Robotic environmental DNA bio-surveillance of freshwater health

ESP sample processing

The ESP operated autonomously, needing only power, communications and fluid connections through its waterproof pressure housing (Fig. 1). Prior to sample initiation, the ESP was purged completely with nitrogen to reduce oxidative reactions (i.e., DNA degradation) from occurring. At the initiation of sampling, a puck (Fig. 1A cutout) loaded with filter material was placed within a clamp. Valves open to the outside allowed a syringe to sequentially pull water through the puck. Once the target volume was filtered, or the filter was loaded with biomass (i.e., ‘clogged’), filtering stopped and excess water was cleared. Five mL’s of RNAlater preservative was then added to the puck, soaking the filter for 10 min before the excess was evacuated and the puck was returned to storage. Preserved pucks were stored at the ESP temperature, which were similar to ambient air temperatures. The upper limit on the amount of time that an ESP device can operate in the field before DNA quality on a puck is comprised is not known but is at least 21 days¹⁰. A constant humidity kept the pucks moist, allowing for easy filter removal once the instrument was recovered.

To get water to the ESP, we designed an external sampling module from which the ESP drew water¹¹. The sampling module was self-draining, and fed by a submersible pump (WSP-12 V-2 M, Waterra USA Inc., Bellingham, W, USA) installed approximately 0.5 to 2 m below the river water line at each deployment site. To reduce possible carry-over contamination, the sampling pumps, tubing and external sampling modules were flushed with river water for 10 min prior to every sample collection. The sampling port of the ESP itself was cleaned with 10% bleach and a 10% tween-20 solution between samples. At the end of each ESP deployment, pucks were manually removed and filters were aseptically recovered into 2.0 mL screw cap centrifuge tubes and stored at − 80 ºC until molecular analyses were performed.

Field deployments

We performed initial ESP feasibility studies in Yellowstone National Park (USA; Fig. 2) in September 2017. Here, our goal was to determine if the ESP could be used to sample DNA of the waterborne protozoa, Naegleria spp., from a freshwater river where these organisms had previously been detected using standard techniques¹². We filled 60-L sterilized carboys with water from the confluence of the Boiling and Gardner rivers. Carboys were transported to a lab at Montana State University (Bozeman, Montana) and connected to ESP samplers via tubing and syringe pumps. Water was passed through each filter (5-µm Diapore filters) until the filter became clogged; six samples were filtered.

We then integrated the ESPs into two USGS streamgages on the Yellowstone River in 2018 and one USGS streamgage on the Snake River in 2019, (Fig. 1B,C) where we tested for DNA of the fish pathogen, Tetracapsuloides bryosalmonae, the causative agent of salmonid fish Proliferative Kidney Disease (PKD). On the Yellowstone River, we installed ESPs at the streamgage near the upstream and downstream extents of a recent PKD outbreak¹³, USGS 06191500 Yellowstone River at Corwin Springs MT and USGS 06192500 Yellowstone River near Livingston MT, described below as Corwin Springs and Carters Bridge, respectively (Fig. 2). On the Snake River, we installed one ESP at the streamgage 1.5 km downstream of Palisades Reservoir near the upstream extent of a recent PKD outbreak, USGS 13032500 Snake River near Irwin ID. The ESP pucks were loaded with 1.2-µm cellulose nitrate filters. We ran two negative controls (1 L of molecular grade water) through the ESP prior to and at the conclusion of deployment to assess for contamination.

Yellowstone River

The ESPs were programmed to collect 1-L samples every 12 h, from Jul 24 to Aug 26 2018, and every 3 h from Aug 27 to Sep 7 2018. The average (± 1 SE) volume filtered per sample was 639 (± 11) mL, indicating that most filters clogged prior to reaching the 1-L target volume. Filter samples were collected at ambient air temperatures ranging from 9.6 to 35.8 °C ((overline{x})  = 18.9) at Carter’s Bridge and 8.3–29.0 °C ((overline{x})  = 17.1) at Corwin Springs. We compared T. bryosalmonae ESP detections to those from manually collected grab samples from shore (6, 250-mL samples per site filtered through 1.2-µm cellulose nitrate filters) collected at weekly frequencies for the entire length of the ESP deployments and at daily frequencies between Aug 27 and Aug 30. Thus, ESP and manual eDNA samples collected at different temporal intervals (3 h, 12 h or weekly) allowed us to evaluate the added value of higher frequency sampling.

We also evaluated the utility of automated high frequency sampling to detect a new invasion by introducing novel DNA of Scomber japonicas (mackerel fish) 100 m upstream of each Yellowstone River streamgage. On Aug 27, we introduced 3 kg of canned S. japonicas 100 m upstream of the water sampling inlet for each ESP. S. japonicas was blended with water, frozen and then placed within metal-wire minnow traps and anchored to the river’s bottom with cement pavers. The ESPs were programmed to sample every 3 h from Aug 27 to Sep 7. Manual grab samples (600 mL) were collected 10 m (n = 3), 100 m (n = 6), and 400 m (n = 3) downstream of the S. japonicas in order to test that S. japonicas DNA was transported downstream past the water sampling inlet of each ESP. Manual grab samples were collected immediately prior to S. japonicas introductions, 3 h post-introduction and then every 24 h for 3 days.

Snake River

The ESPs were programmed to collect 2-L samples every 12 h from Jul 17 to Sep 09 and then every 4 h from Sep 10 to Oct 1, 2019. Manually collected grab samples (three, 2-L samples filtered through 1.5-µm glass fiber filters) and negative field controls (1, 2-L sample of deionized water filtered through 1.5-µm glass fiber filters) were collected every 2 weeks following methods in Sepulveda et al.⁷. Filter samples were collected at ambient air temperatures ranging from 3.9 to 30.2 °C ((overline{x})  = 20.6). To broaden our taxonomic assessment, we tested these samples for T. bryosalmonae DNA, and also for kokanee salmon (Oncorhynchus nerka) and dreissenid mussel (Dreissena spp.) DNA. O. nerka only occur upstream in Palisades Reservoir and at such low abundances that they are not captured by resource managers in annual population surveys⁷. Dreissenid mussels have not yet been observed, but are the principal focus of aquatic invasive species monitoring programs in this region⁷.

Molecular analyses

Filters were removed from the pucks and then shipped frozen to the USGS Upper Midwest Environmental Science Center (LaCrosse, Wisconsin) for DNA extraction and quantitative PCR analyses. Filters were handled and stored in a dedicated room that is physically separated from rooms where high-quantity DNA extraction and PCR product or high-quality DNA is handled. We used the FastDNA SPIN kit for soil to extract DNA on samples from the Boiling River-Gardiner River confluence, following modifications described in Barnhart et al.¹⁴. To extract DNA from Yellowstone River and Snake River samples, we used the Investigator Lyse & Spin Basket Kit (Qiagen, Hilden, Germany) in concert with the gMax Mini genomic DNA kit (IBI Scientific), following manufacturer’s instructions, and eluted in 200 µL of buffer. Samples were extracted as site specific batches and one extraction control was collected per batch. We used previously published assays, limits of detection and methods therein for analyses of Naegleria spp.¹², T. bryosalmonae¹³, S. japonicas¹⁵, O. nerka⁷, and Dreissena spp.¹⁶ (Table 1).

We analyzed all samples in four replicate 25 µL reactions containing 2 µL of template DNA, 1 × Perfecta Toughmix (Quantabio), 400 nM forward and reverse primers, and 100 nM probe. Each plate contained 10 no-template PCR controls (one for each sample) using 2 µL of molecular grade water as the template as well as a standard curve with two replicates of 20,000 and 2,000 copy standards and four replicates of 200 and 20 copy standards. The standards were prepared with synthetic gBlocks (Integrated DNA Technologies) containing the amplicon sequences for each assay. Each sample was also analyzed in three replicates with 200 copies of synthetic gBlock spiked in to check for PCR inhibition. Any sample that indicated less than an average of 60 to 70 copies of targeted DNA in these triplicate samples was considered inhibited. Field and extraction negative controls were analyzed as regular samples. No negative controls amplified.

Analyses

Samples were scored as positive when one or more PCR replicates amplified for the target DNA. We used McNemar’s Exact Test to compare binary qPCR data (detection/non-detection) of T. bryosalmonae and O. nerka DNA between ESP and manually collected samples in the Yellowstone and Snake rivers.

Source: Ecology - nature.com