A real‐time PCR assay to detect predation by spiny dogfish on Atlantic cod in the western North Atlantic Ocean

Abstract Conventional observations show spiny dogfish (Squalus acanthius Linnaeus) rarely eat Atlantic cod (Gadus morhua Linnaeus; 0.02% of stomachs) in the northwestern Atlantic Ocean. Critics express concern that digestion may limit species‐level prey identification, and with recovery from overfishing, dogfish populations may be suppressing cod by competition or predation. This study applied a real‐time PCR TaqMan assay to identify cod in dogfish stomachs collected by cooperating fishing boats during normal trawling operations (May 2014–May 2015; Gulf of Maine, Georges Bank). Conventional methods observed 51 different prey taxa and nearly 1,600 individual prey items, but no cod were observed. Cod DNA was detected in 31 (10.5%) of the dogfish stomachs, with a higher percentage of these from the homogenate of amorphous, well‐digested prey and stomach fluids (20 stomachs or 65%) than from discrete animal tissues (11 stomachs or 35%). Re‐examination of photographs of these 11 tissue samples revealed one whole, partially digested fish that could be recognized in hindsight as cod. Cod DNA was observed in dogfish stomachs year round: in January (1 of 1 trip), February (1 of 1), May (1 of 3), June (0 of 1), July (3 of 4), August (1 of 2), and October (3 of 3). Although these data suggest higher interaction rates between dogfish and cod than previously observed, addressing the population consequences of this predator–prey relationship requires a robust sampling design, estimates of digestion rates by dogfish to account for complete degradation of DNA sequences, and consideration for dogfish scavenging during fishing operations.


| INTRODUC TI ON
Spiny dogfish (Squalus acanthius Linnaeus), after being declared overfished on the northeast U.S. continental shelf in 1998, recovered in about a decade to a higher but variable abundance (Sagarese et al., 2015). Historic diet data suggest that spiny dogfish's role as a piscivorous predator of several groundfish has not been significant (Link, Garrison, & Almeida, 2002), but a recent, simulated food web model suggests that spiny dogfish's role as a competitor and predator has changed as its population has shifted from a depleted to a rebuilt state (Morgan & Sulikowski, 2015). The focus of such debate is often whether spiny dogfish may be regulating the depleted status for DNA.
Our focus here was to quantify dogfish predation to assess its interaction with regional otter-trawling fisheries. Therefore, samples of fish were collected from normal operations of commercial fishing boats in two areas: the Gulf of Maine and on Georges Bank. Such a field sampling approach has limits, because it lacked the statistical design to infer predation rates at more general spatial scales or seasonal periods. However, it begins to address a concern by industry partners that predation rates of dogfish on cod may be higher if a new sampling tool was available. To do so, the molecular results were compared to typical, macroscopic analysis of prey with the same stomach samples. We also consider the potential to apply this approach more broadly: to estimate total population consumption rates by spiny dogfish on Atlantic cod for a comparison to previous work by Link et al. (2002).   , 1977-2017[Smith & Link, 2010). In terms of statistical power, setting aside that these fishery-dependent versus fishery-independent approaches employ different sampling designs (e.g., varying space and nonrandom versus stratified-strata, random-station approaches, respectively), power analysis showed that a threshold of only 2 cod in 295 stomachs would have been significantly more frequent than this reference (i.e., Prob. = .0014 using a glm with a Poisson probability distribution).

| Predator sampling
All samples of dogfish per fishing trip, save one, were dominated by females (total of 245 females, 50 males). Fish ranged in size from 56 to 97 cm total length. Whole dogfish were kept on ice until dissected within 24 hr.

| Prey sampling
Stomachs were excised whole, wrapped individually in a plastic bag, and frozen (−20°C). Later, these stomachs were thawed at 4°C, carefully cut open and the total contents removed and weighed. To minimize cross-contamination of DNA for molecular examination, nondisposable labware were soaked in 10% bleach solution, rinsed with sterile, deionized-distilled water, and autoclaved between individual samples; dissecting tools were scrubbed and flame sterilized between samples.
Stomachs that appeared empty were gently scraped or swabbed for a DNA sample. If not, the gut contents were first examined macroscopically, identified, and separated to the lowest taxonomic level and weighed individually. Care was taken to keep identifiably individual prey whole, so it could be analyzed as a unique sample.
Well-digested prey (WDP) and any remaining stomach fluids (water, small tissues, chyme, mucous, sediment) were separated from the rest of the contents as much as possible and treated as an individual subsample. For reference, all of the remains were photographed.
Invertebrates, vegetation, miscellaneous detritus, and verifiable noncod fish were excluded from further processing.
Processing of the potentially multiple sub-samples within the stomach varied depending on whether a sample was discrete animal tissue or it was WDP or stomach fluids. Discrete animal tissue was firm and intact, such as a whole or partial fish, or pieces of bones, skin, fins etc. Possible contamination of DNA from multiple species was reduced in the final sample by either (1) preferably excising a final tissue sample from the interior of the specimen, or (2) gently scraping the tissue sample to remove surface impurities, soaking in 1%-2% bleach for 60 s, and rinsing in sterile water (Buser, Davis, Jimenez-Hidalgo, & Hauser, 2009;Mitchell, Allister, Stick, & Hauser, 2008). Tissue samples were frozen in 2 ml cryovials at −80°C, and replicate tissues, if available were also preserved in 95% EtOH. The WDP and stomach fluids were treated separately from the discrete animal tissue described above.
This included soft, amorphous, and well-digested remains, ctenophore or ctenophore-like soft tissues, as well as sediment, mucous, chyme, liquid, or a sample recovered from gently scraping the stomach lining.
In order to not miss any target DNA (Rosel & Kocher, 2002), accumulated remains over 50 ml were pulse homogenized with a Waring style blender. A homogenizing probe (Omni Tissue Homogenizer, Omni International, Kennesaw, GA, USA) was used for smaller volumes.

| DNA extraction
A total of 291 dogfish stomachs were examined by molecular techniques for the presence of cod. DNA was extracted from 625 tissue and homogenized dogfish gut samples as well as 22 reference prey tissues (Table 1) for a total of 647 samples.
Extraction of the DNA from samples of discrete animal tissues used the salt precipitation method of Aljanabi and Martinez (1997) or the QIAamp Mini Kit (Qiagen Hilden, Germany). For the homogenized and stomach scrapings (also homogenized), the QIAamp Fast DNA Stool Mini Kit (Qiagen) was used.
Quantification and purity of the extracted DNA was measured spectrophotometrically with a Nanodrop 2000 and the concentrations were adjusted to 50-100 ng/µl. The DNA samples were stored at −20°C.

| DNA quality and gadid screening
Stomachs are highly acidic and degenerating environments for tissues and nucleic acids. Therefore an important step before further processing is to determine the usability of the extracted DNA.
Amplification of DNA from all samples used conventional PCR (cnPCR) (Bastien, Procop, & Reischl, 2008). Universal mitochondrial  Black bubbles indicate locations where cod fishing was actively occurring during sampling. Note, the northernmost sampling location is composed of two separate collections each with one positively identified cod and active cod fishing In order to narrow the focus to cod and cod-like prey, all sample DNA was screened to select only the gadid-related remains for subsequent real-time PCR (King, Read, Traugott, & Symondson, 2008;Rosel & Kocher, 2002). The screening used a cnPCR procedure adapted from Taylor et al. (2002), where a short, 103 bp region of the ATP synthase subunit six (ATPase6) and eight (ATPase8) genes of cod was amplified; Gadid Forward: 5ʹ GCA ATC GAG TTG TAT CTC TAC AAG GAT 3ʹ and Gadid Reverse: 5ʹ CAC AAA TGA GCT CCT CTT CTT GC 3ʹ. Since this target is much shorter than the 16S rRNA gene (500-650 bp) and perhaps more likely to be encountered in degraded samples, the few samples negative (3.4%; Table 2) from the previous step were also included. The cnPCR mix was the same as described for the 16S quality check. Cycling conditions included an initial denaturation at 94°C for 3 min, 35 cycles of denaturation, annealing, extension: (94°C for 30 s, 58°C for 30 s, 68°C for 30 s), a final extension at 68°C for 5 min, followed by 4°C hold. The PCR amplicons were visualized with 3% agarose gel stained by ethidium bromide. Only those samples producing a single 103 bp band were used in real-time PCR assays.  (Taylor et al., 2002). Potential cross-reactions were ruled out by checking the primer pair and probe GenBank using Primer-BLAST and BLASTn.

| Real-time PCR
The specificity of the cod-specific TaqMan probe was checked against reference tissues commonly found in dogfish stomachs (Smith & Link, 2010). All species used for the specificity tests were collected during a spring, 2014, groundfish survey conducted by the Northeast Fisheries Science Center, except for a cusk sample from a 2015 long line survey and a fourbeard rockling from a 2015 spring groundfish survey (Table 1). Real-time PCR assays with the cod-specific TaqMan assay were performed with all 22 reference prey. Only the cod voucher DNA produced a positive signal (Table 2). Subsequently, the same cod DNA was also used as the positive control for all real-time assays.
Reactions were conducted in a final volume of 20 μl containing BSA was also performed. The addition of BSA without further diluting the template was found to be the optimum treatment and was used for all of the subsequent real-time TaqMan assays (Table 3).
To determine the efficiency of the real-time primers and probe, a standard curve of 10 fold serial dilutions of cod DNA was tested. Two rounds of assays were performed with the gadid positive samples. In the first round, a sample was considered positive for cod when 2/3 or 3/3 replicates were called "Present" with a confidence level of at least 99.0%. For the 2nd round, the assays were again repeated in triplicate but only with the samples having 2/3 or 3/3 positive wells. If there were multiple positive samples from the same stomach in the 1st round, only the one with the highest positive signal was repeated. A sample was considered positive for cod when at least 5/6 wells from both the rounds were called "Present" with confidence levels of at least 99.0%.

| Macroscopic examination
In aggregate, 295 dogfish yielded 35.53 kg of stomach contents, comprised of 51 different prey taxa and nearly 1,600 individual prey items ( Figure 2). By weight, the majority of prey taxa were unidentified and were treated as either individual items of well-digested prey could not be distinguished between vertebrate or invertebrate tissue, and 3% were ctenophore-like or miscellaneous but unknown.
The composition of the 289 homogenized subsamples was WDP (69%), mucous or mucous and water (24%), and ctenophora or "ctenophora-like" tissue (7%). However, a well-digested juvenile fish produced a strong positive reaction with the cod probe. After consulting the photo of the fish, portions of the head still retained a mottled skin pattern characteristic of cod juveniles (Figure 3).

| Molecular results
cnPCR with universal mitochondrial 16S rRNA gene primers successfully amplified samples that were not ruled out as noncod by

Results of trials to eliminate failures of the TaqMan Exogenous
Internal Positive Control (IPC) in some of the initial reactions are given in Table 3 In view of these results, BSA was added to all tissue and homogenate real-time reactions without diluting the templates. None of the subsequent reactions resulted in failure of the IPC. (Table 3).
A high correlation (R 2 = .98) was evident between the dilution factor and C t (Figure 4). In addition, the slope produced a very high primer-probe efficiency of 95.41%.
F I G U R E 2 Percent mass (Mass) and frequency of occurrence (FO) of major prey species and groups for spiny dogfish Squalus acanthius identified by macroscopic and molecular examination. The number of stomachs, n, and the number of tows in parentheses are provided. Unidentifiable animal remains denoted by "AR."

F I G U R E 3
Initially classified as unidentified fish remains, real-time PCR identified this as Atlantic cod (Gadus morhua). The mottled pattern on head is characteristic of juvenile cod

| D ISCUSS I ON
There are likely multiple reasons that we observed higher frequencies of cod in dogfish stomachs in these fishery-dependent collections compared to our reference frequencies observed in the fishery-independent bottom trawl survey. First, we anticipated that some cod could be small, well-digested, or both, which would cause them to be overlooked from macroscopic identification used on the bottom trawl survey. We report a single instance of this occurring in the samples we examined here. Second, trace amounts of cod DNA may linger not just in a well-digested mass in the stomach, but in the stomach chyme or mucus, or possibly in the stomach of a prey item eaten by the dogfish, any of which may be detected by the sensitive real-time PCR assay employed in this study. These possibilities increase the period of detectability further, which increases the detection rate per feeding event.
There may be stage-specific or species-specific digestion rates.
Determining an actual feeding rate requires knowledge of the persistence of DNA in dogfish stomachs at a range of temperatures. A third reason that could lead to higher rates of detection in our study is because fishermen report interactions between both species during operations, and while some of these interactions may be active foraging of live prey, some are not. Rafferty, Brazer, and Reina (2012) observed that 3% of cod landed from Georges Bank gillnet fishery was discarded due to spiny dogfish depredation. From the current study, 50% of the sampling trips with cod positively eaten also included observations of fish entrails and potentially discarded fish; thus, scavenging by spiny dogfish was actively occurring, as has been suspected (Hanchet, 1991;Smith, Ford, & Link, 2016), and should be accounted for in future studies.
This behavior and the knowledge of dogfish sampling co-occurring with cod fishing for a subset of the fishing trips used by this study increased our likelihood of positively detecting cod, as anticipated, but it also confounded our ability to expand our results to population-level estimates of cod mortality.
Fishermen see dogfish associated with cod on their fishing hotspots, during a period of rebuilding dogfish and depleted cod.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data are fully accessible on Dryad (https://doi.org/10.5061/ dryad.8w9gh x3jv).