Environmental DNA detection and quantification of invasive red-eared sliders, Trachemy scripta elegans, in ponds and the influence of water quality

Study site

We conducted the field survey in 100 ponds that were located in Himeji, Japan (34°47′–34°54′N, 134°35′–134°45′E, Fig. 1) between July 21 and November 16, 2016. We randomly selected ponds located in each area category, comprising city, rural, and mountain areas (see Tables 1A, 1B, Fig. S1 in SEM). There are few ponds in the southern (city area) and northern areas (mountain) because the distribution of the ponds is biased. We conducted statistical analysis among the three area categories and eDNA concentration, but did not find remarkable patterns of eDNA concentration and detection among the pond locations. The field survey and pond sampling was permitted by the land owners, if needed.

Field survey and sampling

We recorded the presence/absence and number of red-eared sliders, based on visual observations from the shore line for three minutes by an expert (A. Kakuda). We performed surveys during the daytime (1000–1300 h) each day, except for rainy days and one day after rain. From a point within each pond, 500 mL of surface water was collected for eDNA and SS analysis, and a further 100 mL of surface water was collected for Chl. a and water quality analysis. In this study, we conducted a field survey to compare the convenience of the eDNA method with visual surveys; thus, we decided to increase the number of sampling ponds rather than increase the replications at each pond. We selected a water sampling point in the middle section of the ponds, far from the water outflows/inflows. We directly sampled the eDNA using a bleached bottle and added 0.5 mL of benzalkonium chloride (BAC) to avoid a reduction of the eDNA concentration in the samples (Yamanaka et al., 2016). Samples were stored in a cooler box with a ‘cooler blank’. The ‘cooler blank’ contained 500 mL of DNA-free water, which we brought to the field, and it was treated identically to the other water samples, except that it was not opened at the field sites.

Water preparation

Within six hours after water sampling, the samples were filtered onto a GF/F glass filter (47 mm diameter and 0.7 µm pore size, GE Healthcare Japan, Tokyo, Japan). We used separate filters for the eDNA, SS (from 500 mL water), and Chl. a (from 100 mL water) analysis. The filter was then wrapped in commercial aluminum foil and stored at −20 °C until eDNA extraction, or SS/ Chl. a measurement. For eDNA samples, the ‘cooler blank’ and a ‘filter blank’ consisting of DNA-free distilled water were filtered in the same way as the samples. To avoid contamination, each piece of equipment that was used in the water sampling or filtration was soaked in a 10% commercial bleach solution (approximately 0.6% sodium hypochlorite) and rinsed using DNA-free distilled water prior to reuse. The 80 mL of the filtrated samples were stored at −20 °C until further water quality analyses. We lost some of the samples for water quality measurements (see the missing values of Tables S1A, S1B in SEM).

DNA extraction from the filters

The eDNA was extracted from each filter using a DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) based on the method described by Uchii, Doi & Minamoto (2016). Each filtrate was soaked in 400 µL of buffer AL and 40 µL of protease K in a Salivette tube (Sarstedt, Nümbrecht, Germany) and incubated at 56 °C for 30 min. After centrifugation at 5,000 g for 5 min, 220 µL of TE buffer (pH 8.0) was added to each filter, and the tubes were centrifuged again in the same way after being kept for a minute. The 200 µL of buffer AL and 600 µL of 100% EtOH were then added to each filtrate and mixed by pipetting. The mixture was applied to a DNeasy Mini spin column and prepared according to the manufacture’s manual. The sample solution (100 µL of buffer AE) was stored in a 1.5 mL microtube at −20 °C until qPCR analysis.

Quantitative real-time PCR (qPCR)

The eDNA was measured with four PCR replicates using a PikoReal Real-Time PCR System (Thermo Scientific, Waltham, MA, USA). To detect and quantify the DNA of the red-eared slider using qPCR, the mitochondrial cytochrome b gene fragments were amplified and quantified with the following primers and probe: Tse-Kako-A-F (5′-CCTCCAACATCTCTGCTTGA -3′), Tse-Kako-A-R (5′-ATTGTACGTCTCGGGTGATG-3′), and Tse-Kako-A-MGB-P (5′-FAM- CGGAATTTTCTTGGCTATAC -MGB-3′). We used the consensus sequence of Trachemy scripta elegans with the following accession numbers: FJ770617, FR717131, EU787024, AF207750, and HQ442420 in NCBI (http://www.ncbi.nlm.nih.gov/). From the consensus sequence, we artificially synthesized a standard gene, which included the amplicon region and 20 bases each of the upper and lower sequences (5′- ATTCATTGATCTACCAAGCCCCTCCAACATCTCTGCTTGATGGAACTTTGGATCCTTATTAGGTACTTGCCTAATCCTACAAATCCTTACCGGAATTTTCTTGGCTATACACTACTCCCCAGACATTTCACTAGCATTCTCATCAGTAGCCCACATCACCCGAGACGTACAATACGGATGACTTATTCGTAAT-3′). The specificity of the probe and primers was confirmed by Primer-BLAST and testing on Japanese turtles (Mauremys japonica, Mauremys reevesii, and Pelodiscus sinensis). Each TaqMan reaction contained 900 nM of each primer, 125 nM of TaqMan probe, 5 µL qPCR master mix (TaqMan Environmental Master Mix 2.0, Thermo Scientific, Waltham, MA, USA), 0.2 µL AmpErase^® Uracil N-Glycosylase (UNG, Thermo Scientific, Waltham, MA, USA), and 2 µL of the DNA solution. The total volume of each reaction mixture was 10 µL and we performed four replicates for PCR. The PCR conditions were as follows: 2 min at 50 °C, 10 min at 95 °C, and 55 cycles of 15 s at 95 °C and 60 s at 60 °C. These cycles are based on various previously described methods (Thomsen et al., 2012a; Tréguier et al., 2014; Katano et al., 2017; Doi et al., 2017b). The qPCR results were analyzed using PikoReal software ver. 2.2.248.601 (Thermo Fisher Scientific, Waltham, MA, USA), and the standard curves were also automatically estimated by this software. The R² values of the standard curves ranged from 0.960 to 0.989, and PCR efficiency ranged from 73.09 to 118.56%. For the R² values, there is a possibility that we measured the eDNA in the qPCR system with order-level differences because of the high variance in eDNA concentration, from 0.21 to 48,016.44 (copies/L). We defined the limit of detection (LOD) for red-eared slider DNA as one copy per reaction based on a qPCR assay of the four replicates. Each real-time PCR assay included four no template controls (NTCs) and we also measured the cooler and filter blanks with four replicates. We used the average value of the four replicates for each eDNA concentration. All of the above qPCR procedures were based on the MIQE checklist for qPCR (Bustin et al., 2009). We performed the PCR set up and real-time PCR in two separate rooms to avoid contamination.

PCR inhibition test

We compared the Ct shift between the samples and controls with the same number of known target DNA copies, based on the method by Doi et al. (2017b), to confirm the degree of PCR inhibition. Ct is defined as the number of cycles required for enough amplified PCR product to accumulate that it surpasses a threshold recognized by the real-time PCR instrumentation. The Ct is inversely related to the starting quantity of the target DNA in a reaction and is used to calculate this quantity. The presence of PCR inhibitors will shift (delay) the Ct for a given quantity of the template DNA. To confirm the effects of water quality factors on eDNA inhibition, we did not use a PCR inhibition removal kit (e.g., OneStep PCR Inhibitor Removal Kit, Zymo Research) for the qPCR preparation. To test for PCR inhibition in the DNA samples, 1 µL of plasmid including the cytochrome b gene from Trachurus japonicus (1.5 × 10⁴ copies), which is a marine fish that does not inhabit the sampled ponds, was added to the PCR tempelate with 1 µL of DNA-free distilled water. We used the primer and probe set that was reported by Yamamoto et al. (2016): forward primer: 5′-CAGATATCGCAACCGCCTTT-3′; reverse primer: 5′-CCGATGTGAAGGTAAATGCAAA-3′; and probe: 5′-FAM-TATGCACGCCAACGGCGCCT-TAMRA-3′. The PCR conditions were the same as above. Each real-time PCR assay included three no template controls. We used the average value of the replicates for each Ct value. ΔCt ≥ 3 cycles were considered to be evidence of inhibition (Hartman, Coyne & Norwood, 2005).

Water quality analysis

We measured phosphate (PO₄-P), nitrate (NO₃-N), total phosphorus (TP), total nitrogen (TN), dissolved organic matter (DOM), and total organic matter (TOM) from the filtrate, according to the methods of Saijo & Mitamura (1995), using a spectrophotometer (HITACHI U-5100, Hitachi, Tokyo, Japan). The absorbance of DOM was read at 254 nm, using samples not in an autoclave.

SS measurement

We used the GF/F glass filter, which had been burned and dried before the weight was measured by electric balance (Sartorius CPA2252), prior to the SS analysis. The filtrate was dried in the 60 °C automatic oven (Yamato DX402, Yamato, Japan) over 12 h before the weight was measured. After that, we burned the dried filter at 450 °C for 2 h using an electric muffle furnace (Yamato FO410), then measured the weight again in the same way. The SS content was calculated as follows; (450 °C burned weight)–(60 °C dried weight).

Chl. a measurement

We extracted the Chl. a of the filter by immersing it in 99.5% ethanol over 12 h. The extracts were measured at 630, 645, 663, and 750 nm absorbances by the spectrophotometer (HITACHI U-5100). The Chl. a concentration was determined according to the following equation (UNECSO 1969): ${\displaystyle Chl.a\left(mgL-1\right)=\left\{\left(11.64\times E663-2.16\times E645+0.1\times E630\right)\times k\right\}∕V}$

where, k: ethanol for extraction (mL); E663, E645, and E630: each absorbance at 663, 645, 630 nm, excluding the absorbance at 750 nm; and V: water samples (L).

Statistical analysis

We used a linear model (LM) to evaluate the relationship between the red-eared slider eDNA concentration and the number of red-eared sliders observed visually. For the eDNA concentration and the number of red-eared sliders reported by visual observation, we used the data of the sites where red-eared sliders were detected, regardless of the eDNA detection (11 ponds in which red-eared sliders were detected by both methods and 10 ponds in which they were detected only by visual observation; N = 21; Tables S1A, S1B). In addition, we used a multiple linear regression to evaluate the relationship between the eDNA concentration and the following environmental factors: Chl. a, SS, PO₄-P, NO₃-N, TP, TN, DOM, TOM, and estimated number of red-eared sliders (animals/km²; from the result of eDNA concentration and the number of red-eared sliders observed visually). We used the data of all sites in which eDNA was detected (11 ponds in which red-eared sliders were detected by both methods and nine ponds in which they were detected by only eDNA, N = 20, Tables S1A, S1B). Prior to the multiple linear regression, we used a variance inflation factor (VIF) to check the collinearity of the factors. The maximum VIF was 59.685, indicating that co-linearity among the factors would influence the results of the multiple linear regression. Thus, we removed the factors with a VIF >5 to reduce the collinearity effect on the multiple linear regression, resulting in the Chl. a, TP, and number of red-eared sliders remaining in the final linear model (LM) analysis. All statistical analysis and graphics were conducted in R ver. 3.4.1 (R Core Team, 2018) with the “ggplot2” and “car” packages.

The relationships between eDNA measurements and visual observations

Of the 100 surveyed ponds, we detected red-eared sliders in 30 ponds: they were detected in 11 ponds by both visual and eDNA surveys, 10 ponds by only visual survey, and 9 ponds by only the eDNA survey (Tables 1A, 1B). All successful amplifications were considered to be red-eared sliders, because the probe and primers were confirmed to detect only red-eared sliders by Primer-BLAST and were tested by real-time PCR using DNA extracted from Japanese turtles (Mauremys japonica, Mauremys reevesii, and Pelodiscus sinensis) living in the same region. There was a significant positive correlation between the eDNA concentrations in all ponds in which eDNA was detected and the number of red-eared sliders identified by visual observations (LM, R² = 0.48, p < 0.001, N = 20, Fig. 2).

The ΔCt values in 99 of the 100 ponds, except pond No. 33, were lower than 3 (0.27 ± 0.22, mean ± SD), which means that they were lower than the inhibition criteria (Hartman, Coyne & Norwood, 2005). Thus, PCR inhibition was not significant for all samples, but pond No. 33 showed no amplification in the PCR inhibition test. We did not detect any amplifications in the negative controls and equipment blanks, including the cooler and filter blanks.

The relationship between eDNA concentration and water quality

The results of the water quality analysis are shown in Tables S1A, S1B of the SEM [Chl. a: 0.04 ± 0.10 µg L⁻¹, SS: 35.0 ± 44.34 µg L⁻¹, PO₄-P: 0.78 ± 0.87 µmol L⁻¹, NO₃-N: 38.93 ± 18.25 µmol L⁻¹, TP: 1.00 ± 0.99 µmol L⁻¹, TN: 29.53 ± 30.77 µmol L⁻¹, DOM: 0.07 ± 0.04 (abs = 254 nm), and TOM: 0.00 ± 0.01 (abs = 254 nm), mean ± 1 SD]. The VIFs of the LM for each factor are shown in Table S2 in the SEM, of which the VIFs of the Chl. a (4.391), TP (1.870), and estimated number of red-eared sliders (1.801) were lower than 5. The LM, without the factors and with a VIF of >5, showed that Chl. a was negatively related with the eDNA concentration (LM, p < 0.001, Tables 2, 3), while there was no significant relationship between the eDNA concentration and other factors (Fig. 3, Tables S1A, S1B).