Our research involved work with animal subjects (unionid mussels and Round Goby fishes) and was conducted following relevant regulations and standard procedures. The field collections were carried out under Pennsylvania Fish and Boat Commission permits (# 2018-01-0136 and 2019-01-0026). The experimental protocols were approved by Penn State University’s Institutional Animal Care and Use Committee (IACUC# 201646941 and 201646962). All new DNA sequencing data are made publicly available in GenBank (with accession numbers provided in Table 1) and a BioProject (# PRJNA813547) of the National Center for Biotechnology Information40.
Propensity of Round Goby to consume unionid mussels in a controlled lab setting
Stream table setup
We conducted lab experiments to observe the potential predation of juvenile freshwater mussels by the Round Goby, following standard research protocols for work with animal subjects (IACUC# 201646962, Penn State University). We constructed four artificial stream tables in an aquatic laboratory, each measuring 3 × 2 m and featuring two run and two pool sections (each 0.63 × 0.56 × 0.46 m). Water flow was produced using eight Homsay 920 GPH submersible water pumps, which pumped water from a central reservoir tub into each table at the start of each run section. The water flow direction was clockwise for stream tables 1 and 3, and counterclockwise for stream tables 2 and 4. Water pumped into the stream tables exited via two drains located medially of each run section, where it flowed back to the central reservoir tub. Each stream table was filled with a 6 mm layer of substrate consisting of a mixture of sand, gravel (4–6 mm), and crushed stone (size 2B, with an average size of ~ 19 mm). The day before each experiment, field technicians traveled to local streams and collected macroinvertebrates using one minute D-frame kick net samples for each of the four stream tables. The macroinvertebrates and associated substrate were transported back to the facility and were introduced into each stream table system.
Preferential feeding experiments
Before each experiment commenced, juvenile Plain Pocketbook mussels (Lampsillis cardium) were introduced into each stream table (with 165 mussel specimens for experiment 1 and 100 mussels for experiments 2 and 3). This widespread and abundant species is not imperiled in Pennsylvania, and mussels were provided for this study by the White Sulphur Springs National Fish Hatchery located in southeast West Virgina. The mussels were allowed to acclimate in the stream tables for 2 h before commencing each experiment. Ten Round Gobies were introduced into each stream system (stream tables 1 and 2 for experiment 1, and all tables for experiments 2 and 3). The total length (from nose tip to caudal tip) of each fish was measured prior to introduction and after the termination of experiments 2 and 3. Experiment 1 was conducted for 3 weeks, while experiments 2 and 3 were conducted for 8 days. During these experiments, Round Gobies were allowed to exist in the systems and feed preferentially, on the mussels and macroinvertebrates, for the allotted time before each investigation concluded. We acknowledge that in these experiments, the mussel abundances are higher and macroinvertebrate densities lower and less rich than commonly occur in the natural stream environment. Further, the Round Goby fish densities used are much higher than currently in the French Creek watershed, though are comparable to what is currently seen in parts of the Great Lakes basin. Nonetheless, the experiment scenarios allowed us to observe if Round Gobies would consume the mussels when given the choice to feed on a variety of food items.
Evaluation of unionids consumed by fish
Round Gobies were removed from the stream tables upon completion of each experiment. They were euthanized using > 250 mg/L buffered (pH ~ 7) tricaine-S (MS222) solution. The fish were submerged for 10 min beyond the cessation of opercular movement to ensure proper euthanasia, and tissues were collected after we confirmed complete euthanasia—compliant with AVMA guidelines and approved by the IACUC protocol. The Round Gobies were placed in a 10% solution of formalin for preservation, and after 2 weeks, they were rinsed with clean water and were placed in 70% ethanol for long-term storage. After fish were removed from the system, the water was drained, and the substrate was sifted to recover the remaining mussels. Mussels were counted, and live individuals were returned to holding tanks for use in subsequent experiments. To further assess whether Round Gobies had consumed mussels during the investigation, Round Gobies were x-rayed using a Bruker Skyscan 1176 micro-CT scanner. After that, the stomachs of each fish were excised, and the contents examined using a Leica CME dissection scope to confirm the identity of Plain Pocketbook mussels. Contents posterior to the stomach were not analyzed because they could not be reliably counted and identified.
DNA metabarcoding to identify mussel species consumed by Round Goby in a stream setting
Fish and mussel sample acquisition
We collected 39 Round Gobies directly from streams within the French Creek watershed—their newly invaded natural stream habitats—in June 2018. We aimed to quantify which species, if any, of unionid mussels they consumed. Fish collection locations included LeBoeuf Creek at Moore Road and 100 m below the confluence of French Creek and LeBoeuf Creek. The unionid mussel populations and the environmental field settings at these locations are detailed by Clark et al.19. A team of field technicians collected fish by kick seining (3 m × 1 m × 9.5 mm nylon mesh) while moving downstream. Seining was the sampling method of choice compared to electrofishing to avoid possible regurgitation of food items prior to excision of fishes’ stomachs. The stream reaches sampled at each location were between 100 and 200 m in length and included riffle, run, and pool habitats. In addition to fish samples, unionid mussel samples from French Creek were also collected for analysis (under Pennsylvania Fish & Boat Commission collectors permits # 2018-01-0136 and 2019-01-0026). Following standard research protocols (under IACUC# 201646941, Penn State University), the Round Gobies collected were euthanized using buffered Tricaine-S (MS222) solution; and stomachs were excised using sterilized utensils before being placed in sterilized tubes filled with 97% ethanol. After excision of stomachs, fishes were placed in a 10% formalin solution for preservation. After 2 weeks, fishes were rinsed with clean water and transferred to 70% ethanol for storage. The stomach samples were immediately placed in ethanol and on ice in the field. Samples were stored in a freezer before being shipped to the US Geological Survey’s Eastern Ecological Science Center for various molecular ecology analyses. Once the fish and mussel samples arrived at this lab, they were recorded and stored at four °C until analysis.
Primer development
Specific primers targeting a moderately conserved region of the mitochondrial COI gene for 25 species of unionids inhabiting French Creek were designed. Previously a PCR-based amplification method utilizing restriction enzyme digests was used to identify genetic fingerprints of 25 unionid species inhabiting French Creek41. Here, we designed a new degenerate PCR primer set modified with sequencing overhangs to facilitate compatibility with a MiSeq amplicon sequencing method previously designed for 16S Amplicon sequencing. We targeted the locus of the mitochondrial COI gene of unionids known to inhabit the Atlantic Slope Drainage. Consensus sequences were derived using Multalin analysis and a tiling method to identify conserved primer binding regions flanking an ~ 300 bp region of the COI gene. This gene was targeted in part due to the availability of partial or complete sequences representing these target species in the NCBI reference database40. Cytochrome oxidase sequences were downloaded for the 25 unionid mussel species of interest. However, a COI sequence for the Rabbitsfoot (Theliderma cylindrica) mussel was absent from the NCBI database, which required us to sequence this region for an in-house reference (which is described later in the paper). We designed a degenerate primer cocktail specific to all mussel species of interest that amplified a ~ 289 bp product, with forward and reverse primers used for the amplification of unionid specific COI presented as supplemental information (see Table S-230. We evaluated the suitability of the primers using samples from field identified mussels. For primer optimization, PCR was run across a gradient of annealing temperatures to determine suitability. In addition, we used Round Goby DNA as a template to evaluate specificity. In addition to Round Goby stomach samples, mussel samples of several species collected from French Creek were included as positive controls.
DNA extraction from tissue samples
Following the manufacturer’s protocols, tissue samples (including fish stomach and mussel tissue) were extracted with the Zymo Research ZymoBIOMICS 96 MagBead DNA Kit (San Diego, CA). Random samples of DNA extracts were analyzed on an Agilent 2100 Bioanalyzer using a high-sensitivity assay kit. Fragments in the target amplicon range were apparent (albeit not known to be of mussel origin). All samples were stored at − 20 °C until PCR was performed. DNA from both the T. cylindrica and L. complanata samples were analyzed for DNA quality.
Rolling circle amplification of mitochondrial genomes
To acquire COI sequences for T. cylindrica and L. complanata, we subjected archived DNA samples to rolling circle amplification (RCA) followed by amplicon sequencing on the MiSeq. In short, 2 µl of DNA template was added to 2 µl Equiphi29 DNA polymerase reaction buffer containing 1 µl of Exonuclease-resistant random primers (ThermoFisher). Samples were denatured by heating to 95 °C for 3 min followed immediately by cooling on ice for more than 5 min. A volume of 5 µl was added to an RCA master mix containing 1.5 µl of 10 × Equiphi29 DNA polymerase reaction buffer, 0.2 µl of 100 mM dithiothreitol, 8 µl of 2.5 mM dNTPs, 1 µl of Eqiphi29 DNA polymerase (10U) and 4.3 µl of nuclease-free water. The samples were heated to 45 °C for 3 h and then 65 °C for 10 min. Samples were then placed in ice and then frozen at − 20 °C. All RCA products were normalized to 0.2 ng/µl in 10 mM Tris–HCl, pH 8.5. Normalized RCA product was utilized as a template for an Illumina Nextera XT library preparation. Sequencing libraries were prepared following the Nextera XT Library Preparation Reference Guide (CT# 15031942 v01) using the Nextera XT Library Preparation Kit (Illumina, San Diego, CA). Final libraries were analyzed for size and quality using the Agilent BioAnalyzer with the accompanying DNA 1000 Kit (Agilent, Santa Clara, CA). Libraries were quantified using the Qubit H.S. Assay Kit (Invitrogen, Carlsbad, CA) and normalized to 4 nM using 10 mM Tris, pH 8.5. Libraries were pooled and run on the Illumina MiSeq at a concentration of 10 pM with a 5% PhiX spike with run parameters of 1 × 150. Bioinformatic processing of this data is outlined below.
Amplification of the cytochrome oxidase 1 gene
Extracted genomic DNA was used as template for end-point PCR. Samples evaluated were from mussels and round gobies (see supporting Table S-330). The ~ 289 bp COI region was amplified with the mussel primers as follows. The amplification reaction contained 0.15 µM of each primer, 1 µL of the initial amplification product, and Promega Go Taq Green Master Mix following manufacturer recommendations for a 25 µL reaction. The thermocycler program consisted of an initial denaturing step of 95 °C for 3 min, followed by 30 cycles of 30 s at 95 °C, 30 s at 52 °C, and 1 min at 72 °C. Products were subjected to a final extension of 72 °C for 5 min then held until collection at 12 °C. An appropriately sized amplification product was confirmed for each reaction by electrophoresis of 5 µL of the reaction product through a 1.5% I.D. N.A. agarose gel (FMC Bioproducts) at 100 V for 45 min. PCR products were cleaned with the Qiagen Qiaquick PCR purification kit (Valencia, CA) and quantified using the Qubit dsDNA H.S. Assay Kit (Thermofisher Scientific, Grand Island, NY). Samples were diluted in 10 mM Tris buffer (pH 8.5) to a final concentration of 5 ng/µL.
Generation of mock mussel samples
To better understand and minimize sources of error or bias in the taxonomic assignment, we created a mock extraction by mixing sequences from known mussel taxa at defined concentrations. For each mussel, approximately 25-mg of tissue was extracted with the ZymoBIOMICS 96 MagBead DNA Kit (San Diego, CA) following the manufacturer’s protocol. The COI sequence was amplified from each species using the same primer-protocol combination described above. A total of 5 PCR products were mixed at equal concentration (mass/volume) to generate the mock sample (“Mock” hereafter). To confirm the identity of these inputs, each COI region was amplified and sequenced on the Illumina MiSeq during the same run as the Mock and samples.
Sequencing library preparation and quality assessment
Next-generation sequencing was performed on the Illumina MiSeq platform to observe species-specific sequences and determine the diet of the Round Goby. Inclusion of the overhangs on the amplification primers allowed us to utilize the Illumina 16S Metagenomic Sequencing Library Preparation protocol42. Amplicon libraries were prepared following the same manufacturer’s protocol. All samples were indexed using the Illumina Nextera XT multiplex library indices. DNA read size spectra were determined with the Agilent 2100 Bioanalyzer using the Agilent DNA 1000 Kit (Santa Clara, Calif.). Libraries were quantified with the Qubit dsDNA H.S. Assay Kit (ThermoFisher Scientific, Grand Island, N.Y.) and normalized to 4 nM (nM) using 10 mM (mM) Tris (hydroxymethyl) aminomethane buffer pH 8.5. A final concentration of 10 picomolar library with a 6.5% PhiX control spike was created with the combined pool of all indexed libraries. All bioinformatic operations were completed on CLC Genomic Workbench v20 (Qiagen, Valencia, Calif.).
Read filtering, trimming, and RNAseq metabarcoding assembly
FASTQ files from the sequencing runs were imported as paired-end reads into CLC Genomics Workbench v20.0.4 (Qiagen Bioinformatics, Redwood City, Calif.) for initial filtering of exogenous sequence adaptors and poor-quality base calls. The trimmed overlapping paired-end reads were mapped to the 25 target unionid sequences specific for the species of interest. Several mapping iterations were run using different levels of stringency. We utilized + 2/− 3 match-mismatch scoring and set the length fraction to 0.90. Analyses were iterated using different similarity fractions ranging from 0.90 to 0.99. Reads were annotated, and relative abundance was determined using a curated reference library (see supporting Datasets S-1 and S-230).
Source: Ecology - nature.com
