DNA barcoding unveils skate (Chondrichthyes: Rajidae) species diversity in ‘ray’ products sold across Ireland and the UK

Sample collection

Samples labelled as ray or skate wings were initially purchased from fishmongers or fish counters within supermarkets and takeaway ‘fish and chips’ shops from across Dublin (ROI). The retailers were also visited in two distinct periods; January–February and October–November 2010. A total of 58 wings from 10 fish mongers/counters and 9 takeaways were analysed (Table 1 and Supplemental Information). In order to provide a wider view of patterns of skate consumption, 40 additional fresh ray or skate wings were also collected between October 2012 and March 2013 from around the UK; sampling was focused in the south-west of England (i.e., counties of Devon and Cornwall), north-west England (largely in Greater Manchester and Liverpool), Glasgow (Scotland), and Cardiff (Wales). UK samples were limited to ‘fresh’ products as investigations of takeaways in the UK cities failed to identify any that sold skate. Repeat sampling from any single shop was kept to a minimum, although it was necessary in some cases to increase the sample size (but efforts were generally made to sample on different dates or from different sized wings; Supplemental Information 1). Therefore, between January 2010 and March 2013 a total of 98 wings were collected from 49 different retailers across the ROI and UK.

Molecular analysis

Tissue samples were collected following Miller, Jessel & Mariani (2012) and DNA extractions were completed according to a modified chloroform extraction procedure (Petit, Excoffier & Mayer, 1999). DNA barcoding (specifically of COI gene) was selected as the species identification tool (Hebert et al., 2003). Approximately 648 base pairs (bp) of the COI region were amplified by polymerase chain reaction (PCR) using primers designed by Ward et al. (2005): FishF2 (5′-TCGACTAATCATAAAGATATCGGCAC-3′) and FishR2 (5′-ACTTCAGGGTGACCGAAGAATCAGAA-3′). Reaction conditions following Serra-Pereira et al. (2010) were utilized for all purchased skate samples. The PCR products were sequenced by Macrogen Inc, Seoul, and Source Bioscience, Nottingham, and the results were checked by eye in BIOEDIT version 7.1.11 (Hall, 1999) for errors and ambiguous base calls. The sequences were then input into the BOLD system using the species level barcode database to identify the species of each sample and then cross-referenced using BLAST on GenBank (Basic Local Alignment Search Tool, National Centre for Biotechnology Information, Bethesda, Maryland; www.ncbi.nlm.nih.gov/). A 98% similarity criterion was used as a threshold above which identification of unknown samples was deemed reliable (as analysis of barcode records suggests at a divergence of greater than 2% it is very likely that sequences belong to different species, Ward, 2009). Any additional matches within this 98% criterion are also reported in the Supplemental Information. However, in the case of the BOLD system, only additional matches within the Public Record Barcode database are described. This should help ensure that the primary identification is made with a comprehensive database that maximises the chance of finding a strong match, but any additional matches are only made with the more closely regulated dataset, helping to reduce issues associated with the misidentification of submitted records.

Data analysis

Principal component analysis (PCA) of the data was conducted in PRIMER-6 (Clarke & Warwick, 2001), with each retailer representing an individual data point in the ordination. The software was also used to conduct a non-parametric analysis of similarity (ANOSIM) with 999 permutations, and to calculate the contribution of variables to similarity (or similarity percentages; SIMPER). Both utilised the Bray-Curtis similarity measure during calculation. Otherwise, the program defaults were utilised.

DNA barcoding

Good quality sequences were obtained for all 98 skate samples, these varied in length between 494 and 646 bp (average length 617.9 bp). Searches on BOLD and GenBank generally produced clear matches allowing for confident assignment of species and there was good agreement between databases. In fact, all the searches yielded matches that were within 98% similarity to database records, with one exception (Supplemental Information 1). In this case a 100% match on BOLD was made, but no results within the 98% similarity criterion were located in GenBank, so the BOLD identification was utilised. Additionally, in 24 cases a second close match (also with less than 2% divergence to the unknown samples) was also identified. These additional matches originated from closely related sister taxa, but in every case the species were not distributed around the British Isles (they included: Raja maderensis that is restricted to Madeira and the Azores, Raja polystigma that is restricted to the Mediterranean, Raja straeleni and Leucoraja walacei that both have a primarily South African distribution). Therefore, the additional matches were deemed unlikely to be the true source of the samples and the species with the highest degree of similarity from the BOLD species level barcode database was upheld for the identification.

Across the entire study, six species of skate were identified in products sold for consumption (Table 2); 31 blonde rays (Raja brachyura), 28 spotted rays (Raja montagui), 19 thornback rays (Raja clavata), 14 cuckoo rays (Leucoraja naevus), five small-eyed rays (Raja microocellata) and one shagreen ray (Leucoraja fullonica).

Statistical support

The PCA failed to produce highly distinct clusters based on retail type or region, although it does broadly illustrate some important patterns that are evident within the data (Fig. 1, because tests detailed below failed to identify significant differences between samples collected in different seasons, the samples from individual retailers were combined across months). Interpretation of the PCA axis and eigenvectors (Fig. 1 and Supplemental Information 2) suggests blonde rays are the main constituent in sales from takeaways, whilst spotted rays are particularly important in fishmongers, leading to a degree of separation of retailer type in PC1. The PCA also illustrates how cuckoo, shagreen and small-eyed rays are largely absent in samples analysed from the ROI, occurring much more commonly in the analysis of UK ray wings.

The result of the ANOSIM when testing for a difference between the temporally replicated samples collected in January/February and October/November 2010 from Dublin was non-significant (R = 0.073, p = 0.052). As the result was close to the 95% confidence interval, tests were also made between the temporal replicates of the fishmongers (R = 0.091, p = 0.130) and takeaways (R = 0.017, p = 0.298) separately, which were also non-significant. In contrast, the comparison between Dublin fishmongers and takeaways was highly significant (R = 0.182, p = 0.001), with the SIMPER results demonstrating average dissimilarity of 69.89% between samples (with the highest contributions originating from the blonde and spotted rays, each contributing about a third of the total dissimilarity). Comparison of fishmongers in the ROI and the UK was non-significant (R = −0.12, p = 0.995).

Conservation issues

Across the study, no species with a vulnerable status on the redlist (IUCN, 2012), or an EU zero TAC, was identified. Given the severe declines in abundance of these skate species, this may simply reflect the relatively modest sample collection assembled here, but considerable efforts across the British Isles have been made to emphasise the conservation status and TAC limits placed on many skates (e.g., Seafish, 2013). Therefore, the complete absence of species of critical conservation concern may also reflect effective controls by the EU and efforts of fishermen and stakeholders to reduce landings of these groups. Perhaps a more stringent test of such action could be made in regions where threatened species are more locally abundant: e.g., R. undulata remains relatively common around the Channel Isles (Ellis, McCully & Brown, 2012), and analyses focused on this area could form a more rigorous test of effective landing controls. Interestingly, the largest growing species identified (R. brachyura) also occurred most frequently in the study. Maybe more awareness should focus on R. brachyura as size has been suggested as a good proxy for vulnerability to overexploitation (Dulvy & Reynolds, 2002). R. brachyura, R. clavata and L. fullonica that were identified in the purchased wings are already included under the ‘near-threatened’ category in the IUCN redlist. R. brachyura is also listed under the red category in the Marine Conservation Society’s good fish guide (MCS, 2013), which suggests it should not be so extensively consumed. Perhaps guidelines on the maximum size of skate wings sold would increase consumer confidence that the products originate from sustainable sources and reinforce action to protect some of the vulnerable large growing species. This would certainly simplify consumer choice, but could potentially overlook trading or harvesting of juveniles from the large growing species that could go undetected due to their small size.

Variation in patterns of sale

This study represents a relatively intensive effort to collect samples from ray wings across multiple time-points, regions and countries, and goes beyond many previous investigations of seafood and mislabelling that focused on a single city, port or landing. Notwithstanding these efforts, it has resulted in a relatively modest sample collection. In part this reflects the fact that ray wings were not commonly available in some areas (particularly within the UK, where many retailers were visited that did not stock them), but it also reflects a desire to avoid high levels of repeat sampling from any one location or supermarket chain. It is hoped this approach gives the samples analysed a high degree of independence (not just repeatedly sampling from the same supply chains), producing a truly comprehensive examination of the sale of skates. Therefore despite the modest sample size, this study represents the first occasion that an investigation of the patterns of sale of ray wings can be made between retailer types, seasons and regions.

A highly significant difference was identified in species sold between fishmongers and takeaways within Dublin. The additional level of processing in takeaways provides greater opportunity for large wings originating from larger species to be divided before sale. Conversely, wings from smaller growing species may form more attractive sizes for sale in fishmongers. In recent years a number of supermarkets have also made a move towards selling wings from smaller skate species (Waitrose, 2009), marketing them as sustainably sourced ‘ray’ wings. As 28 of the ray wings sampled across the entire study in the fishmonger group originated from supermarket fish counters, this may help explain the difference in species sold. It is also interesting to note that whilst all the samples analysed were labelled as “ray” or “skate” (and all the species identified belonged to the Rajidae family, complying with broad expectations of the labels), one supermarket chain went further and specified its wings originated from three smaller growing species. In the six cases where wings were analysed from this chain two belonged to R. clavata that was not included on the label, therefore, demonstrating a 33% level of mislabelling within this small subset of samples. Overall, the sale of skates under ambiguous names remains a challenge to the consumers’ ability to exercise a right to choose those from less vulnerable groups.

A somewhat surprising result from the data analysis was the failure to identify any significant difference in the patterns of species sold in the January/February and October/November sample collections from Dublin. Many elasmobranch species are highly migratory, particularly with reference to specific nursery areas, with marked seasonal movements between the North Sea and Thames estuary for R. clavata having already been observed (Hunter et al., 2005). Therefore, it may be expected that such seasonal movements would affect capture vulnerability in commercial fisheries, and hence patterns of availability to consumers. One explanation for this result is that retailers will freeze stocks of fish to ensure a steady supply. In the case of skates, previous work suggests this is not common, even within takeaway shops, as the wings do not freeze well so supply was dependent on local fisheries (Miller & Mariani, 2013). However, much remains unknown about the exact provenance of samples sold, behaviour of fisherman and even the movements of skate species (Ellis et al., 2005), such that interpreting patterns of sale of skates remains difficult.

The same problems for interpretation occur when comparing sales of skate species in the different countries (ROI and UK), which were also non-significant. Perhaps this reflects the fact that the sale of ray wings in both countries exploits similar species and stocks of skates. Although in this case the comparison is also confounded by the fact that sample collection was conducted approximately two years apart. The only other study to utilise genetic methods of species identification on skate products was recently conducted by Lago, Vietites & Espineira (2012) in Spain. The 10 samples of whole or winged fresh skate analysed there originated not only from species found around Spain, but included groups restricted to other areas of Europe and even South America. This suggests patterns of sale could vary dramatically between countries or regions and they will not simply reflect levels of abundance in local fisheries.