Environmental DNA (eDNA) assays for the detection of Ridgway’s rail (Rallus obsoletus) in the United States

We designed two novel environmental DNA (eDNA) assays for the detection of Ridgway’s Rail (Rallus obsoletus), and successfully validated each assay using eDNA samples collected from the species’ known distribution within the United States. These assays add to the suite of tools available for the monitoring of this rare and secretive marsh bird, and may help to further elucidate its movement patterns as well as identify important migration corridors. Observed sensitivity of the assays indicates exceptional performance, with limits of detection at ≤ 8 copies of the target eDNA fragment per reaction. Our publication adds to the growing body of literature supporting eDNA surveys as viable tools for bird monitoring endeavors.


Introduction
Within marsh birds, distinction among the clapper/king rail (Rallus crepitans and R. elegans) complex is infamously complicated, but recent phylogenetic analyses have resulted in the elevation of Ridgway's rail to full species status (Rallus obsoletus, formerly R. longirostris obsoletus; Maley and Brumfield 2013;Chesser et al. 2014).In the United States (US), this secretive marsh bird exhibits a limited distribution.It exists in vegetated and shallow saltwater, freshwater, and brackish systems (e.g., marshes, swamps, lagoons, estuaries) of Arizona (AZ), California (CA), and Nevada.Here, it is represented by three subspecies-California Ridgway's rail (R. o. obsoletus), Light-footed Ridgway's rail (R. o. levipes), and Yuma Ridgway's rail (R. o. yumanensis) (Eddleman and Conway 1994;Chesser et al. 2014)-all of which have experienced substantial population decline and Sheena M. Feist sheena.m.feist@usace.army.miltraditional methods, eDNA surveys could boost R. obsoletus detection success as well as increase confidence in determinations of presence/absence.Such developments will be useful for the conservation and management of R. obsoletus, and will aid monitoring efforts aimed at improving our understanding of R. obsoletus occupancy and ecology, including movement, migration, habitat use, and annual life cycles.

Assay design
We secured tissue samples (and one blood sample) from two of the three R. obsoletus subspecies found in the US (Yuma and Light-footed; Table 1).We also secured tissues from 7 non-target Rallidae species that share a distribution with R. obsoletus (Table 1), as well as two close congeners, R. crepitans (of which R. obsoletus was once a subspecies) and R. elegans.We used avian universal primers (Sorenson 2003) to sequence multiple mitochondrial genes (CoxI, ND1, ND2, ND3, ND5, ND6) for all 10 species.Resulting sequences were used for assay design.
To begin, we extracted DNA from tissues supplied by natural history museums (Table 1) using Qiagen DNeasy Blood and Tissue kits.Extractions then underwent PCR amplification for genes of interest using Invitrogen Platinum Taq and associated manufacturer-suggested reaction volumes (at a total volume of 50 μL, with 2 μL of DNA template) and thermal cycler programs (with an annealing temperature of 55 °C and 40 PCR cycles).Resulting PCR amplicons underwent either FastAP Thermosensitive Alkaline Phosphatase (Thermo Scientific) clean-up or EDTA/ Tris-EtOH precipitation before undergoing Sanger sequencing on an Applied Biosystems (ABI) 3500XL using ABI's BigDye Terminator v 3.1 sequencing chemistry.Resulting sequences were assembled and aligned in Geneious Prime v 2019.2.3, where 19 putative R. obsoletus-specific assays were designed with the assistance of the Primer3 (Rozen and Skaletsky 2000) plug-in.Following PCR checks for target-specificity, where numerous putative assays amplified non-target species, two assays were chosen for subsequent optimization and for use in US-based R. obsoletus eDNA surveys: RIRA-ND5-1 and RIRA-ND5-2.Relevant ND5 sequences used during assay development and optimization were submitted to Genbank (NCBI Accession # ON241255 -ON241264).

Assay validation
We next undertook in silico, in vitro, and in situ testing of designed assays (Goldberg et al. 2016).Both assays were first assessed in silico for specificity to R. obsoletus using Primer-BLAST (Ye et al. 2012), as provided online by the National Center for Biotechnology Information (NCBI).Primer-BLAST was executed with no taxonomic or genome locus exclusions, and with the nr database selected (date of query: April 2022).
In vitro qPCR tests for assay specificity were conducted against DNA samples from the species listed in Table 1.All qPCRs had total volumes of 20 μL containing 1X TaqMan Environmental Master Mix, 0.5 μM of each primer, 0.125 μM of the probe, and 1 ng of DNA template.The qPCR We next investigated the performance of our assays using 46 water samples (in situ testing) collected in October 2019 from 36 sampling points spread among four different locales (Suppl.File, Table S1).Locales included two national wildlife refuges (NWRs) with contemporary R. obsoletus satellite-tracking observations (Harrity et al. 2020): Imperial NWR, AZ (n = 21 of 46) and Sonny Bono Salton Sea NWR, CA (n = 4 of 46).Other locales were positioned within Department of Defense (DoD) lands.The first, Naval Weapons Station Seal Beach in Anaheim Bay, CA (NWSSB, n = 18 of 46), has contemporary survey records of R. obsoletus.Additionally, a small number of samples (n = 3 of 46) were taken from Vandenberg Space Force Base, CA (SFB; formerly Vandenberg Air Force Base), which is situated approximately 150 km north of the known northernmost extent of southern coastal populations of R. obsoletus and approximately 300 km south of the known southern extent of northern coastal populations of R. obsoletus.At the two NWRs, samples were taken from within or along the edges of thick stands of reeds or rushes, and guided by observed individual R. obsoletus site use (satellite-tracking) data.At NWSSB, samples were taken from near-shore waters along reaches expected to be used by R. obsoletus (R. Schallmann, personal communication).On Vandenberg SFB, samples were taken from points proximal to riparian, emergent vegetation (e.g., rushes).
When collecting water filtrates we followed protocols published by the US Forest Service's National Genomics Center for Wildlife and Fish Conservation (Carim et al. 2016).We used 0.2 μm cellulose nitrate (CN) filters and aimed to pump 5 L of water from suspected R. obsoletus habitat.However, water turbidity contributed to filter clogging, requiring up to three filters per sample.Resulting sample volumes ranged from 0.15 to 5 L, with a majority (14 of 23) representing ≥ 4.75 L (Suppl.File, Table S1).Filters were preserved in silica beads until DNA extraction.
Water grab samples were taken at a subset of sites, including some sites at which filtrate samples were also procured (Suppl.File, Table S1).Choice of sampling method was for experimental purposes, as well as site-specific ease of application/necessity (e.g., excessive filter clogging, pump battery failure).Grab samples were collected using up to five 50 mL screw-top polypropylene tubes, for total volumes of 200-250 mL (Suppl.File, Table S1).At NWSSB, filtering was extremely difficult due to rapid clogging and only grab samples were obtained at the majority of these sites.Overall, 50% of our samples were filtrate samples and 50% were grab samples (Suppl.File, Table S1).
At the completion of sample collection, we used distilled water to create two, on-site field negative controls, one for each sample type (filter and grab).All samples and controls were kept at 4 °C over the course of the sampling effort (4 days), then shipped overnight on ice to our laboratory facilities, after which they were stored at -20 °C until extraction.
DNA was extracted from samples and controls using a modified CTAB protocol (cetyltrimethyl ammonium bromide; Doyle and Doyle 1987), implementing an initial centrifugation step for all sample types (Guan et al. 2019).For those sampling points where more than one filter was employed (due to clogging), all associated filters were combined into a single CTAB extraction.Two extraction negative controls were included alongside sample CTAB extractions, one for each sample type (filter, grab).
Three qPCR replicates were run for each sample with each assay, following the thermal cycling parameters described previously.Each 384-well qPCR plate included all samples, the field negative controls, extraction negative controls, a qPCR negative control (i.e., no template control), and a template positive control (DNA extracted from the R. obsoletus blood sample in Table 1, used at 1 ng per reaction).A sample was considered positive for R. obsoletus eDNA if one or more of its replicate qPCRs resulted in fluorescence above the critical threshold of detection (C T ).The C T was automatically set by ViiA 7 software.Amplicons from a subset of positive qPCRs were sequenced to verify accuracy to the intended target.Following qPCR, amplicons were cleaned using EDTA/Tris-EtOH precipitation, and then sequenced in both directions using an ABI 3500XL genetic analyzer and ABI BigDye Terminator v 3.1 Cycle Sequencing Kit.

Assay sensitivity and efficiency
The limits of detection (LOD) and limits of quantification (LOQ) for validated assays were assessed using serial dilutions of synthetic DNA templates containing the DNA fragments of interest (gBLOCKs Gene Fragments; Integrated DNA Technologies).Methods used to determine assay-specific LOD and LOQ followed Klymus et al. (2020a, b) and, specifically, Guan et al. (2019).To detail, we used a dilution series representing 128, 64, 32, 16, 8, 4, 2, and 1 gBLOCK copies.LOD was based on a subset of dilutions (16, 8, 4, and 2 copies per reaction; which allowed us to reduce reagent waste) at 24 replicates each.Reactions followed the qPCR protocols used during assay validation but with 1 μl of gBLOCK (at the known concentration) introduced into the 20 μl total volume reaction.LOD was identified as the lowest gBLOCK dilution at which a 95% detection rate was observed (i.e., successful amplification in 23 of 24 replicates).LOQ followed the same qPCR protocols as LOD but elegans).Our in vitro tests demonstrated that both RIRA-ND5-1 and RIRA-ND5-2 will amplify eDNA from these non-target species.When DNA was introduced to the qPCR at 1 ng per reaction, amplifications observed in R. crepitans and R. elegans occurred at much later qPCR cycles than in R. obsoletus (i.e., C T = 23.86-30.54vs. C T = 16.51-18.39,respectively).Additionally, the distribution of R. obsoletus does not overlap with the distributions of these two species (Western US, Pacific Coast vs. Eastern US, Gulf and Atlantic Coasts, respectively).Thus, assay cross-activity should not impair the use of the ND5 assays in US-based R. obsoletus eDNA surveys.
During in situ validation of our assays, detections for R. obsoletus were achieved at 50% of the locations sampled, and in 30.43% of all samples (i.e., 14 of 46 samples; Table 3; Suppl.File, Table S1).Average C T values across positive eDNA detections were 41.22 for ND5-1 and 36.48 for ND5-2.When detection occurred, the average number of positive qPCR replicates (out of a possible three per sample) was two for both assays (Suppl.File, Table S1).Detections were greatest in filtrate samples and with RIRA-ND5-1 (Table 3; Suppl.File, Table S1).No amplification was observed in negative controls (field, extraction, qPCR).
The subset of positive eDNA detections (qPCR amplicons) that underwent Sanger sequencing (8 of 23, or 35%) were confirmed to be R. obsoletus, with all sequences matching their respective targeted ND5 region of interest.

Assay sensitivity and efficiency
Efficiency and sensitivity metrics for RIRA-ND5-1 and are RIRA-ND5-2 are provided in Table 4.

Discussion
The two R. obsoletus eDNA assays designed in this study exhibit high sensitivity to low concentrations of target DNA (LOD ≤ 8 copies per reaction), and have been successfully was based on 8 replicates each across the entire gBLOCK dilution series.LOQ was identified as the gBLOCK dilution at which the observed coefficient of variation (CV) in measured concentrations among qPCR replicates was ≤ 35%.Relative quantification used in CV calculations was based on a standard curve generated alongside LOD and LOQ using triplicate reactions of the following gBLOCK dilutions : 31,250, 6250, 1250, 250, 50, and 10 copies per reaction.Assay-specific efficiency, based on this standard curve, was automatically calculated by the Viia7 software.For each assay, all relevant metrics (LOD, LOQ, amplification efficiency) were produced and assessed using a single 384-well plate.

Assay validation
In silico tests of assay specificity against non-target species (Table 1) indicated little likelihood of primer/probe binding and, thus, minimal likelihood for "detection" of non-target species DNA.Yet, interspecific sequence similarity for ND5 is high between R. obsoletus and two closely-related species, Clapper rail and King rail (R. crepitans and R. Probes possess both an internal quencher (ZEN) and a 3' quencher (Iowa Black, 3IABkFQ) for increased specificity to our target.We note that these assays also amplify in two non-target species, R. crepitans and R. elegans (Clapper rail and King rail).However, the distribution of these non-target species does not overlap with R. obsoletus eDNA assay Primer ID Primer Sequence (5ʹ→3ʹ) hydrodynamics, microbial communities, inhibitory substances; Harrison et al. 2019), as well as our choices in sampling methods.Of the two methods used for eDNA sample collection, detection success appeared to be greatest within on-site filtrate samples.It is difficult to say whether this observation resulted from (1) the factors listed above, (2) differences in sampling volumes across the different sampling methods (i.e., 0.20-0.25 L with grab vs. > 2 L with on-site filtrates), and/or (3) true differences in R. obsoletus occupancy across locations (the largest number of grab samples were taken at NWSSB, which produced no detections regardless of sampling method, but contemporary survey records indicate R. obsoletus should be present).One factor that should be considered is the potential degradation of eDNA in grab samples which may occur between the time of collection and the time of subsequent processing (Curtis et al. 2021).We suspect a combination of factors played a role in our outcomes, and suggest that future efforts investigate and employ best practices for collection and handling of R. obsoletus eDNA samples.These best practices are likely to be situation-specific (i.e., dependent on not only feasibility, but also on R. obsoletus-specific seasonal behaviors and associated habitat/environmental conditions).The use of pilot studies to assist in the determination of best practices is a key recommendation for eDNA studies in general (Goldberg et al. 2016;Thalinger et al. 2021).
Even though we focused our eDNA sampling efforts in habitats supporting year-round R. obsoletus occupancy, our methods could be used to monitor R. obsoletus movements, including the seasonal and life-cycle-dependent use of different habitats, movement corridors, and stop over locations.validated for use in the US.When these assays were applied to field samples, detection success was greatest with the RIRA-ND5-1 assay.Many factors can lead to differential performance of assays in challenging templates like environmental samples (e.g., degradation/fragmentation, primer bias, template competition, and inhibition; Beng and Corlett 2020).As such, we recommend that future eDNA surveys utilize both ND5 assays to increase the probability of R. obsoletus detection.Evidence suggests that the use of multiple assays not only increases the chance of amplification success (Beng and Corlett 2020;Lance and Guan 2020), it also improves eDNA survey reliability (Sepulveda et al. 2020).Additionally, the use of exogenous internal positive controls (Exo-IPC) during qPCR could help determine if inhibitory substances influence eDNA detection rates across assays, sample types, and different environments/locations (Goldberg et al. 2016; but see Huggett et al. 2008;Lance and Guan 2020); additional optimization may be required to ensure Exo-IPC is introduced at an appropriate concentration so as to avoid competition with the eDNA assay (Klymus et al. 2020b).
We detected R. obsoletus eDNA at both of the NWRs sampled, but did not detect R. obsoletus at either of the DoD lands sampled.Our detection success at Imperial NWR supports the modeling efforts of Stevens and Conway (2019), which indicated high quality R. obsoletus habitat was expected to occur near this area.We note, however, the variability in R. obsoletus detection observed across different locations may be influenced by the number of samples collected at each site, site-specific environmental variables that impact eDNA persistence and detection rates (e.g., sunlight, Recent evidence indicates much is still to be learned about R. obsoletus in these respects (Zembal et al. 2013;Harrity and Conway 2020), and eDNA-derived occupancy data could fill some of these knowledge gaps, providing key information to enhance both the conservation and knowledge of R. obsoletus.For such eDNA applications to be informative, we note that the persistence of eDNA (and its movement, particularly in flowing systems) would be necessary factors to consider (Harrison et al. 2019).
As presented, our assays are currently validated for use in US-based R. obsoletus populations, with important caveats due to the use of limited numbers of R. obsoletus tissue/ blood samples and limited R. obsoletus genetic data available on NCBI.Assay utility could be increased via efforts to validate usefulness across greater geographic-and population-based extents (e.g., Mexico), including efforts to confirm assay performance (i.e., match) to any potential variation in intraspecific ND5 diversity within and across these extents.Similarly, we also suggest that the assays be checked for R. obsoletus specificity against co-distributed Central American/Mexican/Baja Californian non-target species, as our efforts largely focused on US-based R. obsoletus detection.With these additional validation steps and following site-specific optimization of eDNA sampling methods, our novel assays could provide further insights into R. obsoletus ecology and, thus, improve conservation actions for this elusive and rare species.

Table 1 Tissues from Rallus obsoletus (Ridgway's rail) and 9 non-target, potentially co-occurring species used in eDNA assay development, acquired from the Florida Museum of Natural History (FLMNH) and the San Diego Natural History Museum (SDNHM; in conjunction with the San Diego State University Museum of Biodiversity). Voucher origin represents the state from which the specimen was collected: California (CA), Florida (FL), and Louisiana (LA). Sample marked by * indicates material was blood rather than tissue, secured by the University of Idaho (or UI).
for 1 min (60 °C for RIRA-ND5-1).All reactions were run on an ABI ViiA 7 Real-Time PCR System.

Table 2
Two novel qPCR assays developed for use in eDNA surveys targeting Rallus obsoletus (Ridgway's rail, RIRA) in the United States.

Table 3
Forty-six Rallus obsoletus (Ridgway's rail) eDNA samples were collected from four locations in Arizona (AZ) and California (CA), using two sampling methods (on-site filtrates and water grab samples).Positive eDNA detections were observed in 14 samples.The number of detections varied across locations and sampling type, with the greatest detection success achieved in filtrate samples

Table 4
Efficiency and sensitivity for our novel Rallus obsoletus (Ridgway's rail, RIRA) eDNA assays, including the limit of detection (LOD) and limit of quantification (LOQ) for each, as determined using a synthetic DNA standard