Food Chemistry 170 (2015) 241–248

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Analytical Methods

Stable isotope and trace metal compositions of Australian prawns as a guide to authenticity and wholesomeness J.F. Carter a,⇑, U. Tinggi a, X. Yang a, B. Fry b a b

Queensland Health Forensic and Scientific Service, PO Box 594, Archerfield, QLD 4108, Australia Griffiths University, Australian Rivers Institute, Kessels Road, Nathan, QLD 4111, Australia

a r t i c l e

i n f o

Article history: Received 8 December 2013 Received in revised form 7 August 2014 Accepted 9 August 2014 Available online 19 August 2014 Keywords: Authentication Country of origin Isotope ratio Seafood Stable isotopes Trace metals

a b s t r a c t This research has explored the potential of stable isotope and trace metal profiles to distinguish Australian prawns from prawns imported from neighbouring Asian countries. Australian prawns were collected mostly from the Brisbane area. Strong differences in Australian vs. imported prawns were evident from both the isotope and trace element data, with the differences most likely occurring because imported prawns are typically reared in aquaculture facilities and frozen prior to sale in Australia. The aquaculture origins are characterised by comparatively; low dHVSMOW, d13CVPDB values, low concentrations of arsenic, zinc and potassium, and high water contents (>80%). Relatively high arsenic and cadmium contents were found within Australian prawns, but the concentrations did not exceed local human health guidelines. Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.

1. Introduction Access to global markets is increasingly competitive for many food producers from various countries as consumers are well informed and, therefore, more selective on the quality and authenticity of food products. As a result, producers of foods (and many other commodities) are increasingly protective of the cachet which the country, or region of origin may bestow upon their produce. In turn, the consumer will pay a premium price for what is perceived to be a superior product, attributed to a region of renown such as; Russian caviar, French wine and Spanish ham. Consumers will also associate positive characteristics with fresh produce advertised with qualities such as ‘‘local’’ or ‘‘free-range’’ and may purchase these products in preference to those perceived as ‘‘imported’’ or ‘‘intensively farmed’’. The elevated prices of premium products will inevitably lead unscrupulous traders to pass cheaper, inferior goods as authentic products to the detriment of the producer and consumer. The result of these practices is that the brand reputation of the genuine product is degraded as counterfeit or substitute goods are unlikely to be of the substance or quality demanded by the customer.

⇑ Corresponding author. Tel.: +61 (07) 3274 9229; fax: +61 (07) 3274 9186. E-mail address: [email protected] (J.F. Carter). http://dx.doi.org/10.1016/j.foodchem.2014.08.037 0308-8146/Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.

There has been a significant interest in promoting increased consumption of seafood (including prawns) because of potential health benefits resulting in an increase in seafood production and marketing and in product quality in various markets. There is also a concern from consumers and health regulators that the products available in a marketplace may not come from genuine and good quality producers and may contain chemical residues, including metal contaminants. In such an uncertain marketplace, many techniques have been developed and applied to protect both producers and consumers of premium brands. Manufacturers will, for example, use packaging with features which are difficult to re-produce, incorporating sophisticated designs, metallic strips, holograms etc. Unfortunately, seafood is typically displayed on ice and sold unpackaged such that the only guarantee of a genuine product is the reputation of the seller. As a means to protect both seller and consumer, food analysts are increasingly looking to stable isotope analysis, often allied with trace metal analysis, as a means to determine geographical origin (Asche, Michaud, & Brenna, 2003; LeBot, Oulhote, Deguen, & Glorennec, 2011). These techniques provide powerful tools to determine the origin of foods and to elucidate other properties such as whether the product is wholesome. The isotopic composition of prawn chitin has been reported to reflect growth conditions related to nutrients and the surrounding water environment (Nielson & Bowen, 2010). The present study set out to determine which, if any, parameters could best discriminate between Australian prawns and those imported from nearby Asian

242

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

countries by examining the stable isotopic and trace metal compositions of shells (chitin), meat, heads and recoverable water. As with many studies of food authenticity we encountered difficulty in obtaining truly authentic samples; with the exception of one sample collected directly (and live) from a local prawn farm. In obtaining other samples deemed ‘‘Australian’’ our strategy was to seek out reputable vendors with an established supply chain. In general, we were not able to determine the specific origin of samples described as ‘‘Australian’’ but in this work we have used the term to identify those samples in which we had reasonable confidence. Samples which claimed to be the products of Malaysia, China or Vietnam were assumed to be authentic since there was no commercial advantage to be obtained, within Australia, by branding products as such. In this paper we have used the term ‘‘imported’’ to identify samples branded as product of these countries although prawns may be imported from as far away as Norway or Colombia. The intention of this study was to identify parameters which may be used to determine the country of origin of prawns. It is envisaged that subsequent studies will test the success of these parameters against regional variations within Australia and against a global survey of prawns and other seafood.

Table 1 Prawn samples obtained for the study. Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

2. Materials and methods 2.1. Samples Ten samples of prawns were initially purchased, prepared and analysed, to develop a methodology to distinguish Australian and imported samples. Each sample comprised approximately 1 kg of uncooked prawns which were either whole (heads and shells) or cutlets (tail). These samples were collected to be representative of native Australian and imported species, with both meat and shell present (Table 1, samples 1–10). Australian samples were collected from the Brisbane area. Sample (1) comprised Tiger prawns collected from a Gold Coast prawn farm said to be tenth generation Australian. These prawns had been raised in water taken from the outflow of a local river (the Logan river) and their diet had been supplemented with feed imported from Thailand. Four samples of feed were provided for analysis. Samples 2 and 3 were obtained from a fishmonger (Scarborough, QLD), said to have been freshly caught by local trawlers. Samples 4–6 were purchased from local supermarkets (Sunnybank, QLD) and were sold as frozen and thawed. Four samples of frozen prawn cutlets were purchased from local retailers (Sunnybank, QLD) and were reported to be products of; Malaysia (7, 8), Vietnam (9) and China (10). These ten samples were used to determine which parameters best distinguished Australian and imported species. Subsequently, the method was applied to seventeen samples of uncooked prawns, of which three claimed to be Australian, two claimed to be imported and the remainder where of unknown origin. For the majority of these samples it was not known if they had been sold fresh, frozen or thawed. Prawns were all larger individuals of >80 mm total length and generally of the family Penaeidae. 2.2. Sample preparation Samples were frozen upon receipt at the laboratory and were defrosted prior to preparation. The meat, shell and head (when present) fractions where separated and coarsely homogenised using a commercial food processor. At this stage of preparation the samples were moist, with a small surface area and were exposed to metal surfaces for approximately 10 s. Previous work within our laboratory has shown that this degree of contact introduces no significant metal contamination. All subsequent stages of

a b c d

Description

State

Country of origin

Gold Coast aquaculture tiger prawns Green tiger prawns Green banana prawns Prawn banana raw large fzn Prawn king green xlrge fzn Prawn banana Aust. green lge frzn Raw prawn cutlets peeled-deveinedtail on Raw prawn cutlets-tail on Frozen raw black tiger prawns Vannamei prawns JW prawn farm Local fresh green prawns Local king prawns Frozen raw vannamei Raw prawns peeled Raw prawn cutlets Prawn meat raw Large green king prawns Large green prawns Raw large king Medium king prawn Green headless prawns Green Pacific king prawn headless Medium green king Large banana prawns Prawn meat raw Prawn green king large

Fresh Fresh Fresh Thawed Thawed Thawed Frozen

Australiaa Australiab Australiab Australiac Australiac Australiac Malaysiad

Frozen Frozen Frozen

Malaysiad Vietnamd Chinad Australiaa Australiab Australiab Importedb Malaysiac Unknownb Unknownb Unknownb Unknownb Unknownb Unknownb Unknownb Unknownb Unknownb Unknownc Unknownc Unknownc

Thawed Frozen Frozen

Thawed Frozen Frozen

Prawn farm. Fishmonger. Major supermarket. Local retailer.

handling, in which either contact times or surface areas where large were performed using metal free materials. Samples were then placed in grip-seal polyethylene bags and frozen at 20 °C. The meat, shell and head components from each sample were dried together for approximately 30 h using a Dynavac FD12 freeze-drier (Seven Hills, NSW, Australia). The weigh loss on drying was recorded and the water from each group of samples was recovered from the cold-trap of the freeze-drier and retained for isotopic analysis. To determine the extent of any isotopic fractionation due to the freeze drying processes a sample of MilliQ grade water (approximately 250 g) was frozen, ‘‘freeze dried’’ and re-collected from the cold-trap. The dried samples were de-fatted with hexane (SupraSolv grade, Merck, Darmstadt, Germany), using a Soxtherm automatic extraction system (Perten Instruments, Australia). Chitin was purified from the dried, de-fatted shells according to published methods (Percot, Viton, & Domard, 2002), summarized as follows. Approximately two grams of de-fatted shell were reacted with 80 mL of 0.25 M hydrochloric acid for 24 h and then washed with de-ionised water to neutral pH. The sample was then treated with 80 mL of 1.0 M sodium hydroxide at 80 °C for 18 h and again washed with deionised water to neutral pH. Samples were finally dried in a vacuum oven at 40 °C overnight. Samples of meat, head, purified chitin or prawn feed were ground to fine powders using a vibratory ball mill equipped with a zirconium oxide mortar and ball (Fritsch, Idar-Ocerstein, Germany). Liquid nitrogen was added to chitin samples to make them brittle for grinding. Grinding times were adjusted to achieve a visibly homogeneous product with the consistency of fine flour. We were aware that the extent of grinding has been reported to affect to amount of exchangeable-hydrogen available within protein samples (Bowen, Chesson, Nielson, Cerling, & Ehleringer, 2005). As a simplified approach we applied a correction for

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

exchangeable hydrogen based on published literature (Chesson, Podlesak, Cerling, & Ehleringer, 2009), discussed below. 2.3. Stable isotopic analysis All isotopic measurements were performed using a Thermo Scientific (Bremen, Germany) Delta VPLUS isotope ratio mass spectrometer (IRMS) coupled to a ConFlo IV interface for working gas introduction and sample dilution. A list of the reference materials used for normalisation is given in the Supplementary data. Hydrogen and oxygen isotopic analysis was performed using a Thermo Scientific Flash 2000 HT elemental analyser (EA) configured for Thermal Conversion (TC/EA) operating at 1400 °C with a helium flow of 100 mL/min introduced via a bottom feed adaptor. Samples of water (0.15 lL) were introduced by an AS3000 autosampler. Six aliquots of water were injected during each analytical run; the result from the first injection was discounted to allow for memory effects associated with the sample introduction and reactor. A sample of purified Brisbane tap water was analysed throughout the analytical sequence to act as an in-house quality control (QC) material and to compensate for any drift in the internal instrument d-scale. Data were normalised to the VSMOW-SLAP scale. The 2H and 18O composition of solid samples were determined using the same instrument parameters. Samples were measured into silver capsules (4  3.2 mm, Elemental Microanalysis, Okehampton, UK) and equilibrated with purified Brisbane tap water of known isotopic composition for 3 days. Samples were then dried over phosphorous pentoxide for three days, under vacuum. Capsules were then crimped closed and introduced into the reactor via a MAS200R auto-sampler. The d-values of the non-exchangeable hydrogen content were estimated from the measured values of the sample and the equilibration water by mass balance (Eq. (2)). The fraction of exchangeable H atoms (fex) for muscle was taken as 0.2 and for chitin as 0.15 (Chesson et al., 2009). For both materials the fractionation factor (a) between H atoms in the equilibration water and sample was taken as 1 (Chesson et al., 2009). This estimation of non-exchangeable hydrogen was considered appropriate for a comparative study.

Rnex ¼ Rmeas  ðRwater  a  f ex Þ=ð1  f ex Þ

ð1Þ

Rnex, non-exchangeable H isotope ratio; Rmeas, measured H isotope ratio; Rwater, equilibration water H isotope ratio. The 13C and 15N composition of solid samples were determined using a Thermo Scientific Flash 2000 HT elemental analyser configured for flash combustion at 950 °C with a helium flow of 100 mL/ min. Samples were enclosed in tin capsules (8  5 mm, IVA, Analysentechnik, Meerbusch, Germany) and introduced into the reactor via a MAS200R auto-sampler. The reactor comprised a single quartz tube containing chrome oxide, reduced copper and silver/ cobaltous oxide. For d13C determination a chemical trap containing magnesium perchlorate was inserted between the reactor and GC column. For d15N determination a chemical trap containing magnesium perchlorate and Carbosorb™ was inserted between the reactor and GC column. During the analysis of solid samples, a sample of chitin (Sigma Aldrich Pty. Ltd., Sydney, Australia) was analysed throughout the analytical sequence to act as an in-house QC material and to correct for any drift in the internal instrument d-scale.

243

overnight to allow slow digestion and gases to evolve. The samples were then sealed and digested using a MarsXpress microwave digestion system (CEM, Matthews, USA) with the digestion program set at 3 stages as follows: stage 1 at power (400 watts), temperature (85 °C), time (14 min); stage 2 at power (800 watts), temperature (110 °C), time (20 min); stage 3 at power (1600 watts), temperature (160 °C), time (10 min). The digested samples were diluted and made up to 40 mL with high purity water (18 mohm, Milli-Q Element System). Digested samples were analysed using a 7700 Inductively Coupled Plasma (ICP)-MS (Agilent Technologies Australia Pty. Ltd., VIC, Australia), with auto-sampler and Integrated Sample Introduction System (ISIS). Operating parameters are given in the supplement. Multi-element and single-element stock standard solutions (10 mg/L) (CHOICE Analytical, NSW, Australia) were used to prepare calibration solutions of 0, 0.1, 1, 10, 100 and 1000 lg/L in 5% HNO3 for screening of 30 elements. A series of blank values containing 5% HNO3 reagent (n = 7) were determined and used for blank correction and limits of detection evaluation. Certified reference material of oyster tissue (NIST, SRM 1566b), mussel tissue (NIST, SRM 2976), and in-house reference materials of freeze dried mussel tissue were used for quality control. For arsenic species determination approximately 0.02 g of freeze-dried sample was accurately weighed into a 10 mL plastic tube. Five millilitre of 0.1 M phosphate buffer solution was added, followed by 40 lL of proteinase K digestion solution (QIAGEN Pty. Ltd., VIC, Australia). The mixture was incubated for 4 h at 37 °C on a rotary unit at 20 rpm. The digested sample was then diluted to 10 mL with high purity water and centrifuged for 10 min at 3000 RPM. The supernatant was filtered through a 0.2 lm PVDF membrane filter before analysis. Separation of arsenic species was performed using an HP1100 High Performance Liquid Chromatograph HPLC (Agilent Technologies Australia) equipped with a G3288-80000 (4.6  250 mm) anionic exchange column (Agilent Technologies Australia). The mobile phase was 2 mM, pH 11 phosphate buffer at a flow rate of 1.0 mL/min. Detection was achieved by monitoring m/z 75 (75As) using the ICP-MS instrument described above. Mixed calibration standards of sodium arsenite (AsIII), sodium arsenate (AsV), dimethylarsinic acid (DMA), monomethylarsonic acid (MMA) and arsenobetaine (AsB) were prepared to contain 0, 0.1, 0.5, 1, 5, 10, 50 lg/L as arsenic. The certified reference material of tuna fish tissue (CRM BCR-627) was used for quality control. The results for AsB (4.16 ± 0.38 mg/kg As) and DMA (0.19 ± 0.02 mg/kg As) were within the certified values of AsB (3.90 ± 0.20 mg/kg) and DMA (0.15 ± 0.02 mg/kg). The isotopic composition and trace metal concentration data obtained during this study are presented in the Supplementary data together with empirical standard deviations. The in-house measurement uncertainty determined for d13C, d15N and d18O measurements were 0.2‰ and for d2H measurements 2.0‰. Any differences between analytical data less than these values were considered insignificant. Data were analysed using R 2.15.2 software environment for statistical computing and graphics (R core team, 2010).

3. Results and discussion 3.1. Physical composition of samples

2.4. Trace metal and arsenic species analysis Approximately 0.3 g of homogenised meat or heads was accurately weighed into PTFE digestion vessels. Four millilitre of high purity nitric acid (69% Seastar Chemicals, Canada) were added to each tube which were then left to stand, at room temperature,

Of the initial suite of ten samples, those sold frozen were found to contain significantly more water in the meat component (typically 84%) compared to samples sold as fresh or thawed (typically 74%), with one sample carrying the declaration ‘‘contains 10% added seawater for freshness’’. Fat was extracted from the samples to

244

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

The meat component was consistently depleted in 18O with respect to the chitin; D18Omeat/chitin ranged from 6.8‰ to 11.6‰ with a median value of 9.2‰. The meat components were also significantly depleted in 2H with respect to the chitin; D2Hmeat/chitin ranged from 25‰ to 50‰ with a median value of 34‰. For both meat and chitin components, the imported samples were depleted in both 2H and 18O with respect to the Australian samples with a clear distinction observed for the meat for this suite of samples. Data obtained from the carbon and nitrogen isotopic analysis of the prawn meat and chitin are summarised in Fig. 2. In contrast to 2 H/18O data, the meat components were consistently enriched in the heavy isotope with respect to the corresponding chitin; D13Cmeat/chitin ranged from 2.5‰ to 4.3‰ with a median value of 3.7‰ while D15Nmeat/chitin ranged from 8.0‰ to 13.7‰ with a median value of 12.4‰. The smallest D15Nmeat/chitin value was observed for sample 1 which was known to be a product of aquaculture. In such a closed system, it is probable that discarded shells will provide a food source for other prawns whereas, in the wild, little of the chitinnitrogen will be recaptured in this way. This effect was not, however, observed for imported cutlets which were assumed to be the product of aquaculture. Samples 2 and 3, claimed to have been locally trawled (Morton Bay, QLD) and were found to have the most positive d15NAIR values (+11.0‰ and +12.0‰) indicative of high nutrient content typified by input from waste waters and observed for biota from nitrogen-enriched rivers and bays in the Brisbane region (Costanzo, Udy, Longstaff, & Jones, 2005; Pitt, Connolly, & Maxwell, 2009). Both the meat and chitin of the Australian samples were consistently enriched in 13C with respect to the imported samples with the exception of sample 1 (shown red) for which the diet consisted of a mixed source of naturally present nutrients and imported feed. All of the prawn feed samples tested had d13C and d15N compositions consistent with the meat component of imported samples. This implies a protein content derived from seafood and explains why the isotopic composition of sample 1 resembled the imported samples more closely that the other Australian samples. Table 2 presents a Pearson correlation matrix for all of the stable isotope measurements. The isotopic composition of all elements showed a strong correlation between the meat and chitin e.g. for d13C measurements r = 0.98. This indicates that equivalent data were obtained from meat and chitin and therefore only one of these components needs to be analysed. The correlation shown for d15N meat/chitin excludes the data for sample 1 since this sample had an atypical D15Nmeat/chitin value (if these data are included

improve grinding characteristics and this extraction proved to be fortuitous since it has been recently reported that the lipid content of fish muscle is highly depleted in 2H with respect to the fat-free tissue (Soto, Wassenaar, & Hobson, 2013). The fat content of the samples was not quantified or analysed; prawns typically contain less than 2% fat (Kirk & Sawyer, 1991). 3.2. Isotopic composition of recovered water To investigate possible isotopic fractionation of water recovered from the freeze drying, a sample of high purity water was frozen, ‘‘freeze-dried’’ and collected from the cryo-trap. A small depletion in both 2H and 18O was observed during freeze-drying (2.5‰ with respect to 2H and 0.2‰ with respect to 18O). These differences were comparable to experimental uncertainty and no corrections were applied to other data. Results for water recovered from prawns fell near the Global Meteoric Water Line (GMWL; left dotted line in Fig. 1) that describes results for most natural waters, according to Eq. (2).

d2 H ¼ d18 O  8 þ 10

ð2Þ

The solid line in Fig. 1 shows the trend for the water recovered from prawns (d2H = d18O  6  15, R2 = 0.87) which was offset from the GMWL. A sample of Brisbane drinking water, which is drawn from various water storages located around Brisbane and the Gold Coast, fell on the GMWL. The solid circle close to this data point corresponds to the sample from the Gold Coast prawn farm implying that the water was drawn from the same source. A sample of Moreton Bay seawater was found to be enriched in 18 O with respect to the GMWL. This can be explained since Moreton Bay is a relatively enclosed body of water subject to significant evaporation. The enclosed circle close to this data point does not, however, correspond to samples supposedly trawled from Morton Bay; these samples appear towards the bottom of the line. Overall, the isotopic composition of the waters recovered from the prawns did not correlate with origin nor distinguish between Australian and imported species. These findings suggest that the post harvest use of water, in freezing or icing may mask any original isotopic signature. 3.3. Isotopic composition of meat and chitin The dotted line to the right in Fig. 1 shows the GMWL off-set by +28‰ which is the fractionation typically associated with biosynthesis (Sternberg & DeNiro, 1983). The majority of meat and chitin samples lie on, or close to, this line.

GMWL 20 Moreton Bay

103 × δ 2HVSMOW

0 -20

Chitin

Brisbane

-40

Water

-60

Meat

-80 25

15

0

5

15

-5

10 × δ OVSMOW 3

18

Fig. 1. Oxygen and hydrogen isotopic data for the prawn components; water (d), meat (j) and chitin (). Australian samples are shown as solid symbols and imported samples as open symbols. Two samples of local water are also shown as blue triangles. (N). Red symbols indicate sample 1, obtained from a prawn farm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

245

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

103 × δ 15NAIR

Meat

Chitin

103 × δ 13CVPDB Fig. 2. Carbon and nitrogen isotopic data for the prawn components; chitin () and meat (j). Australian samples are shown as solid symbols and imported samples as open symbols. Results from samples of prawn food are also shown (N). Red symbols indicate sample 1, obtained from a prawn farm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2 Pearson correlation matrix for the isotopic compositions of water, meat and chitin. Values greater than the critical value (0.632 for n2 = 8) are shown bold. Water 2

Water Meat

Chitin

2

H 18 O 2 H 13 C 15 N 18 O 2 H 13 C 15 N 18 O

H

1.00 0.93 0.47 0.02 0.12 0.18 0.43 0.06 0.20 0.14

Meat 18

O

1.00 0.55 0.26 0.13 0.42 0.55 0.28 0.13 0.31

Chitin

2

13

1.00 0.55 0.36 0.65 0.82 0.57 0.56 0.80

1.00 0.39 0.68 0.38 0.98 0.12 0.55

H

C

the correlation is still significant r = 0.78). No significant correlation was observed between the d2H/d18O compositions of the recovered water and the compositions of the meat or chitin. This may explain why no distinction between Australian and imported species was evident in Fig. 1. In contrast, significant correlations were apparent in the d2H and d18O compositions within the meat and chitin and between the meat and chitin. Both the meat and chitin also showed significant correlation between the d13C and d18O composition. These observations are consistent with recent studies that demonstrate both water and diet are important in determining the H and O isotopic composition of animals (Soto et al., 2013). A model was developed based on d13C/d2H composition because this combination of parameters distinguished between Australian and imported samples for the prawn meat of the initial suite of samples. Fig. 3a shows the basic model used to discriminate Australian and imported prawns, a scatter plot of d2H vs. d13C for prawn meat, with 90% confidence ellipses based on samples of known provenance. The distinction between Australian and imported samples is apparent and all of the unknown samples fall within one or other ellipse. On the basis of this model, 9 unknown samples were identified as Australian and three as imported. The 9 samples identified as Australian were purchased from fishmongers. The three samples identified as imported (16, 17, 26) were all purchased from supermarkets and contained between 82% and 86% moisture which was typical of imported samples. Further, one of these samples was described as ‘‘cutlets’’ and the two others as ‘‘raw prawn meat’’ which were typical of the imported samples. The two samples

15

N

1.00 0.02 0.07 0.40 0.95 0.42

18

O

1.00 0.81 0.72 0.09 0.72

2

13

1.00 0.41 0.18 0.76

1.00 0.11 0.54

H

C

15

N

1.00 0.53

18

O

1.00

obtained from local prawn farms (1 and 11) had d13C values close to the imported samples, but were distinguished by d2H composition. If these samples were allocated a separate group, the Australian grouping would be tighter and better distinguished from the imported samples. Overall, Fig. 3a demonstrates the potential of combined d2H/ d13C measurements to distinguish between Australian and imported seafood. The model may be improved with the inclusion of more authenticated samples from different sources and, possibly the addition of physical parameters, such as moisture content, and further groupings for farmed samples. 3.4. Trace metal composition 3.4.1. Provenance For the first suite of ten samples, the concentrations of 30 elements were determined. Of these, the concentrations of many elements were below the detection limit for the majority of samples and only nineteen elements were included in subsequent calculations. Data analyses were performed using the concentration of metals determined in the dried samples, since imported and frozen samples contained a consistently higher proportion of water, which might bias the concentration data, especially if this water was incorporated post-harvest. To determine which elemental concentrations might discriminate between Australian and imported samples, Principle Component Analysis (PCA) was performed using the L1-norm to increase robustness to outliers compared to traditional PCA (Brooks, Dulá, & Boone, 2013). Concentrations which were reported as below

246

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

(a)

(b)

1

100

11

80

[zinc] mg/kg

103 × δ 2HVSMOW

-40

-60

2 60

22 40

-80

26 20

-25

-20

10 × δ 3

13

-15

CVPDB

5000

10000

15000

20000

[potassium] mg/kg

Fig. 3. Scatter plots of (a) d2H vs. d13C and (b) the concentration of zinc vs. potassium for prawn meat from Australian (d/red), imported (s/blue) and unknown () sources. Confidence ellipses (at 90% confidence) are based on samples of known origin. Dotted lines show equidistance between ellipses. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

the detection limit were substituted with the limit of detection for data analysis. The loading factors for pc1 were dominated by potassium with smaller contributions from arsenic, zinc and copper and the loading factors for pc2 were dominated by arsenic, with contributions from copper and zinc. This suggested that a combination of the concentrations of potassium, arsenic, copper and zinc provides a means to discriminate between Australian and imported samples. Pearson correlation showed significant covariance in the concentrations of copper and zinc (0.82) suggesting that consideration of either element would be sufficient. Canonical Discriminant Analysis (CDA) was applied to the data to determine which suite of measured parameters provided the best classification. The stable isotopic data provided 100% correct classification as did a combination of zinc and arsenic concentration. All other combinations of metal concentrations proved less effective with correct classification typically falling to 87%. The reason for this can be illustrated by Fig. 3b which shows a scatter-plot of the concentrations of zinc vs. potassium. Although the distribution of these data appears visually similar to that of the isotopic data (Fig. 3a) there are a number of important differences. First, the confidence ellipses (drawn at the 90% confidence level) overlap significantly. The confidence ellipse for imported samples encloses samples 16 and 17 but sample 26 lies outside, albeit on the correct side of the dividing line. This was typical of the models that could be generated for other combinations of elements; the imported samples appearing as a small closely defined group and the Australian samples as a much larger dispersion. Second, sample 2 falls closer to the imported grouping, due to a low concentration of potassium. If sample 2 were excluded from the data, the confidence ellipse for the Australian prawns would be smaller and separate from the imported samples. Sample 22 also falls closer to the imported grouping, but was assigned as Australian, based on isotopic composition. Overall, the profile of potassium, arsenic and zinc present in the prawn meat provided a further means to discriminate between Australian and imported samples, albeit with less certainty than isotopic analysis. A combination of stable isotope and trace metal analysis provides enhanced certainty for the origin of seafood.

3.4.2. Health considerations Another facet of this work was to identify potentially harmful concentrations of toxic elements within the edible components of the prawn samples. Fig. 4 presents the concentration of arsenic in the prawn samples adjusted for moisture content and, from this figure, it is easy to understand why samples could be classified as Australian or imported, based on trace metal concentration. There is a clear distinction between Australian and imported samples; the Australian samples containing significantly higher concentrations of arsenic than the imported samples. Within the Australian samples, there is also a distinction between samples purchased as fresh (F) and those purchased as frozen and thawed (T), the later containing a considerably higher concentration of arsenic. The consistently high levels of arsenic in the thawed samples compared to the fresh samples are difficult to explain; the thawed samples were not treated differently in the laboratory in any way that could contribute to arsenic exposure. The high level of arsenic in meat, however, at about 34 mg/kg (Fig. 4) should not cause public heath concern, as arsenic in seafood is predominantly in the form of AsB, a harmless form of arsenic species (Francesconi, 2010; Leufroy, Noël, Dufailly, Beauchemin, & Guérin, 2011). It has been reported that shrimp can accumulate arsenic and the level can be as high as 22 mg/kg (Ruangwises & Ruangwises, 2011). In the present study, 3 samples each of meat and head, with high total arsenic concentrations, were analysed for arsenic species (AsB, DMA, MMA, AsIII and AsV). The species was found to be predominantly AsB which ranged from 104 to 129 mg/kg for prawn meat and 73 to 141 mg/kg for prawn heads (dried weight). No detectable amounts of inorganic arsenic, DMA or MMA were found in these samples. For the six samples purchased whole, the concentration of arsenic in the head was consistent with the concentration in the meat. On average, however, the heads were found to contain approximately five times the concentration of any given metal compared to the meat with much higher concentration factors observed for strontium (37), barium (21), silver (40) and cadmium (53). The concentration of cadmium in the meat ranged from 0.01 to 0.33 mg/kg, whereas the heads for the thawed

247

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

40 35

[arsenic] mg/kg

30 25 20 15 10 5

* * *

0

F

T Australian

imported

unknown

Fig. 4. Concentrations of arsenic in prawn meat (blue/light) and heads (red/dark). Samples which had isotopic, trace metal and physical characteristics (specifically moisture content) consistent with known imported samples are shown (star). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

samples contained up to 28 mg/kg. Pearson correlation showed a very strong correlation between the concentrations of arsenic and cadmium (0.98). The levels of cadmium in prawn meat in this study are comparable to recent study of Australian prawns where values ranged from 0.2 to 0.5 mg/kg (dry weight) (Lewtas, Birch, & Foster-Thorpe, 2013) although higher values (1.2 mg/kg dry weight) have also been reported (Falcó, Llobet, Bocio, & Domingo, 2006). In agreement with previous findings the accumulation of metals in gills and hepatopancreas, the digestive gland located in the prawn head, explains the high metal concentrations including cadmium (Darmono & Denton, 1990; Lewtas et al., 2013; Peerzada, Nojok, & Lee, 1992). The low levels of mercury found in these prawn samples are comparable to other reported values (Falcó et al., 2006; Pastorelli et al., 2012). One of the four samples of prawn feed obtained from the Australian prawn farm was found to contain higher concentrations of virtually all metals, specifically 4.7 mg/kg of arsenic and 1.2 mg/kg of cadmium. The concentrations of most elements were higher in this feed than in the dry weight of prawn meat or head. Of note, three elements were present in higher concentrations in some prawns, compared to this feed, and these were the elements which best distinguished Australian and imported samples; potassium, arsenic and zinc. 4. Conclusions A small convenience sample of prawns was collected to establish the potential of stable isotope and trace metal profiling to distinguish prawns deemed to be Australian and those labelled as the produce of neighbouring Asian countries. The samples deemed Australian comprised fresh samples from prawn farms and fishmongers and frozen/thawed samples from supermarkets. The samples deemed to be imported comprised frozen prawn cutlets. Isotopic analysis of the water recovered from the prawns did not provide a means to discriminate between Australian and imported samples and did not correlate with the isotopic composition of the tissues. This was assumed to be the results of postharvest treatment with water for chilling or freezing. Both meat and chitin were enriched in 18O compared to the recovered water and, for any given sample, the chitin was consistently enriched in both 2H and 18O compared to the meat. The meat from Australian prawns was consistently enriched in 2H with respect to the imported species and it was possible to distinguish the two groups on this basis.

The prawn meat was consistently enriched in both 13C and 15N with respect to the chitin. For both components the Australian samples were enriched in 13C compared to the imported species and this provided another basis for discrimination. Equivalent isotopic information could be determined from either the meat or chitin components of prawns. Overall, a combination of d2H and d13C data for the meat component was found to distinguish Australian and imported species. When this model was applied to prawn meat from a second suit of samples, it correctly classified samples of known origin. Samples of unknown origin which the model identified as imported had physical characteristics typical of other imported samples. Analysis of the trace metal composition showed that a combination of potassium, arsenic and zinc concentrations could distinguish between Australian and imported species. Results for unknown samples were in agreement with the isotopic data. Supermarket samples, sold as frozen/thawed, were found to contain significant concentrations of arsenic in both the edible meat and the heads. These samples also contained high concentrations of cadmium in the heads. Analysis of these data showed significant accumulation of metals in the heads when compared to the meat. Arsenic in both the meat and heads was predominantly in form of AsB, a harmless arsenic species. For further study, it would be useful to undertake the analysis of metal species, particularly arsenic to identify the origin and any associated health risk. Acknowledgements The authors are grateful for funding from Queensland Government; Cabinet Research Project No. RSS12-011(B3). The authors wish to thank Gary Golding for the original idea behind this research. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodchem. 2014.08.037. References Asche, S., Michaud, A. L., & Brenna, J. T. (2003). Sourcing organic compounds based on natural variations measured by high precision isotope ratio mass spectrometry. Current Organic Chemistry, 7, 1527–1543.

248

J.F. Carter et al. / Food Chemistry 170 (2015) 241–248

Bowen, G. J., Chesson, L., Nielson, K., Cerling, T. E., & Ehleringer, J. R. (2005). Treatment methods for the determination of d2H and d18O of hair keratin by continuous-flow isotope-ratio mass spectrometry. Rapid Communications in Mass Spectrometry, 19(17), 2371–2378. Brooks, J. P., Dulá, J. H., & Boone, E. L. (2013). A pure L1-norm principal component analysis. Computational Statistics & Data Analysis, 61, 83–98. Chesson, L. A., Podlesak, D. W., Cerling, T. E., & Ehleringer, J. R. (2009). Evaluating uncertainty in the calculation of non-exchangeable hydrogen fractions within organic materials. Rapid Communications in Mass Spectrometry, 23(9), 1275–1280. Costanzo, S. D., Udy, J., Longstaff, B., & Jones, A. (2005). Using nitrogen stable isotope ratios (d15N) of macroalgae to determine the effectiveness of sewage upgrades: Changes in the extent of sewage plumes over four years in Moreton Bay, Australia. Marine Pollution Bulletin, 51(1–4), 212–217. Darmono, D., & Denton, G. R. W. (1990). Heavy metal concentrations in the banana prawn, Penaeus merguiensis, and leader prawn, P. monodon, in the townsv region of Australia. Bulletin of Environmental Contamination and Toxicology, 44(3), 479–486. Falcó, G., Llobet, J. M., Bocio, A., & Domingo, J. L. (2006). Daily intake of arsenic, cadmium, mercury, and lead by consumption of edible marine species. Journal of Agricultural and Food Chemistry, 54(16), 6106–6112. Francesconi, K. A. (2010). Arsenic species in seafood: Origin and human health implications. Pure and Applied Chemistry, 82(2), 373–381. Kirk, R. S., & Sawyer, R. S. (1991). Pearson’s composition and chemical analysis of foods. Harlow: Longman. LeBot, B., Oulhote, Y., Deguen, S., & Glorennec, P. (2011). Using and interpreting isotope data for source identification. Trends in Analytical Chemistry, 30(2), 302–312. Leufroy, A., Noël, L., Dufailly, V., Beauchemin, D., & Guérin, T. (2011). Determination of seven arsenic species in seafood by ion exchange chromatography coupled to

inductively coupled plasma-mass spectrometry following microwave assisted extraction: Method validation and occurrence data. Talanta, 83(3), 770–779. Lewtas, K. L. M., Birch, G. F., & Foster-Thorpe, C. (2013). Metal accumulation in the greentail prawn, Metapenaeus bennettae, in Sydney and Port Hacking estuaries, Australia. Environmental Science and Pollution Research, 20, 1–13. Nielson, K. E., & Bowen, G. J. (2010). Hydrogen and oxygen in brine shrimp chitin reflect environmental water and dietary isotopic composition. Geochimica et Cosmochimica Acta, 74, 1812–1822. Pastorelli, A. A., Baldini, M., Stacchini, P., Baldini, G., Morelli, S., Sagratella, E., et al. (2012). Human exposure to lead, cadmium and mercury through fish and seafood product consumption in Italy: A pilot evaluation. Food Additives & Contaminants: Part A, 29(12), 1913–1921. Peerzada, N., Nojok, M., & Lee, C. (1992). Distribution of heavy metals in prawns from Northern Territory, Australia. Marine Pollution Bulletin, 24(8), 416–418. Percot, A., Viton, C., & Domard, A. (2002). Optimization of chitin extraction from shrimp shells. Biomacromolecules, 4(1), 12–18. Pitt, K. A., Connolly, R. M., & Maxwell, P. (2009). Redistribution of sewage-nitrogen in estuarine food webs following sewage treatment upgrades. Marine Pollution Bulletin, 58(4), 573–580. R core team. (2010). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Ruangwises, S., & Ruangwises, N. (2011). Concentrations of total and inorganic arsenic in fresh fish, mollusks, and crustaceans from the Gulf of Thailand. Journal of Food Protection, 74(3), 450–455. Soto, D. X., Wassenaar, L. I., & Hobson, K. A. (2013). Stable hydrogen and oxygen isotopes in aquatic food webs are tracers of diet and provenance. Functional Ecology, 27, 535–543. Sternberg, L. S. L. O., & DeNiro, M. J. D. (1983). Biogeochemical implications of the isotopic equilibrium fractionation factor between the oxygen atoms of acetone and water. Geochimica et Cosmochimica Acta, 47(12), 2271–2274.