• Bulté, Grégory
  • Lougheed, Stephen C.


Background and aims

Overwintering is a critical part of the annual cycle of animals living at high latitudes, and selection of overwintering sites (hibernacula) is important to population persistence. Identifying the overwintering sites of aquatic species is challenging in areas where water bodies are frozen for significant parts of the year. We tested whether environmental DNA (eDNA) approaches could help to locate them.

Materials and methods

We conducted environmental DNA surveys of underwater overwintering sites of the northern map turtle (Graptemys geographica ), a species of conservation concern in Canada. We collected water samples under the ice in winter across a mid‐sized temperate lake and used quantitative PCR with a species‐specific probe to quantify concentrations of map turtle eDNA.

Results and discussion

We found localized eDNA signals consistent with known overwintering sites and one previously suspected site. The latter was further confirmed using underwater remote operated vehicle (ROV) visual surveys.


Our study confirms that eDNA can offer insights on a critical part of the annual cycle of aquatic species, for which we know very little.


Study site and species

We conducted this study in Lake Opinicon, a medium‐sized, shallow lake with average depth of 2.5 m, maximum depth around 11 m, and total area of 7.9 km2 located in southeastern Ontario, Canada (Figure 1). Lake Opinicon is typically frozen between late December and early April, except for some open water along the Rideau Canal at the lakes' eastern terminus, and near two creeks on its southeastern shore. The estimated population size of northern map turtle is 1,529 individuals (1.9 turtles/ha; Bulté, Carriere, & Blouin‐Demers, 2010). Overwintering sites along the shoreline of one island (Eightacre Island; Figure 1) have been identified using radiotelemetry (Carrière et al. 2009) and monitored since 2004 as part of a mark‐recapture study. On the tip of a small peninsula toward the southwest end of the lake, we captured turtles ( = 34) between April 21 and May 12 in 2005, 2006, 2007, and 2018 suggesting the existence of another overwintering site nearby (Figure 1) but were never able to confirm the presence of turtles in the winter using radiotelemetry.

Water sampling and eDNA extraction

Water samples were collected during the 2017 and 2018 winters (Figure 1). On 22 February 2017, we sampled a grid of 36 sites spaced approximately 400 m apart across the southwestern portion of the lake (Figure 2a). This sampling was done by author WF without prior knowledge of the location of suspected or confirmed turtle overwintering sites and was designed to serve as an unbiased test of whether systematic sampling of eDNA could reveal the presence of overwintering map turtles.

Between mid‐February and mid‐March of 2018, we surveyed an additional 85 sites (Figure 2b–d) over 7 days (Table 1) with two goals: (a) To examine the relationship between eDNA concentration and distance from its source (known overwintering sites at the shoreline), where we expect an inverse relationship between distance from the shoreline and eDNA concentration. (b) To assess eDNA detectability around the central basin of the lake and at the aforementioned suspected site where positive detections might reveal an unconfirmed overwintering site. Our sampling focused on four areas: (a) the shoreline of Eightacre Island; (b) one transect extending from each of the two sites where turtle presence was visually confirmed via underwater cameras during sampling; (c) shoreline segments in and around the central basin of the lake; and 4) the neighboring shoreline of the suspected overwintering site in the southwestern portion of the lake. All shoreline samples were collected 5 m away from the shoreline.

Quantitative PCR assay detection

To detect map turtle environmental DNA, we used a probe‐based qPCR assay to maximize specificity for single species detection (Goldberg et al. 2016). We designed a map turtle‐specific cytochrome (cyt) primer pair and a TaqMan™ MGB probe (Table 2) using DNA sequences retrieved from GenBank (Table S1). The primers and probe were designed to maximize mismatches with other co‐occurring turtles (Table 1). A 693 bp fragment of cytcontaining the target 99 bp amplicon (details in Supporting Information) was amplified from blood extracts of one individual map turtle and inserted into a pMiniT 2.0 vector (NEB). DNA concentration of the resulting plasmid was then quantified using a Qubit 3.0 fluorometer (Thermo Fisher) and used as a standard in all qPCR assays. The qPCR amplifications were performed on CFX96 Touch™ Real‐Time PCR platform (Bio‐Rad) with reaction cocktails contained the following: 10 μL SensiFAST™ probe NO‐ROX mix (Bioline), 400 nM forward and reverse primer, 200 nM TaqMan probe, 10 μg Bovine Serum Albumin (BSA), 2 μL DNA template with reverse osmosis H2O added to a final volume of 20 μL. The qPCR cycling profile was as follows: 2 min of polymerase activation at 95°C, 45 cycles of two‐step amplification with 10 s of denaturation at 95°C and 20 s of annealing/extension at 60°C. The 137 working samples were analyzed in seven separate 96‐well PCR plates (BioRad), and for each plate, we established the standard curve using a seven‐point tenfold dilution series from 3 × 106 to 3 copies of template per reaction with laboratory blank controls where the DNA template was replaced by ultrapure water. We performed all qPCR amplifications in triplicate.