Global efforts to conserve declining amphibian populations have necessitated the development of rapid, reliable, and targeted survey methods. Environmental DNA surveys offer alternative or complementary methods to traditional amphibian survey techniques. The California tiger salamander Ambystoma californiense (CTS) is endemic to California, where it breeds in vernal pools. In the past 25 y, CTS has faced a 21% loss of known occurrences, largely through habitat loss, and it is threatened by hybridization with an introduced congener. Protecting and managing remaining CTS populations rely on accurately monitoring changes in their spatial distribution. Current monitoring practices typically use dip-net surveys that are time-consuming and prone to false negative errors. To provide a new resource for monitoring and surveying larval CTS, we designed an assay and tested it on water samples collected from 29 vernal pools at two locations in California. We compared environmental DNA results to contemporaneous dip-net surveying results and found the assay agreed with positive dip-net results in 100% of cases. In several instances, we also detected the presence of CTS genetic material in the early spring before larvae hatched, potentially offering a new, earlier detection option for this imperiled species. This assay provides a valuable, noninvasive molecular tool for monitoring the spatial distribution of the CTS in vernal pools.
Amphibian declines in California's Great Central Valley mirror global trends (Fisher and Shaffer 1996), leading to local and regional efforts to protect and recover vulnerable species. The California tiger salamander Ambystoma californiense, (CTS) is endemic to California and faces ongoing threats throughout its range that have led to a 21% reduction in known CTS occurrences since 2002 (U.S. Fish and Wildlife Services [USFWS] 2017). There are three distinct population segments federally recognized by the USFWS: the Central Valley distinct population segment, which is listed as threatened under the US Endangered Species Act (ESA 1973, as amended), and the Sonoma County and Santa Barbara distinct population segments, which are listed as endangered under the same act (ESA 1973). In California, this species is listed as threatened throughout its range under the California Endangered Species Act (1973).
Several specific threats have spurred the protection and recovery of CTS, including habitat destruction and fragmentation, invasive predators, and hybridization with the introduced barred tiger salamander Ambystoma tigrinum mavortium (USFWS 2017). Losses of vernal pools that serve as breeding and spawning habitat from November to April threaten larval and breeding adult CTS; 80–90% of the vernal pool habitat has been lost since Spanish settlement (King 1998). These losses are largely due to urbanization, agricultural land conversion, and other anthropogenic factors. During the spawning season, CTS adults and larvae congregate in vernal pools and ponds and can be readily monitored by visual inspection or dip-netting. Outside of this season, CTS live underground up to 1.86 km from the breeding site (Searcy and Shaffer 2011), making population-level monitoring infeasible. Consequently, CTS monitoring typically occurs during the wet season by using dip-net monitoring or trapping.
Conservation and restoration of amphibian populations rely on ongoing monitoring to provide consistent, accurate data to facilitate status reviews and consequent management decisions. Such spatial distribution monitoring tracks the presence–absence of a species in its habitat and changes in its distribution over time (Greenberg et al. 2018). To date, dip-net surveys are the most common survey method for monitoring the aquatic stage of terrestrial salamanders, although they are not without limitations (Skelly and Richardson 2009). For example, dip-netting causes substantial disturbance to pools when surveyors enter the pool and move a large D-frame mesh and aluminum net through the water column. This process can disturb CTS habitat through destruction of substrate and can result in direct injury to CTS larvae if they become severely entangled in nets or surveyors crush them while walking through pools (Anderson and Davis 2013). The accuracy of dip-netting is also debatable. For example, Curtis and Patton (2010) modeled the detection rate of dip-net surveys on ambystomatid larvae in isolated ponds on the East Coast of the United States and found that detection rates varied across species and throughout the field season, but peaked at only 77%. Furthermore, surveyor movement among pools may increase the risk of spreading diseases such as ranavirus and chytridiomycosis between amphibian populations, the latter of which is implicated in the decline of more than 500 amphibian species (Greenberg and Palen 2019).
Environmental DNA (eDNA) monitoring is an emerging tool that can complement dip-net monitoring of amphibians and alleviate some of the associated concerns. Environmental DNA monitoring is a survey method that processes environmental samples (here, water) for genetic traces of a target species by using specially designed quantitative polymerase chain reaction (qPCR) assays. Worldwide, eDNA has been successfully used to track and monitor salamander species including the great crested newt Triturus cristatus in the UK (Biggs et al. 2015), the endangered hellbender salamander Cryptobranchus alleganiensis in Pennsylvania (Pitt et al. 2017), and the Idaho giant salamander Dicamptodon aterrimus (Pilliod et al. 2014), among others (Vörös et al. 2017; Goldberg et al. 2018; Preißler et al. 2019). The method may be particularly appropriate for CTS monitoring in part because dip-net surveys for CTS must be carried out by trained surveyors carrying federal permits. Collecting water from vernal pools for eDNA sampling requires no permits and minimal on-site training, can frequently be carried out without humans entering pools, and can provide highly accurate results. Here, we developed and field tested a qPCR-based eDNA assay for the detection of CTS from water samples and examined its potential as a survey method to monitor CTS presence–absence in vernal pools in California's Central Valley.
Assay development and optimization
To develop an eDNA assay for monitoring larval CTS, we first obtained representative DNA sequences from CTS and other Ambystomatidae covering the entire mitochondrial genome from GenBank (Grant et al. 2016; Table S1, Supplemental Material) and aligned them using the software program MEGA7 (Kumar et al. 2016). To design the qPCR assay, we identified candidate assays on the Cytochrome Oxidase I mitochondrial gene with Primer3Plus (Untergasser et al. 2007) and PrimerQuest (IDT, Coralville, IA). We then developed several candidate assays with a forward primer, a reverse primer, and a fluorophore-labeled DNA probe. Next, we checked each set of assays against our aligned sequences for the presence of species-specific, single-nucleotide polymorphisms. We selected the candidate assay that had the highest number of interspecific single-nucleotide polymorphisms without compromising optimal reaction kinetics (Table 1). To validate and optimize assays, we ran the assay using tissue-derived California tiger salamander DNA taken from adult salamanders in the Santa Rosa Plains in Sonoma County. The resulting optimized reaction recipe and thermocycling protocol for each sample was 1× TaqMan Environmental DNA Mastermix (Thermo Fisher Scientific, Waltham, MA), 0.9 μM each primer, 0.15 μM probe, 1× bovine serum albumin (Life Sciences), and 6 μL of eDNA template in a reaction volume of 20 μL, with the following thermocycling conditions: an initial denaturing step at 95°C for 10 min, followed by 45 cycles of 95°C for 10 s, and then 56°C for 1 min. To estimate assay specificity, we performed in silico PCR by using ecoPCR (Ficetola et al. 2010) with approximately 70,000 available mitochondrial sequences from all ambystomatid species in GenBank, including the invasive A. t. mavortium. There are no natively occurring Ambystomatidae whose ranges are thought to overlap with CTS (USFWS 2017).
Measurement of limit of detection and limit of quantitation
The limit of detection (LOD) is a parameter used to evaluate the sensitivity of qPCR assays; the LOD is a measure of the lowest concentration of analyte (in this case, target genetic material) detectable in a qPCR assay and is distinguishable from the concentration plateau (Hunter et al. 2016). We used the calculation of the LOD and the limit of quantitation (LOQ) described by the U.S. Geological Survey's Ohio Water Microbiology Laboratory (Francy et al. 2017) based on work by Armbruster and Pry (2008). To determine the LOD and LOQ of the assay, we used synthetic double-stranded DNA fragments (gBlock Gene Fragments; IDT) matching the target amplicon. gBlock gene fragments allow for precise measurement of the number of input DNA copies in each reaction. We produced a standard curve ranging from 900 copies per reaction to 0.6 copies per reaction with the gBlock gene fragments. We replicated each concentration of gBlock standards eight times, and the standard curve included eight no-template controls.
For the purpose of this study, we define “eDNA sample” as a volume of water collected from a vernal pool, passed through a filter, and processed for DNA extraction. A “sampling event” is the process of collecting three replicate water samples, collecting a negative field control sample, and performing a dip-net survey of a single pool at a single visit. We collected our eDNA samples during three wet seasons (January–March in 2016, 2017, and 2018) at two vernal pool complexes regularly monitored for CTS: The Jepson Prairie Preserve in Solano County and the Dutchman Creek Conservation Bank in Merced County (Figure 1).
We sampled pools that ranged in size from 3 m2 to nearly 6.8 km2. We collected replicate eDNA samples in each pool by submerging a sterile 1-L Nalgene container into the pool by hand with a single-use nitrile glove. Between uses, we submerged Nalgene containers in 10% bleach for 30 min, rinsed them in deionized water, and placed them under a UV hood for 15 min. For pools larger than approximately 500 m2, the collector wore sterile, single-use boot covers and waded a short distance into the pool to collect one or more of the replicate samples. For pools larger than approximately 10,000 m2, we sampled a single transect of approximately 10,000 m2. In total, we sampled 29 vernal pools between one and six times each for 51 total sampling events. To evaluate the rate of false positive detections and ensure the assay did not amplify California tiger salamander when not present in a pool, 16 of our sampling events were from pools with no current or historical presence of CTS (Table S2, Supplemental Material). Because CTS has been carefully tracked and monitored for multiple years at these sites, testing historically CTS–negative pools is a reliable way to monitor for false positives. To ensure that any positive amplifications we found were not the result of contamination, we included sterilized water as negative controls alongside each sample.
Filtration occurred concurrently with sample processing, whether in the field or in the laboratory. To filter our eDNA samples, we filtered water samples through a 47-mm-diameter filter (glass fiber, 0.45 or 1.2 μm (Whatman, Maidstone, UK); cellulose nitrate, 0.45 μm (MilliporeSigma, Burlington, MA)) by using a peristaltic pump (Geotech Environmental Equipment, Sacramento, CA) attached to the vacuum flask with silicon tubing. We filtered water until 500 mL passed through the filter or the filter clogged. We filtered all three replicates sequentially. Immediately after field sample filtration, we filtered a 500-mL negative control by using sterilized Nanopure water (Thermo Fisher Scientific, Waltham, MA). Between samples, we replaced all reusable filtration materials (e.g., tubing, filter manifolds) with clean units. We stored contaminated gear in sealed zip-top bags until sterilization in the laboratory with 30 min in 20% bleach, a triple rinse with deionized water, and UV sterilization in a UV hood or crosslinker for 15 min.
We conducted dip-net surveys immediately after eDNA sampling following USFWS survey guidelines (USFWS 2015). We recorded the presence–absence of any CTS larvae in dip nets. The dip-net surveys did not record the presence of CTS eggs in the pools. When pools were more than 10,000 m2, we sampled a transect identically to that used for eDNA sample collection.
Sample processing optimization
To developing the most efficacious field protocol, we refined several steps across the 3 y of sampling (for details, see Figure S1, Supplemental Material). We made all changes incrementally to limit how sampling protocols may confound results. One change included varying filter materials. We used glass fiber filters in 2016 and 2018 and cellulose nitrate filters in 2017. We also varied the how the filters were stored. In 2016, we stored filters dry in silica gel, whereas in 2017 and 2018 we stored filters in the proprietary Qiagen reagent Buffer ATL at room temperature for up to 5 d before DNA extraction. Both dry and buffer storage for filters are effective methods for storing eDNA filters at room temperature (Renshaw et al. 2015; Spens et al. 2017; Majaneva et al. 2018), and cellulose nitrate and glass fiber are both proven filter materials for eDNA (Goldberg et al. 2016).
Additional refinements of our field protocol included varying filtration to test the efficacy of laboratory filtration methods. To do this, we varied the filtering protocols between field and laboratory filtration during repeat visits to the same pools. When samples were filtered in the laboratory, unfiltered water samples were transported in a cooler on ice from the field to the laboratory and stored at 4°C until filtration within 6 h. The costs and benefits of filtering in the laboratory are still being studied, and the ideal method may be ecosystem and assay specific, but the temporary storage of unfiltered water samples up to 24 h is common practice (Handley et al. 2016; Bastos Gomes et al. 2017; Gingera et al. 2017; Hinlo et al. 2017; Fernández et al. 2018; Takahara et al. 2019).
DNA purification and analysis
After filtration, we extracted DNA from filters in a clean laboratory to minimize the potential for contamination. The clean laboratory contained no tissue or high-concentration (tissue-derived or PCR-amplified) DNA from any amphibian or vernal pool species, following the recommendations of Goldberg et al. (2016). We extracted filter-bound DNA with the DNeasy Blood and Tissue kit (Qiagen, Hilden, Germany), with the following modifications: 20 μL of Proteinase K was added to the microcentrifuge tubes containing filters stored in Qiagen reagent Buffer ATL. We cut dry filters into quarters, placed them in separate microcentrifuge tubes, and inundated them with 20 μL of Proteinase K and 180 μL of Qiagen reagent Buffer ATL. We then incubated samples on a rotary incubator (Corning Inc., Corning, NY) overnight (or for at least 12 h). Any filter material remaining after incubation was discarded and not transferred to the spin column. Finally, we used two final elutions of 60 μL of Nanopure water with a 15-min incubation at 56°C. After extraction, all samples were treated proactively for inhibition using the OneStep PCR Inhibitor Removal Kit (Zymo Research, Tustin, CA).
We tested extracted DNA for the presence of our target species by using the optimized protocols on a qPCR machine (CFX; Bio-Rad Laboratories, Hercules, CA). We initially tested seven sampling events, with each of the three field replicates run separately. Of these seven sets of three, all field replicate results from a sample were either positive or negative, indicating perfect agreement between field replicates. As a result, to preserve samples and reduce reagent cost we tested pooling the DNA extracts from the three field replicates from each sampling event. For this test, we ran and accessed four PCR replicates per pooled sample with the following criteria: if one PCR replicate out of four amplified, we re-ran the sample; if two or more replicates amplified, we considered it a positive detection; and if zero of four amplified, we considered it a nondetection. Each plate also included four no-template controls (water blanks) and eight standardized positive gBlock controls.
Limits of detection and quantitation
We used this equation to determine LOD and LOQ. Following the U.S. Geological Survey definition, we calculated the LOD for this assay to be 23 copies per reaction and the LOQ for this assay to be 75 copies per reaction (Table 2).
Field sampling results
Our assay and the dip-net surveys both detected CTS larvae in the same 14 of 51 sampling events, for a 100% agreement rate between positive dip-net results and positive eDNA surveys. In an additional 14 sampling events, our eDNA assay detected CTS when the dip-net surveys did not, suggesting recent presence. In 23 sampling events, neither the dip-net survey nor the eDNA assay detected CTS, including the 16 sampling events from sites where CTS was historically absent. There were no instances where the dip-nets detected CTS but the eDNA assay did not (Table 3).
Protocol optimization results
We compared our detection results between filtration location protocols and filter materials. In this study, the results suggest our assay was robust to sampling methodology, with no evidence that variation in protocol methods affected detection rates (Table S3, Supplemental Material). Because we had no instances where dip-net surveys detected CTS but our eDNA assay did not, any analysis of the impact of protocol variation on the assay must necessarily be speculative, and we cannot conclude that any combination of filter or filtration location outperformed any other.
We developed a new qPCR-based eDNA assay that offers improved detection of the presence of CTS larvae in vernal pools compared with dip-net survey methods. This assay gives highly concordant results with traditional dip-net surveys, with all positive dip-net detections also detected by our eDNA assay. The assay was found to perform well even with some variations in field sampling protocols. Our early-season positive results point to further uses for our assay, including detection of reproductive material (gametes) left behind by adults. The CTS eDNA assay presented here offers an effective new tool to monitor Central Valley CTS larvae that could, in the future, be expanded and used in other geographic areas with additional development.
One advantage of eDNA monitoring for CTS is that it minimizes human disturbance of vernal pools while determining presence–absence. For vernal pool sampling, eDNA sample collections occur without entering pools by using commercially available long-pole samplers or sampling backpacks (Thomas et al. 2018). Environmental DNA sampling can reduce habitat disturbance and may limit disease transmission between sites via human contact with water during dip-net sampling. Furthermore, eDNA sample collection can be performed without the need for federal permits or species-specific training.
We found early-season eDNA detection in samples collected up to a month before dip-net surveys detected the presence of larval CTS. It is possible that the DNA detected in these samples sourced from gametes or embryonic CTS before larval hatch, or from reproductive material or other DNA shed from breeding adults. Larval hatch occurs 10–28 d after breeding (USFWS 2017), after adult salamanders have left the breeding pool. We had no positive detections for CTS in pools considered historically negative for CTS. Therefore, we have no reason to believe that our early-season detections are false positives. In addition, the 14 sampling events that were positive with our eDNA assay, but negative in dip-net surveys, were all in pools that supported populations of CTS larvae later the same year. Although reproductive material is abundant in the pools during breeding season, it would likely degrade rapidly in warm weather; elevated temperatures change several biotic and abiotic factors known to influence the decay of genetic material (e.g., water pH, microbial community abundance; Eichmiller et al. 2016). Elevated early-season temperatures may reduce the abundance (and thus detection) of gametes and produce eDNA false negatives between egg laying and larval emergence, although during this period dip nets would also detect no CTS larvae. Managers will have to decide when to use eDNA surveys based on their needs and questions.
Our assay is designed to detect a locus on the maternally inherited mitochondrial DNA of a CTS salamander. We used mitochondrial DNA because of its high copy number relative to nuclear DNA, which increases the likelihood of a detection in an eDNA sample. However, due to its maternal inheritance, mitochondrial DNA cannot distinguish between a pure CTS and a hybrid CTS × A. t. mavortium that is maternally CTS. Our assay will not detect hybrids that are paternally CTS.
For managers wanting an efficient way to monitor the presence–absence of CTS without dip-netting, we recommend use of this assay with eDNA sampling in spring after larval emergence. However, it is important to remember that our assay cannot provide data about CTS abundance or health, so dip-netting is still necessary when more than presence–absence data is required, such as count data or larval maturity information. Additional field testing of our assay could introduce potential other uses for this survey methodology, such as its use as an early-season predictive tool for pools that are expected to support CTS larvae later in the year. This could allow managers to take early action to protect these pools. As global amphibian decline has intensified the monitoring and management of salamander populations, eDNA has proven to be a successful method for monitoring freshwater amphibians in general and salamanders in particular (Pilliod et al. 2014; Spear et al. 2015; Katano et al. 2017; Preißler et al. 2019). Our assay expands the utility of eDNA to monitoring the threatened CTS, providing managers with an additional highly accurate method of tracking the spatial distribution of larval CTS in the Central Valley.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Table S1. Record of the mitochondrial genetic sequences used to develop and validate our Ambystoma californiense quantitative PCR assay in silico by using MEGA7 (Kumar et al. 2016) and ecoPCR (Ficetola et al. 2010). The scientific name and common name for each species is given, along with the National Center for Biotechnology Information accession number that can be used to reference the original sequence at https://www.ncbi.nlm.nih.gov. When the original sequence is published, the citation is included in the table. These accession numbers represent whole mitochondrial genome sequences for our target salamander species and a variety of other Ambystomatidae. Not all ambystomatid species were available for comparison. No other ambystomatid species is known to natively co-occur with A. californiense (USFWS 2017).
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S1 (35 KB XLS).
Table S2. Record of California tiger salamander Ambystoma californiense environmental DNA and dip-net sampling events undertaken January–March in 2016, 2017, and 2018 in California. This record includes a unique identifier for each pool (pools that begin with D are located at the Dutchman Creek Conservation Bank in Merced County; pools that begin with J are located at the Jepson Prairie Preserve in Solano County); the latitude and longitude of the vernal pool (decimal degrees) and the sampling date; the dip-net sampling results (0 for a nondetection and 1 for a detection); the eDNA assay results (0 for a nondetection and 1 for a detection); the filter type (GF for glass fiber or CN for cellulose nitrate); the filtration protocol (in the field or in the laboratory); the historical status of the California tiger salamander in that pool; the water volume filtered (per biological replicate and averaged across replicates, in milliliters); and pool area (in square meters).
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S2 (45 KB XLS).
Table S3. Results of California tiger salamander (CTS) Ambystoma californiense environmental DNA and dip-net surveys broken down by (a) filter type and (b) filtration protocol. Because our assay has perfect agreement with positive dip-net survey results, all considerations about the effect of various protocols on environmental DNA assay detection rate must be speculative. There are seven cases where there is historical or known presence of the CTS in the pool, but neither the dip-net nor the environmental DNA assay detected the CTS. These cases may speculatively represent failures of our assay to detect CTS. Of these, three assays used a cellulose nitrate filter and four assays used glass fiber. Three were filtered in the field and four in the laboratory. We conclude that there is no evidence that protocol variation impacted the detection rate of our assay.
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S3 (34 KB XLS).
Figure S1. Flowchart of protocol variations for our 51 sampling events (“samples”), each comprising three replicate filters and a negative control filter. In 2016, we used a 0.45-μm glass fiber filter with a reusable 47-mm filter manifold (Advantec, Durham, NC), filtered immediately in the field, and stored dry in silica gel. In 2017, we used a 0.45-μm cellulose nitrate filter (Sterlitech) housed in a single-use filter manifold. We varied our filtration between the field and the laboratory and stored filters in Qiagen Buffer ATL. In 2018, we used a 1.2-μm glass fiber filter (Whatman, Maidstone, UK), did all filtration in the laboratory, and stored filters in Qiagen Buffer ATL. We bleach sterilized all reusable equipment (tubing, filter manifolds, flasks) in 20% bleach for 30 min, triple rinsed the equipment, and then UV sterilized it for 15 min between uses.
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S4 (173 KB JPG).
Reference S5.Grant EHC, Muths E, Katz RA, Canessa, S, Adams MJ, Ballard JR, Berger L, Briggs CJ, Coleman J, Gray MJ, Harris MC, Harris RN, Hossack B, Huyvaert KP, Kolby JE, Lips KR, Lovich RE, McCallum HI, Mendelson JRI, Nanjappa P, Olson DH, Powers JG, Richgels, KLD, Russell RE, Schmidt BR, der Sluijs AS, Watry MK, Woodhams DC, White CL. 2016. Salamander chytrid fungus (Batrachochytrium salamandrivorans) in the United States — developing research, monitoring, and management strategies. Open-File Report 2015-1233. Reston, Virginia: U.S. Geological Survey.
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S5 (821 KB PDF); also available at https://pubs.usgs.gov/of/2015/1233/ofr20151233.pdf.
Reference S2.[USFWS] U.S. Fish and Wildlife Service. 2015. Survey guidelines for listed large branchiopods. U.S. Fish and Wildlife Service, Pacific Southwest Region, Sacramento, California.
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S6 (12.57 MB PDF); also available at https://www.fws.gov/ventura/docs/species/protocols/vpshrimp/shrimp.pdf.
Reference S3.[USFWS] U.S. Fish and Wildlife Service. 2017. Recovery plan for the central California distinct population segment of the California tiger salamander (Ambystoma californiense). U.S. Fish and Wildlife Service, Pacific Southwest Region, Sacramento, California.
Found at DOI: https://doi.org/10.3996/052019-JFWM-041.S7 (750 KB PDF); also available at https://www.fws.gov/sacramento/outreach/2017/06-14/docs/Signed_Central_CTS_Recovery_Plan.pdf.
This project was funded through cooperative agreement R15AC00040 between University of California, Davis, and the USFWS, with additional funding from the University of California Natural Reserve System. We thank the land managers who provided access to study sites: the University of California Natural Reserve System and V. Boucher at Jepson Prairie Preserve and M. Gausse and T. Collins with Westervelt Ecological Services at Dutchman Creek Conservation Bank. We thank and acknowledge the two anonymous reviewers and Associate Editor for thoughtful commentary that greatly improved an earlier version of this survey. We also thank A. Goodbla, A. Benjamin, A. Coen, K. McGee, H. Hwang, P. Tran, J. Bachiero, and D. Prince for help collecting samples in the field. We acknowledge the help of R. Nagarajan, K. Deiner, A. Schreier, A. Holmes, and D. Gille, who provided feedback and guidance.
Any use of trade, product, website, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Citation: Kieran SR, Hull JM, Finger AJ. 2020. Using environmental DNA to monitor the spatial distribution of the California tiger salamander. Journal of Fish and Wildlife Management 11(2):609–617; e1944-687X. https://doi.org/10.3996/052019-JFWM-041
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.