Using quantitative real-time PCR to detect salmonid prey in scats of grey Halichoerus grypus and harbour Phoca vitulina seals in Scotland – an experimental and field study


*Correspondence author. E-mail:


  • 1There is considerable debate over the impact of seal predation on salmonid populations in both the Atlantic and Pacific oceans. Conventional hard-part analysis of scats has suggested that salmonids represent a minor component of the diet of grey seals (Halichoerus grypus) and harbour seals (Phoca vitulina) in the UK. However, it is unclear whether this is an accurate reflection of the diet or due to methodological problems. To investigate this issue, we applied quantitative PCR (qPCR) to examine the presence of salmonids in the diet of seals in the Moray Firth, UK, during the summers of 2003 and 2005.
  • 2Two qPCR assays were designed to detect Atlantic salmon Salmo salar and sea trout Salmo trutta DNA in field samples and experimentally spiked scats. The proportion of scats sampled in the field that were positive for salmonid DNA was low (ª10%). However, the DNA technique consistently resulted in more positive scats than when hard-part analysis was used.
  • 3An experimental study using spiked scat material revealed a highly significant negative relationship between Ct values obtained from the Atlantic salmon qPCR assay and the proportion of Atlantic salmon material added to scats. The Ct value denotes the cycle number at which the increasing fluorescence signal of target DNA crosses a threshold value. Ct values from field-collected seal scats suggested they contained a very low concentration of salmonid remains (1–5%) based on an approximate calibration curve constructed from the experimental data.
  • 4Synthesis and applications. The qPCR assay approach was shown to be highly efficient and consistent in detection of salmonids from seal scats, and to be more sensitive than conventional hard-parts analysis. Nevertheless, our results confirm previous studies indicating that salmonids are not common prey for seals in these Scottish estuaries. These studies support current management practice, which focuses on control of the small number of seals that move into key salmonid rivers, rather than targeting the larger groups of animals that haul-out in nearby estuaries.


There is considerable debate over the potential impact of seal predation on salmonid populations in both the Atlantic and Pacific oceans (Browne et al. 2002; Middlemas, Armstrong & Thompson 2003), many of which are in decline. Conventional analysis of seal scats has generally reported Atlantic salmon, Salmo salar and sea trout, Salmo trutta to be absent from, or to constitute a minor component of, the diets of both grey seals and harbour seals (Halichoerus grypus and Phoca vitulina) in the UK (Pierce et al. 1991b; Hammond et al. 1994a, 1994b; Thompson et al. 1996; Middlemas et al. 2006). It is unclear whether this is due to salmonids forming a small part of the diet, or the result of salmonid otoliths being under-represented compared with those of other species (Boyle, Pierce & Diack 1990). An additional problem occurs where it is difficult to distinguish between closely related salmonid species once the hard parts have been subject to digestion (Kvitrud et al. 2005; Middlemas et al. 2006).

To overcome some of the problems with these previous studies, the use of molecular techniques to identify salmonid prey in seal scats has been explored. Such methods have mostly involved conventional PCR amplification of cytochrome oxidase III, cytochrome b or mitochondrial 16S ribosomal RNA genes, followed by either restriction fragment length polymorphism (RFLP) (Purcell et al. 2004; Parsons et al. 2005) or denaturing gradient gel electrophoresis (Deagle et al. 2005) to identify different prey species represented.

DNA analyses employing non-invasive sampling are now widely used in molecular ecology, population genetics and conservation biology. In marine mammal research, molecular scatology is applied to reveal sex, individual identity and genetic population structure (Tikel, Blair & Marsh 1996; Reed et al. 1997; Parsons et al. 1999; Parsons 2001). In recent years, DNA-based techniques have also proved to be a valuable tool in studying predator/prey interactions in a number of species (see review by Symondson 2002). DNA extracted from stomach contents and scats has been used to identify both predators and prey species where this has not previously been possible using conventional techniques (Hansen & Jacobsen 1999; Jarman et al. 2002; Nyström et al. 2006). In pinnipeds, a molecular approach to identifying diet components in a predator/prey system offers a complementary and/or alternative approach to classic hard-part analysis of otoliths and other skeletal structures (Purcell et al. 2004; Deagle et al. 2005; Parsons et al. 2005). Conventional hard-part analysis is subject to several biases, mainly caused by partial digestion of hard parts of prey species (Pierce & Boyle 1991; Tollit et al. 1997a,b; Bowen 2000; Cottrell & Trites 2002; Tollit et al. 2003) and absence of hard parts in the scats analysed (Bowen 2000; Browne, Laake & DeLong 2002; Orr et al. 2004).

The present study employed quantitative PCR (qPCR) to detect Atlantic salmon and sea trout DNA in seal scats. Spiked scat material was used to assess the applicability and accuracy of this technique. The field study targeted estuarine areas of the Moray Firth (North Sea, UK) that are inhabited by mixed populations of harbour and grey seals known to consume salmonids (Middlemas et al. 2006).

Materials and Methods

design of primer/probe sets for real-time qpcr, their optimization and specificity

All primers and probes were designed using the primer express software from Applied Biosystems (Warrington, UK) and supplied by the same company. A universal salmonid forward primer, species-specific reverse primers and species-specific probes to distinguish Atlantic salmon and sea trout were designed from published cytochrome b gene sequences (accession numbers AF133701 and M64918). Primer sets, Salmonid Cyt B FOR 5′-CGG AGC ATC TTT CTT CTT TAT CTG T-3′ and S. salar REV 5′-ACT CCG ATA TTT CAG GTT TCT TTA TAT AGA-3′ or S. trutta REV 5′-CTC CGA TAT TTC AGG TTT CTT TAT ATA GG-3′, were capable of amplifying a 96- or 95-bp fragment, respectively, of the cytochrome b gene from each species. Sequences of Atlantic salmon and sea trout specific Taqman MGB probes were: S. salar Cyt B Probe 6-FAM-CGA GGA CTT TAT TAT GGT TC-MGB, Salmo trutta Cyt B Probe VIC-CGA GGA CTC TAC TAT GGT TC-MGB.

To provide a positive control for qPCR, an additional combination of universal primers and a probe was designed to amplify a 142-bp fragment of the cytochrome b gene from both harbour seal and grey seal. Primers Seal Cyt B FOR 5′-TCA TCC GTT ATC TTC ACG CAA AT-3′ and Seal Cyt B REV 5′-GCT ATG ACG GTG AAT AAG AGG ATA ATG-3′, and probe Seal Cyt B Taqman MGB Probe 6-FAM-TTC ATC TGC CTA TAC ATG CA-MGB were designed from published cytochrome b sequence (accession numbers X82306 and NC-001602).

The efficiency of target gene detection by each primers/probe combination was tested on a 10-fold dilution series of Atlantic salmon, sea trout or harbour seal DNA. Triplicate reactions were performed at each dilution in order to generate a standard curve. The qPCR reaction contained 2·5 µl genomic DNA (extracted as below), 900 nm each primer, 250 nm probe, 1 × Taqman master mix (Applied Biosystems) and distilled water (Sigma, Gillingham, UK) in a final volume of 25 µl. The concentration of all qPCR reaction reagents was the same for all assays conducted in this study. Cycling conditions consisted of a single cycle to allow UNG (Uracil-N-glycosylase) digestion of previously amplified potential contaminant products (50 °C for 2 min, 95 °C for 10 min) followed by 45 cycles of denaturation and annealing/extension (95 °C for 15 s, 60 °C for 1 min). These standard conditions were employed for all assays conducted in this study.

To test specificity of primers/probe sets, assays with genomic DNA originating from 31 different prey species recorded in the diet of harbour or grey seal (Table 1) were conducted using each primer set and probe. Genomic DNA was extracted from fish tissue using a cetyl trimethyl ammonium bromide (CTAB)/chloroform method (Saghai-Maroof, Soliman & Allard 1984).

Table 1.  Main groups of prey species found in the diet of harbour seal (Phoca vitulina) and grey seal (Halichoerus grypus) (information from Hammond, Hall & Prime 1994a, 1994b; Thompson et al. 1996; Tollit & Thompson 1996)
Common speciesScientific name
SalmonidsAtlantic salmonSalmo salar L.
Sea troutSalmo trutta L.
GadidsCodGadus morhua L.
WhitingMerlangius merlangus (L.)
HaddockMelanogrammus aeglefinus (L.)
PoorcodTrisopterus minutus (L.)
PoutTrisopterus luscus (L.)
Norway poutTrisopterus esmarkii (Nilsson 1855)
SaithePollachius virens (L.)
LingMolva molva (L.)
TuskBrosme brosme (Ascanius 1772)
ClupeidsHerringClupea harengus L.
SpratSprattus sprattus (L.)
PleuronectidsFlounderPlatichthys flesus (L.)
PlaicePleuronectes platessa L.
Common dabLimanda limanda (L.)
TurbotScophthalmus maximus (L.)
BrillScophthalmus rhombus (L.)
Dover soleSolea solea (L.)
Lemon soleMicrostomus kitt (Walbaum 1792)
OthersMackerelScomber scombrus L.
DragonetCallionymus lyra L.
ScadTrachurus trachurus (L.)
CatfishAnarhichas lupus L.
GreaterHyperoplus lanceolatus (Le Sauvage 1824)
ArgentineArgentina sphyraena L.
BullroutMyoxocephalus scorpius (L.)
LumpsuckerCyclopterus lumpus L.
Red gurnardAspitrigla cuculus (L.)
Grey gurnardEutrigla gurnardus (L.)
Conger eelConger conger (L.)
CephalopodsOctopusEledone cirrhosa (Lamarck 1798)
SquidLoligo sp.

A preoptimized internal positive control (IPC) assay (Taqman Exogenous Internal Positive Control Reagents, Applied Biosystems) was used to check for the presence of inhibitors in scat samples and to distinguish true target negatives from PCR inhibition, as described by McBeath et al. (2006).

field collection of seal scats

Scats from harbour and grey seals were collected shortly after low tide from sandbanks in the Cromarty Firth (57°62′ N, 4°35′ W) and Findhorn Bay (57°66′ N, 3°63′ W) (North Sea, UK). Samples were collected in 2003 and 2005 (Table 2) during May and late June and July, when peaks in salmonid remains were expected (Middlemas et al. 2006). Counts of both harbour and grey seals were made prior to scat collection. Where an area contained >90% of one species, the scats were attributed to a single seal species. Scats taken from mixed groups, and areas where seals had moved before counting could take place, were noted as unknown seal species. Each scat was placed in an individual sealed polythene bag to prevent cross-contamination among samples. In 2003, a 2-g subsample of each scat was stored in 100% ethanol (ethanol absolute, Riedel-de Haën, Seelze, Germany) for subsequent DNA extraction, and the remaining faecal material was stored at –20 °C prior to analysis of residual hard parts. Scats were thawed and each was washed through a nest of sieves (5·0, 1·0, 0·5, 0·25 mm) to allow collection of hard parts. To preserve larger amounts of faecal material in 2005, hard parts were removed by physically breaking apart the scat using fine forceps. The remaining scat material was stored at –80 °C for subsequent DNA extraction. Prey species were identified from sagittal otoliths using a published identification guide (Härkönen 1986) and an in-house reference collection. Bones (vertebrae and jaw) were also used to identify the presence of salmonids (Watt, Pierce & Boyle 1997).

Table 2.  Locations and date of collection of seal scats in the Moray Firth (North Sea, UK) used for analyses of hard parts and molecular detection of the Atlantic salmon (Salmo salar) and sea trout (Salmo trutta), and number of scats attributed to different groups (harbour seals or mixed group)
LocationDate of collectionTotal number of scats collectedHarbour seal scatsMixed group scats
Cromarty FirthMay 200328280
July 200346460
Findhorn BayMay 200335350
July 2003522626
Findhorn BayJune 2005909
July 200543043

dna extraction from field seal scat and qpcr

Approximately 350 mg material from each scat subsample collected in 2003 was washed in distilled water prior to DNA extraction (DNAce Spin Stool Kits, Bioline, London, UK). The scats collected in 2005 were thawed to a consistency at which it was possible to mix them thoroughly prior to extraction, to compensate for subsampling error. Three subsamples (≈200 mg) were transferred to a 2-ml microfuge tube and milled using a TissueLyser (Qiagen) at 20 Hz for 2 min in the presence of ASL buffer (QIAamp DNA Stool Mini Kit, Qiagen, Crawley, UK). Genomic DNA was extracted from these samples using a QIAamp DNA Stool Mini kit (Qiagen, Crawley, UK) and quantified by fluorometry (Quant-iT PicoGreen dsDNA Assay Kit, Invitrogen, Paisley, UK).

The qPCR assays to detect seal, salmon and sea trout DNA were performed using the standard protocol described above. In 2003, both 1 : 10 and 1 : 100 dilutions of DNA extracted from faecal samples were used in qPCR. In cases where only one of these dilutions showed amplification in a qPCR, the seal scat was still marked as positive for the target DNA. The IPC assay was run on samples that failed to detect the seal target DNA in both dilutions, as described above. In 2005, 0·5 ng DNA (neat template) was used as template for all the samples to allow comparison among the scats. Three out of 52 samples were excluded from the qPCR analysis due to the extraction of very low concentrations of genomic DNA. Those samples demonstrating inhibition with neat template were reanalysed using a higher dilution as template for qPCR.

The amount of DNA in the logarithmic phase of amplification was used as a relative measure of how much prey DNA was present in the scat sample. Ct values denote the cycle number at which the increasing fluorescence signal of target DNA crosses the threshold set in the logarithmic phase of amplification.

dna analyses of scats from captive-fed seals

Scats of captive grey seals fed exclusively on Atlantic salmon diet or a mixture of herring and sprat diet (non-salmon diet) were collected and stored at –80 °C (Parsons et al. 2005). Scat samples were mixed thoroughly, and spiked experimental material was prepared by combining salmon-exclusive and non-salmon-fed seal scats to obtain material containing 1, 5 and 25% of scat from seal fed exclusively on Atlantic salmon (hereafter referred as scats 1, 5 and 25% positive for Atlantic salmon).

Approximately 200 mg spiked scat was transferred to a 2-ml microfuge tube and milled using a TissueLyser at 20 Hz for 2 min in the presence of ASL buffer. DNA was extracted from the resulting homogenate using the QIAamp DNA Stool Mini Kit. Genomic DNA was quantified using the Quant-iT PicoGreen dsDNA Assay Kit. Approximately 0·5 ng DNA was used as template in the qPCR detection of seal and salmonid target DNA. The qPCR reaction and cycling conditions are described above. The IPC assay was performed on serial dilutions of templates from each spiked experimental group, as described above. Forty-five subsamples from each spiked experimental group were processed.

The proportion of Atlantic salmon in the experimental spiked scats was log10-transformed before testing its relationship with Ct values obtained from the Atlantic salmon and seal qPCR assays using the general linear model (GLM) in r (R Development Core Team 2006).


optimization of primers/probe sets for qpcr

Each assay exhibited high efficiency, with slope values (ΔS) of 3·57, 3·34 and 3·36 for the Atlantic salmon, sea trout and seal assays, respectively. Each reaction series also showed a high correlation between cycle number and dilution factor (R2 > 0·994 for salmon and sea trout; R2 = 1 for seal).

Specificity of the seal primer/probe set was tested against a dilution series of DNA extracted from salmon and sea trout, and no amplification resulted from templates derived from salmonids. The salmon and sea trout primers/probe sets were also specific to their intended target species, and showed no amplification with DNA extracted from seal or any of the other species listed in Table 1.

detection of salmonids from 2003 field seal scat samples

A total of 87 samples from the Findhorn Bay and 74 samples from the Cromarty Firth were processed for the detection of seal, Atlantic salmon and sea trout DNA (Table 2). Scats collected from the Cromarty Firth site were exclusively from harbour seals. The Findhorn Bay sampling site was used by both species of seal, and half of the 52 scats collected in July 2003 were of unknown origin (Table 2). The seal positive control qPCR assay failed in 22 samples from Findhorn Bay (25·2%) and five samples from Cromarty Firth (6·8%) in both 1 : 10 and 1 : 100 dilutions. The IPC assay did not detect the presence of inhibitors in these samples, therefore the qPCR failure may be due to DNA degradation in the analysed scats.

DNA from Atlantic salmon was detected in four scats from Findhorn Bay and five scats from the Cromarty Firth. The sea trout qPCR assay detected target DNA in four scats from the Findhorn Bay and nine scats from the Cromarty Firth (Table 3). Two scats collected in the Cromarty Firth in July 2003 showed the presence of DNA from both salmonid species. 74% of the Atlantic salmon and sea trout positives showed very high Ct values (over 40), suggesting that only a very small amount of the target DNA was present in the scat. It was impossible to estimate Ct values for detection of salmonid target in neat samples due to the low efficiency of the assay in detecting template at such a low concentration. The remaining Atlantic salmon and sea trout positives showed Ct values between 32·11 and 39·41 (in 1 : 10 dilutions) and 35·12–42·80 (in 1 : 100 dilutions). The fact that Ct values were relatively high in all cases suggests a low quantity of prey salmonid DNA in these samples.

Table 3.  Number of scats analysed that showed positive detection of Atlantic salmon (Salmo salar) and sea trout (Salmo trutta) using qPCR
LocationDate of collectionTotal number of scats analysedSalmo salar detectedSalmo trutta detected
Cromarty FirthMay 20032812
July 20034647
Findhorn BayMay 20033520
July 20035224
Findhorn BayJune 2005900
July 20054321
Total 2131115

For all four of the month/site combinations, a higher proportion of scat samples contained salmonid DNA than contained either otoliths or bones (Fig. 1). DNA also provided evidence of salmonids (salmon and trout) in the May 2003 Cromarty Firth samples, which were negative using the other techniques (Fig. 1). The number of scats containing sea trout DNA in the July 2003 Findhorn Bay samples was high enough to test for differences between samples identified as harbour seals compared with those from a mixed group containing harbour and grey seals. A significantly higher percentage of the harbour seal scats tested positive for sea trout DNA compared with those from mixed seal groups (25% vs. 0%, Fisher's exact test, P = 0·031).

Figure 1.

Percentage of scat samples tested that contained evidence of salmonids using three different techniques (otoliths, grey bars; bones, black bars; qPCR detection of prey DNA, white bars) in scats collected in 2003. Sample sizes are given in Table 2.

Differences in the diet of harbour seals have been shown to be linked to foraging habitat (Tollit et al. 1998), and sample sizes allowed the occurrence of sea trout to be investigated in relation to that of sandeel and flatfish otoliths in the combined data set (all months/sites; Table 2). Sandeel otoliths were found in significantly fewer scats that tested positive for sea trout DNA than in those that tested negative (9 vs. 51%, Fisher's exact test, P = 0·010); in contrast, there was no difference in the occurrence of flatfish otoliths between scats with sea trout DNA and without (64 vs. 60%, Fisher's exact test, P = 1). These results suggest that the occurrence of sea trout in scats is associated more with the occurrence of flatfish than of sandeels (Fig. 2).

Figure 2.

Percentage of scats collected in 2003 containing sandeel and flatfish otoliths broken down into those testing positive for the presence of sea trout DNA (grey bars) or those testing negative (white bars). Note that all sites and months have been combined (n = 101).

dna analyses of scats from captive-fed seals

To relate Ct values from qPCR to the proportion of target DNA in the whole scat sample, an experiment was carried out with spiked seal scats. Forty-five subsamples from each group of salmon-spiked experimental seal scat (scats 100, 25, 5 and 1% positive for Atlantic salmon) were processed for seal and salmon detection by qPCR. All IPC assays indicated no inhibition of qPCR using neat template derived from subsamples from the spiked scat material. The seal positive control qPCR assay failed in only one subsample of 1% salmon-positive spiked scat. There was no significant relationship between Ct values from the seal assay qPCR and the proportion of Atlantic salmon in each spiked experimental group (GLM, F1,176 = 2·21, P = 0·139; Fig. 3a).

Figure 3.

Relationship between proportion of Atlantic salmon in experimental spiked scats and Ct values obtained from seal (a) and Atlantic salmon qPCR (b) assays. Ct values denote number of cycle at which an increasing fluorescence signal of target DNA crosses the threshold set in the logarithmic phase of amplification.

The salmon assay detected target DNA in all subsamples of experimental scat 100, 25, 5 and 1% positive for Atlantic salmon. There was a highly significant negative relationship between Ct values from the salmon assay qPCR and the proportion of Atlantic salmon in the experimental spiked scats (GLM, F1,176 = 2713, P < 0·001; Fig. 3b). Serial dilutions were prepared for 10 subsamples of each spiked experimental group, and these data indicated that the salmon assay was efficient enough to detect even a very low concentration of salmon DNA. For example, in the 1 : 10 dilution of template 1% positive for Atlantic salmon, seven out of 10 subsamples showed amplification of target DNA, indicating a final concentration of Atlantic salmon as 0·1%. In 1 : 100 dilution of the same template, one out of 10 subsamples showed amplification of target DNA, showing the possibility of detecting Atlantic salmon in template containing only 0·01% of this prey species.

detection and relative quantification of salmonids from 2005 field seal scat samples

Three subsamples from each of the 49 samples from Findhorn Bay were processed for detection of Atlantic salmon and sea trout DNA. The number of subsamples per scat was increased compared with the 2003 analysis because only a limited number of scats were found positive for salmonid remains. This might have led to underestimation of prey content as the homogenization of scat material was not performed prior to DNA extraction. Scats from the 2005 collection were also marked as mixed samples, originating from a mixed harbour and grey seal population. Inhibition was detected in neat template of 14 scat samples. Seven of these samples showed inhibition in all three subsamples; the rest of the samples showed inhibition in only one of three subsamples. A 1 : 10 dilution of template was used for detection of Atlantic salmon and sea trout for all 14 inhibited scat samples. The seal positive control qPCR failed in three scat samples. All three subsamples were negative in one scat, and two scats showed no amplification of seal DNA target in one subsample.

Only two of the 49 Findhorn Bay scat samples tested positive for Atlantic salmon DNA, with all three subsamples of these being positive (Table 3). The qPCR analysis was performed for another 10 subsamples from each positive scat. Atlantic salmon DNA was detected in all 10 and nine from 10 subsamples, respectively. Ct values from individual subsamples within the positive scat were variable, ranging from 38·03–41·71 and 34·71–44·87. Comparing these values with values obtained from the spiked scat experiment, the concentration of Atlantic salmon in positive seal scat, which had been homogenized prior to subsampling, was estimated to be around 1 and 1–5%, respectively. The sea trout assay resulted in only one scat sample positive for the target DNA out of 49 scats tested (Table 3); however, only one of three subsamples was positive, with a high Ct value of 43·18. When compared with the Ct values from the spiked scat experiment, the concentration of sea trout in field scat samples was estimated to be around 1%. No remains of salmonid hard parts were found in any of the 2005 samples.


Our study reports the first application of qPCR to detect marine mammal prey in scats, and reveals data previously unobtainable using conventional PCR (Purcell et al. 2004; Deagle et al. 2005; Parsons et al. 2005). Additionally, qPCR provides a more sensitive tool than conventional hard-part analysis, consistently detecting more scats containing salmonids than the traditional analyses. The Taqman assays presented here demonstrated high specificity in the detection of target DNA. The seal assay showed no amplification from salmonid-derived material, and both Atlantic salmon and sea trout assays were unable to amplify DNA extracted from 29 different fish species recorded in the diet of harbour seal and grey seal (Hammond et al. 1994a, 1994b; Thompson et al. 1996; Tollit & Thompson 1996).

In the 2003 field scat samples, high levels of PCR inhibitors hindered detection of prey DNA in the neat template in most samples. A high level of DNA degradation, shown by the failure to detect the positive control seal DNA in 17% of scat samples from 2003, might indicate that ethanol is not the best preservative for scat samples. Storage of scats at –80 °C was shown to be a better method for storage and preservation, as the presence of PCR inhibitors in the neat template of 2005 was reduced to ≈20% of scats, and detection of the control seal DNA failed in only 6% of scats. Also, the seal scat sampling protocol was shown to affect results of the qPCR assays, and might in some cases lead to underestimation of the proportion of salmonids consumed by seals. As DNA from prey species is known to be distributed unevenly within a scat, a thorough homogenization, followed by subsampling (minimum of three subsamples per scat) of scats after removal of the hard parts, is recommended to prevent underestimation of consumed prey.

The experimental study was designed to assess the potential use and accuracy of the qPCR assay to detect salmonids from seal scats, and these data were used to estimate salmonid remains in scats obtained from the Moray Firth area in 2005. The experimental trial on spiked seal scats confirmed high efficiency in detecting seal and Atlantic salmon DNA in scat material. Based on the Ct values for detection of seal DNA, similar quantities of seal DNA were present in each of the spiked scat experimental groups. This also confirms that mixing scats prior to analysis is a useful approach to avoid subsampling error, as previously stated by Deagle et al. (2005). The seal DNA in scat samples originates from the cells of the rectal lining, and is expected to be localized mainly on the surface of scats. The fact that no significant differences were found in Ct values for seal detection among the spiked scat experimental groups indicates a uniform distribution of seal DNA, and most likely prey DNA, in mixed scats. The Atlantic salmon qPCR assay performed using the spiked scat experimental material revealed high consistency in detection of the target DNA, even at concentrations as low as 5 and 1% of Atlantic salmon-fed scat in the entire scat. All 45 subsamples in each spiked experimental group showed amplification of target DNA, suggesting a uniform distribution of prey DNA in mixed spiked scats. Performing the qPCR assay on a dilution series revealed that it is possible to detect as little as 0·01% Atlantic salmon in seal scats using the qPCR assay designed in this study.

Based on Ct values of Atlantic salmon from the experimental study, an approximate calibration curve could be constructed and the concentration of Atlantic salmon in field scat samples estimated. However, estimating the quantity of prey in field scats based on this is only relative. This is because no information is available on the rate of DNA degradation for different prey species during digestion, or degradation of prey DNA in scats in the environment prior to collection. Furthermore, seals may not consume whole specimens of fish, but may only eat soft tissue (Pierce & Boyle 1991; Orr et al. 2004). The limitations of this approach mean it is impossible to accurately relate concentration of prey DNA to the number of prey specimens consumed by an individual seal. However, future application of the Taqman assays on scat material derived from a controlled seal-feeding trial might help to reduce these problems and enable more accurate quantification of prey species in seal scats.

Analysis of seal scats collected from the field in 2005 revealed that only two samples out of 49 tested positive for Atlantic salmon DNA, and one sample tested positive for sea trout. Based on the calibration curve from the experimental study, the predicted concentration of the salmonid target in whole scat samples was about 1–5 and 1% for Atlantic salmon and sea trout, respectively. This suggests that these types of prey were not common in the samples. If salmonids represent only an infrequent or ‘pulse’ component of diet, detection of target DNA using conventional PCR is restricted to scats produced within 48 h of ingestion (Deagle et al. 2005). However, the qPCR technique has been shown to detect target DNA, even in cases where conventional PCR failed (Snow et al. 2006), so using the qPCR approach a ‘pulse’ component of diet might be detectable for longer than 2 days after ingestion. A higher number of scats that were positive for salmonids was found in the 2003 sampling season. However, it was difficult to relate Ct values from these samples to values from the spiked scat experiment, as a different amount of template was used in the qPCR reactions due to the presence of PCR inhibitors. Also, only 2-g subsamples of scats were taken to be analysed in 2003, so the calibration curve from the spiked scat experiment cannot be used in this case.

A parallel conventional analysis of hard parts was performed on the scat samples collected in both years. The occurrence of salmonid hard parts in the scat samples showed a pattern similar to that of the genetic analysis. There were, however, a number of important differences, notably the ability to differentiate between Atlantic salmon and sea trout. Additionally, salmonid DNA was detected in a greater number of scats than hard parts, including positive results from areas and months where otoliths and bones were absent (Cromarty Firth, May 2003; Findhorn Bay, 2005). This shows that analysis of DNA is better for detecting rare prey species than the conventional analysis of hard parts.

The results illustrate the potential use of DNA alongside conventional analyses in providing insights into the feeding ecology of seals. For example, qPCR analyses support previous hard-part analyses and confirm that salmonids are relatively scarce prey items at these study sites. The results suggest that sea trout were more prevalent in the diet of harbour seals compared with a mixed group containing both grey and harbour seals, although definite identification of seal species would be required to confirm this finding. This result is in agreement with previous diet work (Hammond et al. 1994a, 1994b; Tollit & Thompson 1996; Middlemas et al. 2006) and observations showing that harbour seals are more common than grey seals in rivers (Carter et al. 2001; Middlemas et al. 2006). In addition, the results of this study suggest that predation on sea trout is linked to predation on flatfish rather than sandeels. The occurrence of these prey groups in the diets of harbour seals in the Moray Firth region has been shown to be linked to the seals’ use of different water depths and sea bed sediments (Tollit et al. 1998). This result therefore potentially provides information on where seals are consuming sea trout (Tollit et al. 1998; Bowen et al. 2002).

Just as the classical hard-part approach involves a bias in quantification of prey in seal scats (Pierce & Boyle 1991; Tollit et al. 1997a,b; Bowen 2000; Browne et al. 2002; Arim & Naya 2003; Orr et al. 2004), quantification based on DNA techniques is complex. Deagle et al. (2005) designed a captive feeding trial with Steller sea lions, Eumetopias jubatus, to assess the reliability of conventional PCR amplification of prey DNA from seal scats, and also tried to estimate diet composition by quantifying the amount of DNA present in scats, using library screening. They concluded that DNA techniques are accurate; however, they uncovered a possible bias when estimating quantity of prey, especially when unequal meals have been ingested during a day, and suggested pooling DNA extracted from scats with patchy diet to ensure a more accurate proportional estimate.

For field studies, the results derived from the controlled trial by Deagle et al. (2005) suggest that multiple scat samples from individual seals would provide better insight into feeding habits. Linking scats to individual seals may be possible using microsatellite analysis, as described by Reed et al. (1997). This would allow identification of specialist seals which is of importance for management of predator–prey interactions (Graham et al. 2005). Moreover, in other predator/prey systems, differences between prey consumption of an individual predator and that of the whole population were highlighted, reflecting differences in fitness and reproductive success within the predator population, or their social dominance (Fedriani & Kohn 2001; Gende & Quinn 2003).

In conclusion, an important step in the fisheries management of seal–salmonid interaction in the Moray Firth is to determine accurately the occurrence of salmonids in the seal diet (Middlemas et al. 2003). Using conventional techniques, it is not possible to determine unequivocally whether a low occurrence of salmonid remains is due to their scarcity in the diet or the fragility of their hard parts compared with those of other prey species (Boyle et al. 1990). Our DNA-based approach allows the relative contribution of Atlantic salmon and sea trout in the seal diet to be determined. With only approximately 10% of scats testing positive for salmonid DNA, our DNA analysis supports previous suggestions that salmonids are rare in the diet of seals in the Scottish estuaries (Pierce et al. 1991a, 1991b; Tollit & Thompson 1996; Middlemas et al. 2006). Consequently, these results support current management practice, which focuses on control of the small number of seals that move into key salmonid rivers, rather than targeting the larger groups of animals that haul-out in nearby estuaries (Butler et al., in press).


The authors would like to thank Sandy Mitchell, Leanne Reid and Jenny Gaube for help during laboratory work, and staff of the Sea Mammal Research Unit at the University of St Andrews for providing the salmon-fed seal scats for the experimental study.