Trace element fingerprinting of cockle (Cerastoderma edule) shells can reveal harvesting location in adjacent areas

Determining seafood geographic origin is critical for controlling its quality and safeguarding the interest of consumers. Here, we use trace element fingerprinting (TEF) of bivalve shells to discriminate the geographic origin of specimens. Barium (Ba), manganese (Mn), magnesium (Mg), strontium (Sr) and lead (Pb) were quantified in cockle shells (Cerastoderma edule) captured with two fishing methods (by hand and by hand-raking) and from five adjacent fishing locations within an estuarine system (Ria de Aveiro, Portugal). Results suggest no differences in TEF of cockle shells captured by hand or by hand-raking, thus confirming that metal rakes do not act as a potential source of metal contamination that could somehow bias TEF results. In contrast, significant differences were recorded among locations for all trace elements analysed. A Canonical Analysis of Principal Coordinates (CAP) revealed that 92% of the samples could be successfully classified according to their fishing location using TEF. We show that TEF can be an accurate, fast and reliable method to determine the geographic origin of bivalves, even among locations separated less than 1 km apart within the same estuarine system. Nonetheless, follow up studies are needed to determine if TEF can reliably discriminate between bivalves originating from different ecosystems.

shall be traceable at all stages of production, processing and distribution, from catching or harvesting to retail stage". More recently, the European regulation (EC) No 1379/2013 "on the common organization of the markets in fishery and aquaculture products" further contributes to the implementation of seafood traceability and requires that the category of fishing gear or production method (i.e. caught or farmed) is provided together with geographic detail of the catch area. However, this information is not always available to end consumers and is prone to fraudulent use (e.g. mislabelling of place of origin). Consequently, even conscientious buyers aware of the potential hazards associated with the consumption of bivalves may not be able to securely purchase this highly-prized seafood. It is critical to develop and validate reliable techniques that allow competent authorities to trace the origin of traded bivalves to ultimately fight fraud and prevent notable risks to public health 10 .
The taxonomic identification of bivalves and the traceability of their fishing location are often difficult to achieve, as bivalves are commonly processed after collection (e.g. precooked, canned). Most studies have relied on molecular techniques, including PCR 11,12 , FINS 13,14 and DNA barcoding 15,16 , for species identification. Molecular tools, particularly microsatellites 17,18 , as well as biochemical methods, such as fatty acids [19][20][21] and stable isotopes [21][22][23] , have also been used to assess geographical origin of bivalves. Trace element fingerprinting (TEF) of bivalve mineral structures may also be useful to distinguish populations or stocks [24][25][26] . Trace elements are influenced by environmental features of each ecosystem 27 and are recorded in hard structures (e.g. shells, statolith and otoliths) of marine organisms 28 . Several trace elements are found in a wide range of marine species 29 , with the most common being aluminium (Al), barium (Ba), calcium (Ca), cobalt (Co), chromium (Cr), copper (Cu), magnesium (Mg), manganese (Mn), lead (Pb), zinc (Zn) strontium (Sr) and uranium (U). TEF of sea snails larvae 28 and shells from bivalve larvae 30 and adults 26 have been successfully used to distinguish specimens from geographically close populations (20-50 km). However, it remains unclear whether this geochemical approach has enough resolution to discriminate specimens from adjacent areas (<1 km apart) within the same aquatic system.
The present study aimed to validate TEF of shells from fresh bivalves as a proxy to discriminate the origin of specimens collected from adjacent areas of the same estuarine system. It is important to highlight, that unlike previous studies on TEF that use laser ablation of a small part of the larval or early juvenile shells of bivalves 28,30 , the present study uses the whole shell of adult bivalves. The rationale for using this approach was to somehow minimize the temporal variability of TEF in the shells of adult specimens. We used cockle (Cerastoderma edule) as a model species due to its economic importance as a fishery resource 31 , with the coastal lagoon Ria de Aveiro (Portugal) being selected as the collection site due to its diverse tidal system and important role in Portuguese bivalve fisheries 31 . Once cockles are usually fished by hand or hand-raking, this study also aimed to test if the use of metal rakes could induce some type of metal contamination and be a source of bias for TEF. The following hypotheses were tested i) TEF of C. edule shell does not differ with fishing method (i.e. hand-raking vs. by hand), and ii) TEF of C. edule shell is similar among different locations within the same coastal lagoon.

Material and Methods
Study area and cockle collection. Cerastoderma edule with a shell length > 25 mm (i.e. commercial size) (likely displaying an age of 3+ years; the species lifespan may be up to 6 years 32 ) were collected during June 2013 in five different locations of Ria de Aveiro distributed among Mira (M1 and M2), Espinheiro (E1 and E2) and Ílhavo (I) Channels (Fig. 1). All locations play an important role on the fishery of C. edule in Ria de Aveiro, which usually exceeds 1000 tons per year in this region 31 . Two fishing methods were used to collect twenty specimens of C. edule at M1: ten by hand-raking and ten by hand (n = 10 *2). Subsequently, ten specimens were collected by hand on the other locations: M2, E1, E2 and I (Fig. 1). All samples were stored in aseptic bags kept refrigerated during sampling and brought to the laboratory and frozen at − 20 °C for later processing.
Shell preparation. Volumetric polyethylene material and micropipettes with plastic tips were used to prepare collected shells for trace elements analysis 33 . Plastic bottles, ceramic coated blades and tweezers kept in 2-5% solution of DECON 90 over 2 h were washed with running water, immersed in 10% of HNO 3 for 24 h, washed with Milli -Q (Millipore) water and dried in a laminar flow hood. The preparation for ICP-MS analysis was performed in a class 100 (ISO class 5) clean room. The valves were separated and the organic tissues were removed using ceramic coated blades and tweezers. The right valve was transferred to a previously acid-washed plastic bottle and the left valve discarded.
Samples were soaked in 20 mL high-purity H 2 O 2 (30% w/v) (AnalaR NORMAPUR, VWR Scientific Products) overnight (14-16 h) to remove organic matter from the shell including the periostracum. After organic matter removal, the valve was rinsed in Milli -Q (Millipore) water three times. Digestion of entire valves was performed with addition of 20 mL of high-purity concentrated (70% w/v) HNO 3 (Trace metals; Sigma-Aldrich). To avoid having Ca masking the concentrations of the remaining elements 34,35 , the resulting solution was diluted with Milli -Q (Millipore) water to a final acid concentration of 2% HNO 3 .   55 Mn and 66 Zn were determined through inductively coupled plasma mass spectrometry (ICP-MS), on a Thermo ICP-MS X-Series equipped with a auto sampler CETAX ASX-510, Peltier Nebulizing Camera Burgener nebulizer, nickel cones and the CeO+ /Ce+ ratio was optimized at < 2%. The concentrations of 48 Ca, 24 Mg and 88 Sr were determined by inductively coupled plasma optical emission spectrometry (ICP-OES) on a ICP-OES Jobin Yvon Activa M equipped with auto sampler JY-AS500 and Burgener Mira Mist nebulizer.

Statistical analysis.
Concentrations of trace elements of the shells were standardized to Ca and all data analyses were carried out on the element ratios (X: 48 Ca) 25,30,36 . To assess if fishing method significantly affected TEF, a resemblance matrix based on the normalized Euclidean distance was calculated 37 for a one-way analysis of similarity (ANOSIM) 38 , which calculates a global R statistic that assesses the differences in variability between groups when compared to within groups and checks for the significance of R using permutation tests 38 . Differences among fishing locations for each elemental ratio were assessed using a one-way analysis of variance (ANOVA), and Tukey's HSD pairwise comparisons when significant differences were observed (p < 0.05). Similarity percentages (SIMPER) were calculated to quantify the contribution of each trace element to the dissimilarities recorded among locations. Only trace elements that cumulatively contributed up to 80% of the dissimilarities recorded were selected 39 . A Canonical Analysis of Principal Coordinates (CAP) 39 was performed to test if TEF could be used to predict the fishing location of collected specimens. CAP is a constrained ordination tool that discriminates locations defined a priori and determines the level of misclassification among sampling locations. Appropriate axis (m) was applied by maximizing the leave-one-out allocation success (m = 5) 40 . This approach tests how well locations were discriminated using CAP. To quantify the effect of each trace element to potential differences recorded among locations, Spearman correlation were calculated for all trace elements and the CAP axes. Only the trace elements with a correlation coefficient (r) > 0.30 were considered. ANOVAs were performed using GraphPad Prism 6 (GraphPad Software. Inc., San Diego, CA, USA), and multivariate analyses were performed using PRIMER v6 with the add-on PERMANOVA + .

Results
Five trace elements ( 137 Ba, 24 Mg, 55 Mn, 207 Pb and 88 Sr) were detected in C. edule shells from Ria de Aveiro, with Mg and Sr denoting the highest ratios to Ca (Fig. 2). While no differences between specimens collected by hand or by hand-raking were detected (ANOSIM, p = 0.268, R = 0.025), significant differences among locations were observed for each trace element ratio ( Fig. 2; one-way ANOVA, p < 0.05 for all trace elements; Table 1, summarizes ANOSIM results). The ratios of Mn and Ba were significantly higher at location M2 (p = 0.0001 and 0.0001, respectively). In contrast, the Mg ratio was lowest for C. edule shells from M2 (p = 0.0004). The Pb ratio was only significantly higher (p = 0.0001) at location I, whereas the Sr ratio was also higher at this location but only significantly different from shells collected at E2.
Pairwise comparisons revealed significant differences among locations, apart from those within the Espinheiro Channel, i.e. E1 and E2 (ANOSIM, p = 0.059, R = 0.123). SIMPER analysis showed that the dissimilarity among locations was associated to five elemental ratios: Mg, Sr, Pb, Ba and Mn. (Table 2). Mg and Sr were always among the elements that most contributed for the variability between location M1 and locations from Espinheiro Channel (E1 and E2). Mg and Sr varied significantly between M1 and Espinheiro (p = 0.0001 and 0.021, respectively) and SIMPER revealed that these elements explained more than 55% of the differences recorded between these locations (Table 2). Specimens from location I were significantly different from other areas due to their concentrations of Pb (Fig. 2). SIMPER analysis revealed that Pb alone accounted for 28 to 53% of all differences recorded between location I and all other locations (Table 2). Ba and Mn together contributed for more than 43% of the differences recorded among specimens collected in M2 and other locations.
TEF differences among locations were strong enough to accurately assign collected specimens to their fishing location. The leave-one-out procedure resulted in an average CAP classification of 92% (Table 3), i.e. 92% of the specimens were correctly assigned to their origin. Locations M1, M2 and I had the highest percentage of correct classification (100%), whereas two replicates from E1 and E2 were misclassified, which led to 80% correct classifications. Vector overlay of Spearman correlations of TEF with CAP axes are shown in Fig. 3. Vectors of Ba and Mn ratios were positively correlated with samples from location M2, Mg ratio with areas E1 and E2, and Pb and Sr ratios associated with samples from location I (Fig. 3). C. edule from area M1 were not associated with a particular trace element.

Discussion
In general, adult bivalves display a reduced locomotor ability, being their aragonitic shells potential biogenic archives of marine ecosystems environmental fingerprints 41 . This feature prompted the use of trace elements of bivalve shells to assess their geographic origin. TEF has been successfully used to geographically distinguish populations of blue mussel M. edulis 26 , black mussel M. galloprovincialis and California sea mussel M. californianus 29,30 , soft shell clam Mya arenaria 36 and Olympia oyster Ostrea lurida 42 . This geochemical tool also allowed to distinguish juveniles of green-lipped mussel (Perna canaliculus) ~13 km apart 43 , and to record differences in scallop shells (Argopecten irradians) within a small bay (~10 km 2 ) 44 . The present study shows, for the first time, that TEF of bivalve shells can be used to assign the origin location of bivalves with a resolution <1 km.
Scientific RepoRts | 5:11932 | DOi: 10.1038/srep11932 While Mg and Sr ratios were relatively higher than Ba and Mn (Fig. 2), the latter ratios were among the most important to differentiate locations ( Table 2). The presence of Ba and Mn with elevated concentration, as observed in M2, have been already reported for Isognomon ephippium 45 , Mercenaria mercenaria, Spisula solidissima 46 and M. edulis 47 . Such high concentration in Ba and Mn are usually associated with freshwater inputs and nutrient runoff to estuarine systems, which ultimately causes phytoplankton blooms, particularly diatoms [46][47][48][49] . It is possible that the environmental conditions at M2, which is located more upstream and has stronger riverine input, causes diatom blooms more often than the conditions observed at others locations such as M1, which is located near the inlet 50 . Ba and Mn end up in bivalves' tissue and shell as  a consequence of the ingestion of Ba and Mn-rich particles associated with such diatom blooms. Heavy metals are also incorporated in calcite and aragonite shells of bivalves 51,52 . The high levels of Pb in the shells from location I are likely associated with anthropogenic impacts, particularly acute pollution from boats using leaded gasoline. Note that location I is relatively close to the commercial harbour of Aveiro. Cockle shells from each location displayed a different TEF, with the exception of stations E1 and E2 that showed no statistical differences between each other (Table 1). Nevertheless, CAP results showed a success of 80-100% to identify the origin of cockles collected from Ria de Aveiro (Table 3). The important to highlight that the misclassifications were solely associated with locations E1 and E2. It is the potential of TEF for geographical traceability purposes as we were able to identify the origin of cockles using this statistical tool (CAP) in, at least, 80% of the cases. Nevertheless, the average 92% correct classification is still higher than results by Sorte et al. 26 for the blue mussel M. edulis in the Gulf of Maine (68%) and by Becker et al. 30 for the congeners mussels M. californianus and M. galloprovinciallis in Southern California, USA (56%). The latter and other studies 26,29,30,42,43 have also shown that Ba, Mn, Mg, Pb and Sr play an important role discriminating specimens among areas, as observed here through the magnitude of the vectors of the standardized discriminant functions (Fig. 3).
The chemical nature of the trace elements deposited over time in bivalves is determined by metabolic efficiency and environmental conditions 53 . As this study was conducted in Ria de Aveiro, which is a highly dynamic tidal-system with notable spatial variability in environmental conditions 54 , it is likely that different fingerprints are associated with contrasting environmental conditions recorded at each channel (Fig. 1). The spatial variability here recorded for TEF of cockle shells thus stresses the potential of this method to validate screening for fraudulent use of origin certification. However, temporal variability in environmental conditions may also change TEF and interfere with this traceability tool. Indeed, it has already been shown that seasonal and annual variation may change the TEF of bivalves and other biogenic carbonate structures such as fish otoliths 29,30,55 . In opposite, Carré et al. 56   the marine bivalve species Mesodesma donacium and Chione subrugosa. This study aimed to validate a tool for origin certification of bivalves and not to study the temporal variability of TEF. Consequently, we used cockles with similar size and, therefore, similar age, in order to minimize any bias associated with potential differences in the age of selected specimens. The analyses performed in this study used the whole shell and, consequently, averaged the present and the past elemental fingerprints of cockles. While this approach may have the TEF over multiple years, notable differences among sites were still recorded (Table 1), which emphasizes the robustness of this method for geographical traceability purposes. However, if legal authorities aiming to fight the fraudulent mislabelling of origin location want to minimize this potential temporal bias associated with the analysis of the whole shell, they may rather monitor the elemental fingerprint of the outer margins of bivalve shells from each fishing location or, those more prone to fraud, as these will reflect the most recent elemental fingerprints from the location where they were collected. By comparing the fingerprint of the investigated shells with monitoring data and/or samples from the different sites in the same season, legal authorities may minimize the effect of temporal variability and, ultimately, use of this tool to expose fraudulent situations. Although TEF fails to detect differences associated with fishing method, this information would be potentially relevant for legal authorities to manage bivalve trade, from fishing to the end consumer, as fishermen using hand-raking usually collect larger volumes of bivalves. The effect of environmental conditions on the TEF of bivalves occurs within a relatively long time frame, i.e. within weeks or months 57 , which likely explains the lack of differences associated with fishing methods. The effect of fishing method on TEF, if any, would probably occur within a very small time frame as fishing duration usually takes less than one hour.
Most traceability tools have been focused on issues associated with species mislabeling 13,58,59 or with identification of geographical origin of specimens separated by distances higher than 20 km 26,30 . However this study shows, for the first time, that TEF can be a fast, reliable and accurate method that may be used for origin certification of bivalves collected from locations less than 1 km apart. While this is probably associated with the high environmental variability observed within Ria de Aveiro, it is still unknown if TEF is a reliable tool to accurately identify the origin of bivalves collected from different ecosystems with similarly high variability. Follow-up studies are already being developed to clarify if TEF can be used to discriminate between bivalve shells from specimens originating from distinct ecosystems (from tens to hundreds of km apart). An additional benefit of TEF is that there is no post-harvesting shift and/or degradation associated with bacterial action as recorded for biochemical and molecular methods. The present approach may also play a relevant role on the conservation and management of cockle populations being exploited, namely in the fight against illegal/unreported fishing.