized water. A no-template control in which distilled water was substituted for DNA template was included in each analysis. For all qPCR experiments, a reaction was considered positive if the amplification curve crossed the assigned threshold during the exponential phase.

| RESULTS
The results of the BLAST search indicated the potential for amplification of 10 nontarget mollusk species (Table 5). We aligned sequences of these species with the western pearlshell assay to examine the number of mismatches with each component of the assay. There were a minimum of eight mismatches, with at least one mismatch in each primer and four mismatches in the probe ( Table 5) (Table 4). The assay did not detect western pearlshell DNA in samples from nine streams in Montana where mussels were not previously observed in historical surveys and therefore not expected to occur (Table 4). In

| DISCUSSION
The assay we developed for the western pearlshell mussel is both efficient and specific, and effectively demonstrates the utility of repurposing eDNA sample collections. Environmental DNA samples collected for a previous independent eDNA survey from sites adjacent to traditionally identified populations produced consistent results (Table 4), even though none were specifically collected for western pearlshell.
The samples were collected for a highly mobile, midwater species (bull trout), which has a very different life history than the sessile mussels. for bull trout (Isaak, Young, Nagel, Horan, & Groce, 2015). Thus, lower elevation mussel beds identified by historical surveys were not sampled during the bull trout eDNA surveys (Figure 1). This result is likely to be common: Existing eDNA samples collected to detect one species will most often not completely replace the need to collect new data at qPCR analyses should take place in a room separate from extraction and PCR setup to avoid the risk of contaminating samples with PCR product. To minimize degradation of eDNA samples prior to processing, we recommend storing sample filters in silica desiccant in a cool, dark location immediately upon collection, and processing or placing them a freezer within 2 weeks . To minimize degradation after processing, one study in forensic science suggests that archived DNA extracts should be stored frozen in TE buffer at −80°C or below, or dried, amended with a trehalose additive, and stored at room temperature or −80°C (Smith & Morin, 2005). In addition, lo-bind or siliconized storage tubes should be used to minimize DNA binding to the tube walls, and repeated freeze-thaw cycles should be avoided.
While these are some general recommendations for long-term storage of DNA, more research is needed to determine the temporal stability of DNA from environmental samples stored at these conditions.
We envision that in addition to augmenting current surveys, eDNA samples will ultimately provide a snapshot of historical conditions for retrospective surveys. However, to fully take advantage of the multitude of eDNA samples available for assaying nontarget species (e.g., repurposing), this will require carefully archiving metadata in an online database and archiving the sample itself in a way to avoid degradation and contamination. These needs are likely best met via dedicated institutions with proper curation experience and facilities. While these institutions require investment, the potential value of the archived data in terms of information and cost efficacy is enormous.

DATA ACCESSIBILITY
All data are fully available without restriction. Genetic sequence data were obtained from GenBank, and accession numbers are listed in the manuscript and tables.