1.2. Reptile eDNA
Despite breakthroughs in assessing density in fish and amphibian species, there remains a dearth of studies quantifying aquatic non-avian reptile populations with eDNA under field conditions [
66]. This gap in the literature is notable because turtles are among the most at-risk vertebrates, with over 60% of modern species listed as threatened, endangered, or extinct [
67,
68]. To our knowledge, most eDNA studies on non-avian reptiles that heavily use aquatic habitats focus on detecting the presence of snakes (five studies) and turtles (nine studies) (
Figure 1). An attempt was also made to find West African crocodile (
Crocodylus suchus) and Nile monitor (
Varanus niloticus) with eDNA metabarcoding methods, but presence has not yet been detected successfully [
69]. For this study, 500 mL of water was filtered through a 0.45 µm nitrocellulose Whatman filter [
69]. DNA was extracted with an ethanol precipitation and EZNA Tissue DNA kit and then amplified with the 12SV5.1 universal primer to target all vertebrates [
69]. While human, dog, and teleost DNA were detected, reptile and amphibian eDNA were not [
69].
The first notable aquatic reptile eDNA study was on Burmese python (
Python bivittatus) in South Florida [
23]. These snakes inhabit semi-aquatic slough, coastal habitats, and marsh prairie [
71]. This study extracted 15 mL water volumes from each site with a sodium acetate precipitation protocol and extracted eDNA with Qiagen’s QIAamp DNA Micro Kit [
23]. After successfully detecting python presence from aquatic eDNA using penned snakes, field sites with previously sighted pythons were tested [
23]. Field sites yielded positive eDNA detection where
P. bivittatus had been detected previously, and no eDNA was detected at one site where a python had not been detected previously [
23]. Further research detected eDNA in terrestrial samples under field conditions in sites monitored via radio telemetry [
66]. In this study, 950 mL of water was filtered using 0.45 µm cellulose nitrate filters near radio-tracked snakes [
66]. This yielded a 58% detection rate using a quantitative PCR (qPCR) Taqman assay [
66]. Another study confirmed these results by sampling sites with radio telemetry and burrow camera tracking [
72]. However, unlike previous studies, soil samples were taken instead of water samples. Soil samples (~0.5 g) were extracted with a modified phenol-chloroform protocol and amplified with conventional PCR for
P. bivittatus using a
CytB primer [
72]. Importantly, the study found eDNA degraded over time in soil, becoming undetectable four to seven days after snake presence [
72]. This supports the use of soil eDNA tools for a temporally sensitive presence detection measure. Overall, eDNA can detect
P. bivittatus in field settings, both for aquatic and terrestrial samples.
Additional aquatic snake studies have focused on the threatened eastern massasauga rattlesnake (
Sistrurus catenatus) [
73]. Fifty mL of water was taken from crayfish burrows, typical
S. catenatus overwintering refugia, in occupied field sites [
73]. Samples were centrifuged and filtered with a 1.4 µM cellulose acetate filter, extracted with a Qiagen Qiashredder and Qiagen DNeasy Blood and Tissue kit, and run through the Zymo PCR Inhibitor Removal Columns [
73]. A species-specific Taqman assay targeted the cytochrome oxidase I (
COI) region of
S. catenatus mitochondrial DNA (mtDNA) [
73]. Despite known local abundance and collecting water within a meter of a snake, only two of 100 environmental samples amplified positively with eDNA, compared to 12 positive snake detections with traditional methods within a 2 m radius [
73]. This assay was not effective in determining
S. catenatus presence compared to visual confirmation.
Similarly, giant garter snake (
Thamnophis gigas) eDNA assays were created for presence detection [
74]. These snakes inhabit canals and marsh-like habitats [
75]. In this study, laboratory experiments detected
T. gigas eDNA from tanks with snake skin and snake feces in water, but curiously eDNA was not detected in tanks with live snakes in the water [
74]. This qPCR Taqman assay used 1 L water samples which were filtered with a 0.45 µm nitrocellulose filter, although precipitation with sodium acetate methods were also tested [
74]. The Taqman assay amplified the
ND4 region of mitochondrial DNA (mtDNA), although cytochrome B (
CytB) and NADH dehydrogenase 2 (
ND2) regions were also considered [
74].
T. gigas eDNA was not detected in water at field locations, despite capture with traps at the same sites [
74]. Overall, this assay was not effective at detecting
T. gigas eDNA in field settings compared to traditional methods.
With metabarcoding primers, redbelly snake (
Storeria occipitomaculata), northern watersnake (
Nerodia sipedon), and milksnake (
Lampropeltis triangulum) eDNA presence was detected in Canadian lakes and rivers [
70]. For this assay, 1 L water samples were filtered through 1.2 µm glass microfiber filters (Whatman GF/C), and DNA was extracted with the Qiagen QIAshredder and DNeasy Blood and Tissue kit [
70]. Metabarcoding assays targeted the
CytB and
COI regions of mtDNA [
70]. These results suggest that metabarcoding can effectively detect snake eDNA, although more studies are needed to confirm this finding. Overall, results have been mixed for detecting the presence of snakes with eDNA (
Table 1) and, to our knowledge, no studies have yet attempted to quantify snake eDNA. It is possible that the more time snakes spend in water, the more likely aquatic eDNA will be able to detect snake presence; however, more research is needed to support this relationship.
Previous work has assessed the ability of eDNA to detect the presence of aquatic turtle species in a variety of habitats. In a marine aquarium, a green sea turtle (
Chelonia mydas) was present but not detected when using eDNA metabarcoding methods [
79]. Water samples were split into 1 L subsamples and filtered through 0.22 µm durapore membrane filters [
79]. The filters were divided into groups, and DNA was extracted using either the DNeasy Blood and Tissue Kit or PowerSoil DNA Isolation Kit [
79]. The vertebrate-specific
12S rRNA gene was targeted for a metabarcoding assay, and the mtDNA
CR of
C. mydas was specifically targeted with conventional PCR [
79]. While no turtle eDNA was detected with metabarcoding methods, the species-specific PCR assay detected the presence of
C. mydas. This suggests species-specific assays may be more sensitive to detection of a particular species as compared to metabarcoding methods.
Similarly, eDNA assays were developed for multiple captive native Canadian turtles, and eDNA from red-eared slider turtle (
Trachemys scripta) was successfully detected in a single artificial pond (
Table 1) [
24]. One-liter water samples were filtered with 1.2 µm Whatman GF/C glass microfiber filters and DNA was extracted out with the Qiagen DNeasy Blood and Tissue kit [
24]. The
COI region was targeted with a PowerSYBR qPCR assay [
24]. All PCR replicates amplified, showing the ability to detect
T. scripta presence in the field [
24]. The increased testing of this assay in other bodies of water may help determine if this assay is more broadly applicable to other bodies of water. Importantly, this assay may help to detect where
T. scripta has become an invasive species [
83].
Additionally, an eDNA assay was developed to detect alligator snapping turtle (
Macrochelys temminckii) presence in both lentic and lotic environments in the southeastern USA [
78]. Filtered water volumes varied according to water sample clogging rates, so two 1.5 µm microfiber glass filters were used to collect between 850 and 1500 mL of field water in total [
78]. The Qiagen QIAshredder and DNeasy Blood and Tissue kits were used to extract out DNA which was amplified by a Taqman probe targeting the
CR of mtDNA [
78]. Technical replicates amplified 1/6 to 2/3 replicates, but technical replicates never amplified in all triplicate or sextuplicate reactions [
78]. Furthermore, the replicates that did amplify had a high Cq value (>38), indicating a low eDNA concentration [
78]. It is possible that turtle eDNA concentrations may approach the limit of detection in some cases.
In India, several imperiled turtle species (
Chitra indica,
Nilssonia gangetica, and
N. nigricans) were detected in a temple pond using eDNA methodology [
80]. Small water samples of 15–20 mL were collected and preserved with sodium acetate [
80]. The DNA was extracted with the QIAamp Tissue Extraction Kit, and DNA was amplified with PCR targeting the
COI region of turtle mtDNA (~650 bp), not specific to species [
80]. Amplicons were sequenced to and confirmed to detect these three turtle species with BLAST [
80]. This provides support that turtle species can be detected with eDNA, even with large amplicon sizes.
In Southeast Asia, the southern river terrapin (
Batagur affinis) was detected in river samples in Malaysia [
82]. Water samples (5 L) were collected close to a turtle nesting beach and 250 mL were filtered through a 0.45 µm cellulose nitrate filter [
82]. The filters were kept in T1 lysis buffer and then extracted with the NucleoSpin Tissue Kit. Species-specific primers for
B. affinis amplified the
CytB region of the mtDNA genome with conventional PCR which yielded DNA [
82]. This eDNA detection corresponded to the presence of at least one radio-tracked individual within 1 km (
Table 1). This indicates that eDNA could be effective in detecting critically endangered turtles in their natural habitat. Potentially, this could aid in conservation efforts in Malaysia and other similar habitats.
Beyond presence detection, site-occupancy models in slow-flowing streams in the southeastern USA quantified the minimum number of eDNA samples needed to determine presence of the endangered flattened musk turtle (
Sternotherus depressus) [
77]. This study filtered 1 L of water through 0.80 µm cellulose nitrate filters [
77]. Species-specific primers amplified part of the
16S rDNA using an EvaGreen qPCR assay [
77]. This study found the warm season (May–September) yielded higher eDNA detection rates for
S. depressus, which likely corresponds to turtle activity [
77]. Four replicate samples were needed in the warm season for a 95% detection probability versus 14 during the cool season. This supports the idea that sampling should be targeted to organism biology and active periods. Additionally, it shows how many replicates may be needed to efficiently detect
S. depressus under field conditions.
The density dependence of threatened European pond turtles (
Emys orbicularis) in natural ponds was also investigated using eDNA in Switzerland [
81]. To collect eDNA from ponds, 90 mL of water was filtered through a 0.22 µm filter, and DNA was extracted with the QIAamp Tissue Extraction Kit or sodium acetate precipitation methods [
81]. A SYBR Green assay was used to amplify part of the
E. orbicularis CytB gene. No correlation was found between turtle density, number, or biomass and eDNA abundance, although sites with vegetation and shallow waters yielded more turtle eDNA [
81]. Detection probabilities ranged from 25–100%, and precipitation yielded increased detection (7 ponds versus 5 ponds) [
81]. While unable to relate biomass to eDNA abundance, this study does support eDNA as a way to detect this threatened turtle under field conditions.
In Canadian riverine environments, the sensitivity of eDNA detection of at-risk wood turtles (
Glyptemys insculpta) was tested [
70]. As previously mentioned, 2 L water samples were filtered with 1.2 µm glass microfiber filters. However, qPCR Taqman assay targeted the
COI region for
G. insculpta abundance. The presence of
G. insculpta was detected and correlated with turtle abundance from visual surveys. Furthermore, when using eDNA-metabarcoding methodology and universal primers, both
G. insculpta and common snapping turtles (
Chelydra serpentina) were detected. However, these metabarcoding methods did not detect
G. insculpta eDNA in all rivers where qPCR eDNA methods detected this species [
70]. This supports a trade-off between single-species and metabarcoding methods; Taqman qPCR assays may be specific and sensitive but limited to one species, while metabarcoding is less specific but can detect multiple species within a sample. Indeed, this is also seen in the aforementioned aquarium mesocosm study [
79].
Finally, the eastern box turtle (
Terrapene carolina) presence was detected using metabarcoding methods on an Illinois river, though turtle presence was not confirmed with an actual specimen [
76]. Surface water (~50 mL) was sampled, centrifuged, and then extracted with the Qiagen DNeasy kit [
76]. Twelve primers were used to amplify eDNA, of which the amphibian primer amplified a
CytB region of the mtDNA genome, detecting
T. carolina when analyzed after Illumina sequencing [
76]. This study adds support that turtles can be detected with metabarcoding, despite previous attempts which have been less successful. However, metabarcoding likely can only detect presence [
84]. Taken together, the total of these studies illustrates success in detecting turtle eDNA in aquatic systems, indicating promise for using this population monitoring technique in this increasingly imperiled group. However, work remains to be done in achieving consistent and sensitive detection, and obtaining eDNA quantities that reflect abundance.