Science of the Total Environment 494–495 (2014) 18–27

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Persistent toxic substances in Mediterranean aquatic species

Roberto Miniero a,⁎, Vittorio Abate a, Gianfranco Brambilla a, Enrico Davoli b, Elena De Felip a,Stefania P. De Filippis a, Elena Dellatte a, Silvia De Luca a, Roberto Fanelli b, Elena Fattore b, Fabiola Ferri a,Igor Fochi a,1, Anna Rita Fulgenzi a, Nicola Iacovella a, Anna Laura Iamiceli a, Dario Lucchetti c, Paolo Melotti d,Ivo Moret e, Rossano Piazza e, Alessandra Roncarati f, Alessandro Ubaldi c,Stefano Zambon e, Alessandro di Domenico a

a Department of the Environment and Primary Prevention, Italian National Institute for Health, Rome, Italyb Department of Environmental Health Sciences, Istituto di Ricerche Farmacologiche “Mario Negri”, Milan, Italyc Istituto Zooprofilattico Sperimentale del Lazio e Toscana, Rome, Italyd School of Veterinary Medical Sciences, EAEVE Certificate, Camerino University, Matelica, Italye National Research Council, Venice, Italyf Centro di Ricerca Interdipartimentale sulle Tecnologie e l’Igiene delle Piccole Specie, Department of Food Science, Alma Mater Studiorum, Bologna University, Ozzano dell’Emilia, Italy

H I G H L I G H T S

• PTS concentration levels in Mediterranean fish and fishery products were studied• Data were collected on many contaminants in wild and farmed species of commercial interest• Most samples were compliant with EU regulatory limits except for high predators.• Mediterranean Sea areas with fish at background contamination levels were identified.• The study outcome was evaluated in the light of the Marine Strategy Framework Directive.

⁎ Corresponding author. Tel.: +39 06 49902067.E-mail address: roberto.miniero@iss.it (R. Miniero).

1 Temporary co-worker.

http://dx.doi.org/10.1016/j.scitotenv.2014.05.1310048-9697/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 21 March 2014Received in revised form 28 May 2014Accepted 28 May 2014Available online xxxx

Editor: Adrian Covaci

Keywords:Persistent organic pollutantsPersistent toxic substancesMediterranean SeaFishCrustaceansFood safety

Fish and fishery productsmay represent one of themain sources of dietary exposure to persistent toxic substances(PTSs) such as polychlorinated dibenzodioxins, dibenzofurans, and biphenyls; polybromodiphenyl ethers; organ-ochlorine pesticides; perfluorooctanoic acid and perfluorooctane sulfonate; and inorganic mercury and methylmercury. In this study, PTS contamination of Mediterranean fish and crustaceans caught in Italian coastal waterswas investigated in order to increase the representativeness of the occurrence database for wild species. The ob-jectives were to verify the suitability of regulatory limits for PTSs, identify background concentrations values, ifany, and examine the possible sources of variability when assessing the chemical body burdens of aquatic species.Twelve wild species of commercial interest and two farmed fish species were chosen. Excluding methyl mercury,chemical concentrations found in wild species fell generally towards the low ends of the concentration rangesfound in Europe according to EFSA database and were quite lower than the tolerable maximum levels establishedin the European Union; farmed fish always showed contamination levels quite lower than those detected in wildspecies. The data obtained for wild species seemed to confirm the absence of local sources of contamination in thechosen sampling areas; however, species contamination could exceed regulatory levels even in the absence of spe-cific local sources of contamination as a result of the position in the foodweb and natural variability in species' life-style. A species-specific approach to themanagement of contamination in aquatic organisms is therefore suggestedas an alternative to a general approach based only on contaminant body burden. A chemical-specific analysisperformed according to organism position in the food chain strengthened the need to develop this approach.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Fish and fishery products may represent one of the main sources ofdietary exposure to the following persistent toxic substances (PTSs):polychlorinated dibenzodioxins (PCDDs), dibenzofurans (PCDFs)

19R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

(together also known as “dioxins”), and biphenyls (dioxin-like: DL-PCBs;non-dioxin-like: NDL-PCBs); polybromodiphenyl ethers (PBDEs);perfluorooctanoic acid and perfluorooctane sulfonate (PFOA, PFOS); or-ganochlorine pesticides (OCPs) such as aldrin, cis- and trans-chlordane,2,4′- and 4,4′-dichlorodiphenyldichloroethane (2,4′-DDD and 2,4′-DDD), 2,4′- and 4,4′-dichlorodiphenyldichloroethylene (2,4′-DDE and4,4′-DDE), 2,4′- and 4,4′-dichlorodiphenyltrichloroethane (2,4′-DDTand 4,4′-DDT), dieldrin, endrin, heptachlor, cis- and trans-heptachlorepoxide, hexachlorobenzene (HCB), α-hexachorocyclohexane (α-HCH), β-HCH, and γ-HCH (lindane). The Mediterranean area is alsoconsidered to be a critical environment for coastal populations due to apotential overexposure to inorganic mercury (Hg) and methyl mercury(MeHg) when diets rich in fish are consumed (INRAN, 2012; UNEP,2002).

Food safety andmanagementmeasures in the European Union (EU)received much attention in recent years. On the basis of dioxin(2,3,7,8-T4CDD) toxicity equivalency factors (TEFs) adopted by theWorld Health Organization (WHO) in 2005 (Van den Berg et al.,2006), Regulation (EU) 1259/2011 updated the maximum tolerablelevels (MLs) in dioxin toxicity equivalents (TEQ units) for PCDDs +PCDFs and PCDDs + PCDFs + DL-PCBs (TEQTOT) in food, including fishand alike products. The Regulation also established de novo MLs forthe sum of the six “indicator” NDL-PCBs 28, 52, 101, 138, 153, and 180(Σ6NDL-PCBs). Recommendation 2013/711/EU updated action levels(ALs) for PCDDs + PCDFs and DL-PCBs in food; in the case of fish andfishery products, an AL is available for the farmed produce. Regulation(EC) 1881/2006 provides MLs for Hg (expressed as total Hg, HgTOT) infish and fishery products. No EU MLs concerning the same food areavailable for the remaining organohalogenated compounds of interestin this study (aldrin, chlordane, DDD, DDE, DDT, dieldrin, endrin, hepta-chlor, heptachlor epoxides, HCB, HCH, PBDEs, PFOA, and PFOS),although certain Member States have their own regulatory limits.

In early 2004, the Italian Ministry for Education, University, andResearch financed a 5-year study (2004–2008) to characterize thechemical contamination and nutritional quality of fish and crustaceanscaught in Italian coastal waters. The objectives of the study were(a) to increase the representativeness of chemical occurrence databasefor wild as well as farmed species collected in the same samplingareas so as to allow a statistically significant comparison of contaminantlevels, (b) to compare these levels with regulatory limits and possiblyprovide suggestions for risk reduction strategies, (c) to explore thevariability of experimental material in relation to possible influencingfactors, and (d) to identify geo-referenced background values andtheir association with the Mediterranean fishing areas classified by theUnited Nations Food and Agriculture Organization (FAO). This last ob-jective addressed the scope of theMarine Strategy Framework Directive(MSFD) (Directive 2008/56/EC), setting a framework within which EUMember States shall take thenecessarymeasures to achieve ormaintaingood environmental status in the marine environment by the year2020 at the latest.

2. Materials and methods

2.1. Sample collection and preparation

In order to characterize the contamination of Mediterranean aquaticspecies caught in relevant fishery zones presumably unexposed todirect contamination sources, the following areas were selected(Fig. 1): Southern Adriatic Sea (longitude 17° 20′ 42 E, latitude 41° 00′52 N, 10–40 miles off the coastline before Monopoli port); Ionian Sea(longitude 14° 54′ 625 E, latitude 36° 23′ 135 N, 15 miles off Portopaloport); and southern Tyrrhenian Sea (longitude 15° 46′ 16 E, Latitude38° 17′ 54 N, 15 miles off the coastline, before Bagnara Calabra). Theseareas are all part of FAO Zone 37 (www.fao.org) and correspond to 18,19, and 10 geographical sub-areas, respectively. The selection of sam-pling areas was critical as one of the objectives of the study was to

collect aquatic species in the most relevant fishing areas in terms ofcatches (over 50% of the 17 kton offish caught each year are from south-ern Italy) (IREPA, 2012). These fishing areas are less affected by anthro-pogenic pressures such as densely populated coastal towns andindustrial settlements than those in the northern Adriatic andTyrrhenian basins.

The sampling design covered many of the fish and crustacean spe-cies available on the Italian market (ISMEA, 2007). Twelve wild aquaticspecies, including two crustacean species and two farmed fish species,were chosen for this study (Table 1). The samples from Adriatic, Ionian,and Tyrrhenian seas are identified with the acronyms MO, PP, and BC,respectively (cf. Supplementary Material (SM), Table SM-1). In general,the same species were collected from the aforesaid areas with only fewdifferences. Farmed gilthead sea bream and sea bass specimens wereprovided by COOP Italia consumer cooperative from offshore farmingplants relatively close to the areas where wild species were caught.Specimenswere obtained during 2007–2008.Wild fish and crustaceanswere caught using bottom trawls, bottom trammel nets, and seine netsplaced on the surface and at mid-water.

Sampled fish and crustaceans were stored in tanks with sea waterice; they were transported to the laboratory, where specimens weremeasured and their weight recorded. Fish specimens heavier thanaround 100 g, were skinned, gutted, beheaded, and filleted: filletamounts of equal weight from individuals of the same species (andfrom the same site) were pooled together to obtain a composite sample.For smaller fish species (e.g., anchovy) and crustaceans, compositesamples were made by using the edible portions of individuals of thesame species after removal of the head, tail, and visceral package. Eachcomposite sample was homogenized in a stainless steel jacketed cut-ter/mixer (Stephan UMC5) and aliquots were stored at −20 °C untilanalysis.

2.2. Chemical analysis

The quantification of PCDDs, PCDFs, and non-ortho DL-PCBs (conge-ners 77, 81, 126, and 169) was performed by adapting US EPAMethods1613-B (1994) and 1668-B (2008), using high-resolution gas chroma-tography coupled with high-resolution mass spectrometry (HRGC-HRMS). High-resolution gas chromatography coupled with low-resolutionmass spectrometry (HRGC-LRMS)was used for the followingchemicals: mono-ortho DL-PCBs (congeners 105, 114, 118, 123, 156,157, 167, and 189); 30 NDL-PCBs (congeners 18, 28, 31, 33, 49, 52, 66,70, 74, 91, 95, 99, 101, 110, 128, 138, 141, 146, 149, 151, 153, 170,174, 177, 180, 183, 187, 194, 196, and 203; cumulative analytical con-centration, Σ30NDL-PCBs); 13 PBDEs (congeners 28, 47, 49, 71, 85, 99,100, 153, 154, 183, 197, 206, and 209; cumulative analytical concentra-tion, Σ13PBDEs). The analytical procedure described by Pacini et al.(2013) for the determination of PCDDs, PCDFs, and PCBs infishwas suit-ablymodified for PBDEs. In linewith European regulatory requirements,the cumulative concentrations of PCDDs, PCDFs, and DL-PCBs areexpressed as upper bound (UB) dioxin (2,3,7,8-T4CDD) toxicity equiva-lents (TEQ units). In addition, the parameter Σ7PBDEs is reportedproviding the sum of the seven priority compounds identified by theEuropean Food Safety Authority (EFSA, 2011): congeners 28, 47, 99,100, 153, 154, and 183. Limits of quantification (LOQs) were normallybetter than 0.1 pg/g fw for PCDD, PCDF, and non-ortho DL-PCB conge-ners, and 0.01 ng/g fw for mono-ortho DL-PCB, NDL-PCB, and PBDEcongeners.

The aforementioned eighteenOCPswere determined by an in-housevalidated HRGC-MS/MS method operating in the selective reactionmonitoring (SRM) mode. LOQs were generally better than 0.5 ng/g fwfor each OCP.

PFOA and PFOS were analyzed by liquid chromatography (LC)coupled via an electrospray interface (ESI) operating in the negativeion mode with a triple quadrupole mass spectrometer (MS) as de-scribed by Paiano et al. (2012). The limit of detection (LOD) was ca.

Fig. 1. Areas of the Mediterranean Sea selected for the study and respective reference fishing ports: Southern Adriatic Sea, Monopoli; Ionian Sea, Portopalo; Southern Tyrrhenian Sea,Bagnara Calabra. Fishery areas are also identified according to theUnitedNations Food and Agriculture Organization. (For interpretation of the references to colour in this figure, the readeris referred to the web version of this article.)

Table 1Fish and crustacean species from the Mediterranean Sea selected for the study.

Species Length(cm)

Weight (g) Lipids (%) Trophiclevela

Speciecharacteristics

Feeding habitsa

Bullet tuna (Auxis rochei Risso) 39–41 748–1120 1.2–4.7 4.31 Pelagic, neritic Carnivorous with preferences for fish and cephalopodsEuropean sea bass (Dicentrarchus labrax L.) 30–34 323–412 2.6–5.5 3.8 Farmed –

European anchovy (Engraulis encrasicholus L.) 14–22 23–27 1.3–2.6 3.5 Pelagic, neritic Omnivorous with preferences for animal materialEuropean hake (Merluccius merluccius L.) 20–30 75–280 0.28–1.2 4.17 Demersal Carnivorous with preferences for fish and cephalopodsSurmullet (Mullus surmuletus L.) 15–20 70–108 1.7–10 3.41 Demersal Omnivorous with preferences for animal materialNorway lobster (Nephrops norvegicus L.) 14 55–73 0.60–0.70 2.6b Benthonic Manly detritivorous (www.fao.org)Rose shrimp (Parapenaeus longirostris L.) 12–23 12–120 0.89–1.4 3.31b Benthonic Carnivorous with preferences for polychaetes, small

crustaceans, and gastropods (www.faoadriamed.org)Atlantic bonito (Sarda sarda L.) 45–60 1090–2260 1.8–4.2 4.5 Pelagic, neritic Carnivorous with preferences for fish and cephalopodsAtlantic mackerel (Scomber scombrus L.) 32–33 157–160 3.8–4.2 3.9 Pelagic, neritic Carnivorous with preferences for decapods and fishEuropean pilchard (Sardina pilchardus Walb.) 17 28 3.2–4.7 3.2 Pelagic, neritic Omnivorous with preferences for animal materialGilthead (Sparus aurata L.) 28–32 345–445 2.5–5.5 3.42 Farmed –

Altantic bluefin tuna (Thunnus thynnus L.) – 100,000 2.0–2.6 4.44 Pelagic, oceanic Carnivorous with preferences for fish and cephalopodsAtlantic horse mackerel (Trachurus trachurus L.) 21–31 45–258 1.1–1.9 3.6 Pelagic, neritic Omnivorous with preferences for animal materialSwordfish (Xiphias gladius L.) – 45,000 2.2–3.4 4.49 Pelagic, oceanic Carnivorous with preferences for fish and cephalopods

(a) From Stergiou and Karpouzi (2002) and from http://www.fishbase.org.(b) From http://seaaroundus.org/distribution/LinkGroup.aspx?GroupCode=10&Class=1&LME=1.

20 R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

21R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

0.05 ng/g fw for both compounds. The analytical procedure was inaccordance with the method developed and cross-validated withinthe EU-financed PERFOOD project (PERFOOD, 2009).

MeHg was determined by HRGC-MS/MS after digestion with a 6-Mhydrochloric acid solution according to the analytical procedure report-ed by Ipolyi et al. (2004); HgTOT determination was performed by flow-injection cold-vapor atomic absorption (FI-CV-AAS) using a FIAS 100flow-injection mercury system, after digestion with concentrated nitricacid and hydrogen peroxide. LOQs were respectively ca. 0.03 μg/g fwand better than 0.01 μg/g fw.

Details on analytical procedures are reported in SM.

2.3. Statistical analysis

PCDD+ PCDF, DL-PCB, NDL-PCB, and PBDE data distributions werestatistically analyzed as cumulative analytical and TEQ concentrations,as appropriate, in accordance with performance criteria established byRegulation (EU) 252/2012. For OCPs, PFOA, PFOS, HgTOT, and MeHg,the data distribution of each chemical was studied only when the sub-stance was quantified in at least 50% of samples. This criterion is morestringent than that adopted by EFSA on the basis of WHO guidelines(GEMS/Food-EURO, 1995). In order to distinguish species body burdens,a hierarchical cluster analysis (CA) was adopted; to confirm its findings,the general linearmodels (GLMs) approachwas applied (Garson, 2012).The data of each chemical or chemical family considered in the CAwerelog-transformed; outliers were checked with the Grubbs' test. In theapplication of GLMs, latent variables (“unobservable” attributes) wereutilized. Latent variables are samples characteristicswhich are notmea-suredbut improve data interpretation by reducing thedimensionality ofdata (Borsboom et al., 2003; cf. SM).

3. Results and discussion

3.1. Chemical contamination levels

Table 2 summarizes the statistical descriptors for PCDDs + PCDFs,DL-PCBs, TEQTOT, Σ30NDL-PCBs, Σ6NDL-PCBs, Σ13PBDEs, Σ7PBDEs, 4,4′-DDE, 2,4′-DDT, 4,4′-DDD, PFOS, HgTOT, and MeHg in wild and farmedaquatic species. Data sets were not normally distributed (Shapiro andWilk's W statistics, P b 0.05); however, their log-transformed versionswere on thewhole characterized by P values larger than 0.05: therefore,the hypothesis that the data were from normally distributed popula-tions was not rejected in these cases. Chemical concentrations deter-mined in each sample are presented in Table SM-1. All the remainingcontaminants—aldrin, cis- and trans-chlordane, 2,4′-DDD, 2,4′-DDE,4,4′-DDT, dieldrin, endrin, heptachlor, cis- and trans-heptachlorepoxide, HCB, α-HCH, β-HCH, and γ-HCH, and PFOA—were not report-ed in Tables 2 and SM-1 as not detected or detected only sporadically(non-detected frequency N 50% of samples).

Asmost environmental and food data concerning PCDDs, PCDFs, andDL-PCBs available from the literature are expressed in accordance with1997 WHO TEFs—also adopted for the previous regulatory measures—the data are reported according to both 1997 and 2005 WHO TEFs,respectively, identified as TEQ97 and TEQ05 (Van den Berg et al., 1998,2006).

3.1.1. Polychlorodibenzodioxins (PCDDs), polychlorodibenzofurans(PCDFs), and dioxin-like polychloro-biphenyls (DL-PCBs)

PCDD and PCDF congeners were measured in detectable concentra-tions in the majority of wild and farmed species, with the exception of1,2,3,7,8,9-H6CDF and 1,2,3,4,7,8,9-H7CDF that were never quantified.The prevalent congeners were 2,3,7,8-T4CDF and 2,3,4,7,8-P5CDF,detected in nearly all samples. Important TEQ contributions derivedprimarily from, in descending order, 2,3,4,7,8-P5CDF, 1,2,3,7,8-P5CDD,and 2,3,7,8-T4CDF, on account of their high TEFs. Forwild species, cumu-lative concentrations of PCDDs + PCDFs ranged from 0.0654 to 1.08

pgTEQ05/g fw, with Q.50 and ‹X› estimates, respectively, of 0.157 and0.170 pgTEQ05/g fw. In farmed species, PCDD + PCDF contaminationlevels (0.0296–0.0925 pgTEQ05/g fw) were below the Q.25 valueestimated for wild species; as TEQ97 values, they were an order of mag-nitude below the concentrations determined in the same type of farmedfish recently collected in Turkey (Cakirogullari et al., 2010).

As to DL-PCBs, more than 99% of their cumulative analytical concen-trations in wild and farmed species was due to mono-ortho DL-PCBs118, 105, and 156. In wild and farmed species, DL-PCB concentrationswere found to be in the ranges 0.0267–11.8 and 0.109–0.459pgTEQ05/g fw, respectively; all the concentrations detected in farmedspecies falling below the Q.50 value estimated for wild species (0.535pgTEQ05/g fw).

TEQTOT figures for wild and farmed species were in the ranges0.0922–12.8 and 0.139–0.551 pgTEQ05/g fw, respectively; all theconcentrations measured in farmed species falling below the Q.50

value estimated for wild species (0.700 pgTEQ05/g) and below thevalues determined in farmed fish from Turkey (Cakirogullari et al.,2010).

With reference to Regulation (EU) 1259/2011, no PCDD + PCDFvalues appear to exceed the applicable ML: for wild and farmed species,the observed XMAX values are approximately 3 and 40 times lower thanthe regulatory limit. For TEQTOT, all wild species but one are well belowthe pertinent ML, exceeded only in the case of a bonito sample (12.8pgTEQ05/g fw), whereas all farmed fish exhibit lower contaminationlevels.

3.1.2. Non-dioxin-like polychlorobiphenyls (NDL-PCBs)NDL-PCBs were quantified in almost all samples, the predominant

congeners being (in descending order) 153, 138, 180, and 187. In wildspecies, in 38 out of 40 samples Σ30NDL-PCBs and Σ6NDL-PCBs werefound in the ranges 1.47–96.5 and 0.768–48.3 ng/g fw, respectively.The highest NDL-PCB levels, exceeding the pertinent ML of 75 ng/g fwestablished for Σ6NDL-PCBs, were found in two Atlantic bonito samples(99.9 and 148 ng/g fw, corresponding toΣ30NDL-PCBs values of 207 and413 ng/g fw, respectively). The contamination of farmed fish wasparticularly low, with XMAX values for Σ30NDL-PCBs and Σ6NDL-PCBswell below the corresponding Q.50 values estimated for wild species(13.5 and 6.15 ng/g fw). These findings are indeed quite lower thanthose reported by Cakirogullari et al. (2010) for the same type of farmedfish: together with the other findings of these authors, the influence ofthe feed on the presence of priority contaminants in fish is therebyreinforced.

3.1.3. Polybromodiphenyl ethers (PBDEs)Most of the 13 PBDE congeners were quantified in the samples ana-

lyzed, with the exception of PBDE 71, 85, 183, 197, and 206 for whichquantitative results were obtained in less than 40% of the samples ana-lyzed. PBDE 47 was the most abundant congener, followed by PBDEs100, 209, 49, 99, and 154 in descending priority order. Table 2 showsthe statistical descriptors for the distributions of PBDEs in wild andfarmed aquatic species. For wild species, Σ13PBDEs (Σ7PBDEs) concen-trations ranged from 0.0709 (0.0606) to 3.34 (2.98) ng/g fw, with Q.50

and ‹X› values of 0.434 (0.344) and 0.497 (0.409) ng/g fw, respectively;in farmed fish, concentrations were found to be below the Q.75 value ofthe distribution observed for wild species.

3.1.4. Organochlorine pesticides (OCPs)For wild species, 4,4′-DDE was the predominant component in all

the samples analyzed, with concentrations ranging from less than 0.1to 39.2 ng/g fw and Q.50 and ‹X› of 2.16 and 1.62 ng/g fw, respectively.DDT—as the analytical sum of 2,4′-DDD, 4,4′-DDD, 2,4′-DDE, 4,4′-DDE,2,4′-DDT, and 4,4′-DDT—ranged from 1.00 to 44.7 ng/g fw. These valuesare in agreement with data obtained in other studies on wild aquaticspecies caught in the Mediterranean Sea (Stefanelli et al., 2004;Storelli et al., 2009; UNEP, 2002). As 2,4′- and 4,4′-DDT, the major

Table 2Descriptive statistics of chemical concentrations (upper bound values on freshweight)measured in aquatic species from theMediterranean Sea. Values rounded off to threefigures. Due tothe limited number of samples analyzed, the high percentiles (Q.90 and Q.95) were not estimated, while Q.75 has been reported only for the wild species.

Chemical a Nb Unit XMIN Q.25 Q.50 ‹X› CI95% Q.75 XMAX

Wild speciesPCDDs + PCDFs 40 pgTEQ05/g fw 0.0654 0.104 0.157 0.170 0.137–0.210 0.226 1.08

pgTEQ97/g fw 0.0721 0.127 0.183 0.205 0.163–0.256 0.286 1.48DL-PCBs 40 pgTEQ05/g fw 0.0267 0.298 0.535 0.588 0.395–0.874 1.31 11.8

pgTEQ97/g fw 0.0474 0.370 0.737 0.779 0.529–1.15 1.64 15.0TEQTOT

c 40 pgTEQ05/g fw 0.0922 0.416 0.700 0.796 0.566–1.12 1.54 12.8pgTEQ97/g fw 0.125 0.502 0.933 1.03 0.730–1.44 1.94 16.5

Σ30NDL-PCBs d 40 ng/g fw 1.47 6.66 13.5 14.8 10.0–21.9 33.4 413Σ6NDL-PCBs e 40 ng/g fw 0.768 3.27 6.15 7.32 5.00–10.7 17.5 148Σ13PBDEs f 40 ng/g fw 0.0709 0.211 0.434 0.497 0.355–0.694 1.06 3.34Σ7PBDEs g 40 ng/g fw 0.0606 0.173 0.344 0.409 0.289–0.580 0.889 2.984,4′-DDE 42 ng/g fw b0.1 0.404 2.16 1.62 – h 5.69 39.22,4′-DDT 42 ng/g fw 0.0524 b0.3 b0.3 0.376 – h 0.569 3.264,4′-DDD 42 ng/g fw 0.0608 0.100 0.130 0.241 – h 0.553 3.31PFOS 42 ng/g fw 0.0890 0.405 0.810 0.784 0.584–1.05 1.74 5.96HgTOT 42 μg/g fw 0.0550 0.136 0.223 0.280 0.208–0.376 0.623 2.00MeHg 42 μg/g fw b0.03 0.0967 0.186 0.199 0.141–0.283 0.543 1.93

Farmed speciesPCDDs + PCDFs 10 pgTEQ05/g fw 0.0296 0.0606 0.0699 0.0674 0.0531–0.0855 – 0.0925

pgTEQ97/g fw 0.0317 0.0716 0.0801 0.0790 0.0607–0.103 – 0.114DL-PCBs 10 pgTEQ05/g fw 0.109 0.158 0.245 0.234 0.171–0.318 – 0.459

pgTEQ97/g fw 0.133 0.186 0.304 0.280 0.205–0.381 – 0.545TEQTOT

c 10 pgTEQ05/g fw 0.139 0.222 0.305 0.303 0.228–0.402 – 0.551pgTEQ97/g fw 0.164 0.261 0.376 0.361 0.271–0.481 – 0.659

Σ30NDL-PCBs d 10 ng/g fw 2.59 3.27 5.64 4.92 3.58–6.76 – 9.32Σ6NDL-PCBs e 10 ng/g fw 1.03 1.28 2.15 2.00 1.46–2.73 – 3.67Σ13PBDEs f 10 ng/g fw 0.227 0.273 0.399 0.388 0.291–0.516 – 0.803Σ7PBDEs g 10 ng/g fw 0.162 0.215 0.320 0.303 0.228–0.403 – 0.6054,4′-DDE 10 ng/g fw 0.600 0.600 1.22 1.12 – h – 2.602,4′-DDT 10 i ng/g fw b0.4 – – – – – ≈0.84,4′-DDD 10 i ng/g fw b0.1 – – – – – ≈0.5PFOS 10 i ng/g fw b0.05 – – – – – ≈0.06HgTOT 10 μg/g fw 0.0330 0.0512 0.0580 0.0643 0.0483–0.0855 – 0.135MeHg 10 μg/g fw 0.0220 0.0300 0.0385 0.0394 0.0305–0.0511 – 0.0680

(a) PFOA was characterized by non-detected frequencies of ca. 52% and 100% in wild and farmed species samples, respectively. Wild species sample descriptors (ng/g fw): XMIN b 0.05;X.50 b 0.05; XMAX = 1.89.(b) Number of samples included in statistical evaluation.(c) PCDDs + PCDFs + DL-PCBs.(d) Sum of PCBs 18, 28, 31, 33, 49, 52, 66, 70, 74, 91, 95, 99, 101, 110, 128, 138, 141, 146, 149, 151, 153, 170, 174, 177, 180, 183, 187, 194, 196, and 203.(e) Sum of PCBs 28, 52, 101, 138, 153, and 180.(f) Sum of PBDEs 28, 47, 49, 71, 85, 99, 100, 153, 154, 183, 197, 206, and 209.(g) Sum of PBDEs 28, 47, 99, 100, 153, 154, and 183.(h) Log-transformed data distribution not normal (P b 0.05; cf. text).(i) Non-detected frequency ≥ 60%.

22 R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

constituents of technical DDT mixtures, are metabolized to DDE iso-mers, the ratio of 4,4′-DDE to DDT was used as an indicator of DDT ex-posure in aquatic species: a ratio lower than the critical value of 0.6was considered to be indicative of recent exposure to DDT(Cakirogullari and Secer, 2011; Coelhan et al., 2006; Stefanelli et al.,2004). For those samples inwhichDDT and related compoundswere al-most all measured, the ratio was in the range 0.6–0.9, thus confirmingthat theMediterranean areas under study were apparently not exposedto recent sources of DDT. In farmed fish, DDT analogues were generallyfound at concentrations lower than 1 ng/g fw, with the exception of4,4′-DDE which was present in the range 0.600–2.60 ng/g fw.

3.1.5. Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS)PFOAwas found at concentrations higher than ca. 0.05 ng/g fw only

in a fewwild species, generally living in close contact with the sedimentbed (shrimps and lobsters) and in the neritic zone (such as surmullets,anchovies, and sardines). The concentrations detected in crustaceans(0.175–1.89 ng/g fw) were higher than those found in fish(0.0515–0.209 ng/g fw), suggesting a potential role of sediments intransferring PFOA to the human food chain.

In farmed fish, PFOS concentrationswere always lower than 0.05 ng/gfw. In wild species, PFOS was measured at concentrations ranging from0.0890 ng/g fw to 5.96 ng/g fw, with Q.50 and ‹X› values of 0.810 and

0.784 ng/g fw, respectively. European anchovies and Atlantic mackerelswere the most contaminated species, despite their relatively low tro-phic level (TL), indicating that biomagnification does not occur for thischemical. In line with other reports concerning the MediterraneanSea, the higher PFOS concentrations observed in coastal rather than inpelagic fish may be related to inputs fromwastewater treatment plants(Sánchez-Avila et al., 2010).

3.1.6. Total mercury (HgTOT) and methyl mercury (MeHg)MeHg was measured in all wild species with the exception of some

anchovy samples in which it was present at concentrations below0.03 μg/g fw. The distribution frequency for this contaminant showeda Q.50 value of 0.186 μg/g fw and a concentration range from less than0.03 to 1.93 μg/g fw. As to HgTOT, six samples exceeded the MLs of 0.5or 1.0 μg/g fw but more than 50% of the samples had lower HgTOT con-tents (Q.50 = 0.223 μg/g fw). The high HgTOT findings confirm that theMediterranean Sea is a geological hot spot for mercury (Brambillaet al., 2013). In farmed fish, MeHg and HgTOT concentrations were allbelow the Q.25 values of the corresponding distributions observed forwild species. As known, MeHg is found at significant levels especiallyin predator fish tissue, where it is generally the most abundant Hgform. In agreement with this, the MeHg fraction found in this study incarnivorous species ranged from 53% to 100% of HgTOT content.

0.50

0.00

−0.50

−1.00

−1.50

−2.00

0.50

0.00

−0.50

−1.00

−1.50

Lo

g[M

eHg

]L

og

[MeH

g]

Y = (1.108 ± 0.023) X – (0.73 ± 0.12)

R = 0.979 PR < 0.001F = 2305 PF << 0.001N = 100

Y = (1.116 ± 0.056) X – (0.080 ± 0.033)

R = 0.959 PR < 0.001F = 398 PF << 0.001N = 37

−1.00 −0.75 −0.50 −0.25 0.00 0.25

−1.50 −1.00 −0.50 0.00 0.50

Log[HgTOT]

(b)

(a)

Fig. 2. Log10[MeHg, μg/g fw] vs. log10[HgTOT, μg/g fw]. Linear regression lines are shown to-getherwith their 95% confidence limits (continuous lines) and 95% confidence predictabil-ity boundaries. In (a), data points are exclusively those obtained from the FISR study; in(b), these data points were combined with those from an investigation on the Orbetellolagoon, central Italy (Miniero et al., 2013) (cf. Table SM-2).

23R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

However, the MeHg fraction in omnivorous fish such as European an-chovies averaged around 30%. Because of their feeding habits, the latterspecies incorporate Hg preferentially from zooplankton, in which inor-ganic Hg is the predominant form (da Silva et al., 2011).

In consideration of the high toxicity of MeHg, its correlation with thecorresponding HgTOT content, regulated in fish and fishery products byRegulation (EC) 1881/2006,was of particular interest. A linear regressionstudywas carried out between theMeHg andHgTOT concentrations fromtwodata sets: thefirst consisted of samples from the 10wild aquatic spe-cies analyzed in this study (N= 37; Table 1); the second contained alsodata on wild fish from the Orbetello lagoon, Italy (N = 37 + 63;Table SM-2) (Miniero et al., 2013). MeHg and HgTOT concentrationswere entered only when quantified and [MeHg] × [HgTOT]−1 ≤ 1.3, as-suming that ratiosmuch greater than 1would be unrealistic but allowingfor measurement uncertainties. Fig. 2 shows the correlation lines for thetwodata sets, highly significant in both cases. The difference in predictivepower between the two data sets—quite enhanced when N = 100—isprimarily due to the difference in the number of samples. The aforesaidresults suggest that this type of relationship could be employed in themanagement of MeHg contamination in fish.

3.2. Comparison with PTS data of European fish and fishery products

In linewith Recommendations 2004/705/EC and 2004/704/EC advo-cating a periodic monitoring of PCDDs, PCDFs, and PCBs in Europeanfood and feed, the presence of the chemicals in the products of interestwas recently re-evaluated by EFSA (2012a). The statistical analysis ofEFSA's food category “Muscle meat fish” was carried out on data offish specimens collected in 1995–2010 in 26 European States. PCDD,PCDF, and PCB data shown in Table 2 were comparedwith the distribu-tions observed in Europe for the aforementioned food category and thespecial sub-categories “Other wild caught fish” (not including char, her-ring, salmon, and trout) and “Other farmed fish” (not including salmonand trout) (Table 3). The Q.50 and ‹X› TEQTOT figures reported for wildspecies in Table 2 are 1.4 and 3.1 times lower than the correspondingEuropean upper bound values estimated for “Muscle meat fish,” andare comparable with the corresponding estimates obtained for the spe-cial sub-category “Other wild caught fish”. The Q.50 and ‹X› values forfarmed fish species are remarkably lower.

As to NDL-PCBs (Σ6NDL-PCBs), the ‹X› estimated for wild species inthe present study falls between one half and one third of theEuropean mean contamination levels reported for “Muscle meat fish”and for the sub-category “Other wild caught fish” (Table 3). Althoughthe Q.50 figure reported in Table 2 is higher than the correspondingEuropean figure, most of the findings of this study appear to be compa-rable with European distributions: for instance, the XMAX value of148 ng/g fw is well below the Q.99 estimates of European data (205and 303 ng/g fw, respectively). For farmed fish, the ‹X› calculated forΣ6NDL-PCBs is one seventh of the European value estimated for “Musclemeat fish,” and one fourth when compared with the correspondingvalue estimated for EFSA's sub-category “Other farmed fish”.

The seven congeners selected for Σ7PBDEs, together with PBDE 209,were identified as of primary interest for analytical assessment based onthe composition of marketed technical PBDE mixtures and occurrencein the environment and food (EFSA, 2011). The ‹X› estimates for wildand farmed species, 0.409 and 0.303 ng/g fw, respectively, were approx-imately one fifth, or less, of the mean Σ7PBDEs concentration for “Fishand other seafood” (Table 3) estimated by these authors from EFSA(2011) data.

A direct comparison of the occurrence of PFOA and PFOS in theMed-iterranean species with the database recently published by EFSA (EFSA,2012b) is basically hampered by the different performances of the ana-lytical methods used. In the EFSA's Opinion, only 4% and 33% of PFOAand PFOS results, respectively, were reported to be above the LOD in“Fish meat,” thus not allowing satisfactory statistics. The ‹X› concentra-tion for PFOS in marine fish muscle was reported to be in the order of

2.5 ng/g fw: this figure is quite higher than our average estimates(ca. 0.8 ng/g fw).

3.3. Multivariate analysis of chemical contamination

Fig. 3 exhibits the CA dendrogram obtained by considering the vari-ables PCDDs + PCDFs, TEQTOT, Σ6NDL-PCBs, Σ7PBDEs, 4,4′-DDE, PFOS,and HgTOT. On the whole, the data show a good degree of species-specific clusterization. The samples of the two farmed fish areclusterized in Cluster 1a, below an inter-centroid distance of 2.5. Allhake samples are grouped in Cluster 1c, below a distance of 3.1; thisgroup also includes some surmullet and anchovy samples, fish speciesalso present in Cluster 1b at approximately the same distance. Despitetheir heterogeneity, Clusters 1c and 1b are joined at a distance of 3.9.Additionally, they are joined to Cluster 1a at a distance of 4.2, suggestinga contamination profile possibly comparable to that of farmed fish. Onthe whole, Cluster 2 comprises predatory fish belonging to trophiclevels higher than 4. At a distance of 3.4, the small but fairly homoge-neous Cluster 3 includes only crustaceans. In Clusters 1 and 2 three sam-ples seem to be misclassified: MO B5 (Atlantic bonito) and PP A6(Atlantic mackerel) (Cluster 1), and PP B3 (European pilchard) (Cluster2). The concentrationsmeasured inMO B5 and PP A6 did not test out asoutliers or extreme values: they were not included in Cluster 2 likely

Table 3Summary of statistics for concentrations of selected POPs in fish and fishery products reported by EFSA (2011, 2012a, 2012b). Data, expressed as upper bound (UB) estimates, concernspecimens collected in European countries over the period 1995–2012.a Values rounded off to a maximum of three figures.

Chemical Unit Food categoryb Food subcategoryb Q.50 ‹X› Q.99

TEQTOTc pgTEQ05/g fw “Muscle meat fish” – 0.95 2.50 19.4

“Fish meat, eels excluded/Other wild caught fish” 0.74 1.78 11.8“Fish meat, eels excluded/Other farmed fish” 2.60 6.45 48.6

Σ6NDL-PCBs d ng/g fw “Muscle meat fish” – 3.79 14.8 205“Fish meat, eels excluded/Other wild caught fish” 3.46 19.4 303“Fish meat, eels excluded/Other farmed fish” 6.85 8.27 37.5

Σ7PBDEs e ng/g fw “Fish and other seafood(including amphibians, reptiles, snails, and insects)”

– – 2.14 f –

PFOA ng/g fw “Fish and other seafood” “Fish meat” – 0.64 (0.10) g –

PFOS ng/g fw “Fish and other seafood” “Fish meat” – 2.5 (2.1) g –

(a) Sampling periods: 2001–2009 for PBDEs; 1995–2010 for PCDDs, PCDFs, and PCBs; 2006–2012 for PFOA and PFOS.(b) As classified in EFSA Opinions.(c) PCDDs + PCSDFs + DL-PCBs.(d) Sum of PCBs 28, 52, 101, 138, 153, and 180.(e) Sum of PBDEs 28, 47, 99, 100, 153, 154, and 183.(f) Derived by these authors based on the mean congener concentrations reported by EFSA (2011).(g) In parenthesis the lower bound (LB) estimate.

24 R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

due to the natural variability of data. On the contrary, PP B3 resulted tobe a Cluster 1 outlier (P b 0.05) for several chemicals (e.g., PCDDs +PCDFs, DL-PCBs, NDL-PCBs, and PBDEs) as concentrations fell in generalat the high ends of distributions: the apparently consistent behaviouracross these chemicals suggests that the sample analyzed was obtainedfrom a specimen that had possibly fed on different food sources met

0 2

European sea bass

European sea bass

Gilthead

Gilthead

Bullet tuna

Atlantic bonito

Surmullet

European anchovy

European hake

Atlantic horse mackarel

Bullet tuna

Rose shrimp

Surmullet

European anchovy

European hake

European sea bass

Gilthead

Atlantic bonito

European anchovy

Surmullet

European hake

Atlantic horse mackarelAtlantic bonito

Norway lobster

European anchovy

Surmullet

European hake

European sea bass

European sea bass

Gilthead

Gilthead

Atlantic bonito

Atlantic mackerel

Surmullet

European anchovy

European pilchard

Rose shrimp

Bullet tuna

Atlantic mackerel

European hake

Surmullet

European anchovy

European pilchard

European hake

SwordfishAltantic bluefin tuna

1a 1b

Di

Clusters

Fig. 3.Dendrogramobtained by cluster analysis of the following log-transformed data sets: PCDDs4,4′-DDE (ng/g fw), PFOS (ng/g fw), and HgTOT (μg/g fw) (original concentrations available from

during migrations (www.fao.org). However, in PP B3, the low MeHgconcentration is comparable with the MeHg concentrations measuredin anchovies and other organisms grouped in Cluster 1, a fact possiblyconfirming that the chemicals dealt with bio-magnify in the foodchain following different accumulationmechanisms fromwhich the un-expected positioning in Cluster 2 may derive.

4 6 8 10

2

3

1

1c

stance (arbiratry units)

+ PCDFs (pgTEQ05/g fw), TEQTOT (pgTEQ05/g fw),Σ6NDL-PCBs (ng/g fw),Σ7PBDEs (ng/g fw),Table SM-1). The y-axis defines the inter-centroid distances.

25R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

On the whole, the three clusters described above appear to be wellseparated. Cluster 1 is constituted by a heterogeneous group of speciescharacterized by relatively low contamination levels and comprises spe-cies living in coastal shallow waters, such as the anchovy and the horsemackerel, and/or characterized by demersal behaviour such as the hakeand the mullet (Table 1). In addition, the collected specimens of thesespecies had relatively small sizes (b300 g). The body burden of farmedspecies depends on feed contamination but the farming plants are usu-ally set in relatively shallowwaters. The incomplete subdivision of clus-ters 1b and 1c may be due the low sample number as the respectivefood habits are really different (Table 1). The aforesaid species resultlinked in the CA output because all of them live in waters most influ-enced by sediments and anthropogenic actions. Cluster 2 includeslarge-sized, long-lived, and predatory species such as bonito, bullettuna, Atlantic mackerel, Atlantic bluefin tuna, and swordfish. All thesespecies are strong swimmers and normally travel long distances,which can expose them to very different contamination sources. Cluster3 is formed only by crustaceans, that live and feed on the bottom.

To validate CA results, the measured (length and weight) and latentvariables (all the others, except lipid content) listed in Table 1were test-ed, alone and in combination, by GLMs for their possible influence onchemical data sets. Each sample was grouped according to its nominalfeature (e.g., omnivorous vs. carnivorous) by considering the trendsobserved in the CA outcome: results are summarized in Table 4. The la-tent variables that resulted most significant and predictive are “speciescharacteristics” and “trophic level,” which were entered in GLMsanalysis as “species characteristics” × “trophic level” and “species char-acteristics” × “weight” (Woodward, 1999); single factors such as thesampling area and the food choice were also tested. Sampling areasare not characterized by statistically significant differences: this maybe interpreted as due to the absence of area-specific contaminationsources, that is, the selected sampling areas represent a contaminationbackground. According to GLMs outcome, PCDDs + PCDFs andΣ6NDL-PCBs log-transformed data reflect closely the CA findings; how-ever, log[HgTOT] and log[PFOS] data appear to deviate from such a pat-tern. This likely occurs because of different chemical accumulationmechanisms and a diffuse Hg distribution in the Mediterranean area(UNEP, 2002). In particular, MeHg, themost important chemical speciesinvolved in fish body burden, was shown to accumulate as a MeHg-cysteine complex in muscle tissues of piscivorous fish (Kuwabaraet al., 2007), while PFOS concentrationwas found to be positively corre-lated with the liver protein content in organism tissues or fluids (Hoffet al., 2003). However, despite the wide range of HgTOT demersal con-centrations, the bulk of farmed, demersal, and pelagic/neritic/coastallog[HgTOT] values is relatively narrow and confirms the CA findings.The pelagic/neritic and pelagic/oceanic concentration range is widerthan that discussed above but is not in contrast with the CA findings.On the contrary, HgTOT concentration range in crustaceans is significant-ly different from those observed for log[PCDDs + PCDFs] and log

Table 4Statistical significance of linear regression models (LRMs) and influencing factors.a Each P estim

Item PCDDs + PCDFs TEQTOTb Σ6N

(pgTEQ05/g fw) (pgTEQ05/g fw) (ng

Cluster (1, 2, 3), P 0.0000 (57.7) 0.0000 (52.2) 0.0Species, P 0.0000 (67.7) 0.0000 (65.6) 0.0Species characteristics × trophic level,e P 0.0000 (61.8) 0.0000 (58.4) 0.0Species characteristics × weighte P 0.0007 (32.7) 0.0001 (39.3) 0.0Sampling area, P 0.102 0.375 0.5

(a) Original data expressed on fresh weight. Regressions carried out on log-transformed data s(b) PCDDs + PCDFs + DL-PCBs.(c) Sum of PCBs 28, 52, 101, 138, 153, and 180.(d) Sum of PBDEs 28, 47, 99, 100, 153, 154, and 183.(e) Cf. text.

[Σ6NDL-PCBs]: however, the observation is based on only three samplesand merely suggests that sediments can be a significant Hg exposuresource for these organisms. On the whole, the factors listed in Table 1appear to influence the organism body burden of the chemicals consid-ered. This suggests that the management of chemical contaminationshould follow a fish group-specific strategy such as the one adoptedfor HgTOT (Regulation (EC) 1881/2006).

Several data are available regarding the occurrence of these prioritycontaminants in fish and fishery products but often the comparability isseverely affected by the different strategies adopted in the studies, as inthe following examples: an unidentified origin of samples, such as thosegenerically obtained fromfishmarkets of theMediterranean area (Bocioet al., 2007; Perelló et al., 2012); differences in the species collected (BenAmeur et al., 2013; Koenig et al., 2013; Storelli et al., 2009); differencesin the matrices analyzed (Serrano et al., 2008; Storelli et al., 2009);chemical data reported in various ways with no or scarce possibility toconvert them (Borghesi et al., 2009; Focardi and Renzi, 2012;Harmelin-Vivien et al., 2012). However, in the following examples, a ge-neric comparability between our data and those reported in the litera-ture is feasible. A pool of pilchards collected from fishing harbours ofthe Ebro river delta (Spain) showed a high TEQTOT concentration (2.55pgTEQ05/g fw) (Parera et al., 2013), which is strictly comparable withone of our results (sample PP B3: 2.82 pgTEQ05/g fw). A survey carriedout at the end of the 1990s (Bayarri et al., 2001) throughout the AdriaticItalian coastline appears of interest to compare our findings. In theBayarri et al.'s paper, PCDDs + PCDFs and Σ30NDL-PCBs were detectedin anchovy and red mullet samples at concentration ranges of0.23–0.56 pgTEQ97/g fw and 33.6–132 ng/g fw, respectively. The con-gruent PCDD + PCDF findings of this study (0.120–0.590 pgTEQ97/gfw) appear to include the aforesaid range; however, Σ30NDL-PCBs con-centrations (3.96–32.9 ng/g fw) are lower than the former ones in pos-sible agreement with the downward contamination trend observed forPCBs in the last decade (Szlinder-Richert et al., 2009). In the Bayarriet al.'s paper, European mackerel samples were reported to have con-centrations of PCDDs + PCDFs and Σ30NDL-PCBs, respectively, in theranges of 0.59–1.07 pgTEQ97/g fw and 94.1–177 ng/g fw, both beingcomprised in the corresponding chemical ranges of this study's Cluster2 fish (0.0921–1.48 pgTEQ97/g fw and 7.90–413 ng/g fw, respectively),but above the pertinent Q.50 values of 0.266 pgTEQ97/g fw and29.9 ng/g fw. Although such comparison can be influenced by the limit-ed number of Cluster 2 species, the different sampling strategy adopted,and the time span between the two sampling periods, the aforemen-tioned paper is themost comprehensivework carried out along the Ital-ian coastline. According to another investigation, theΣ6NDL-PCBs levelsin anchovy, Atlantic mackerel, European pilchard, redmullet specimenscollected along the Italian Adriatic coastline (Abruzzi Region) fell in therange 14.9–19.9 ng/g fw (Perugini et al., 2004), which is in the higherend of the Σ6NDL-PCBs range measured in Cluster 1 wild fish(1.44–20.4 ng/g fw). Two factors may explain this observation: in the

ate is associated with the paired df-adjusted R2 value × 100 (in parenthesis).

DL-PCBs c Σ7PBDEs d 4,4′-DDE PFOS HgTOT

/g fw) (ng/g fw) (ng/g fw) (ng/g fw) (μg/g fw)

000 (53.3) 0.0000 (53.4) 0.0000 (57.0) 0.285 0.0005 (26.8)000 (64.3 0.0000 (64.4) 0.0000 (71.4) 0.0000 (78.7) 0.0000 (69.8)000 (60.9) 0.0000 (58.5) 0.0000 (59.3) 0.0000 (81.5) 0.0000 (50.1)000 (45.3) 0.0003 (36.0) 0.0000 (48.1) 0.0000 (73.6) 0.0000 (54.6)05 0.890 0.0831 0.995 0.771

ets.

26 R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

paper referred to, sampling site was close to the coastline and thereforeparticularly influenced by the anthropogenic action, while samplingoccurred 6–7 years before our sampling was carried out.

4. Conclusions

The fish and crustacean samples analyzed in our study appear to de-rive from sampling areas away from direct anthropogenic impact andmay be viewed as reflecting a “normal” or background contaminationstatus: this could represent a valuable information for risk assess-ment/management of the chemicals of interest. Furthermore, the iden-tification of background concentrations in fish and fishery products canallow a follow-up in time trend studies of Mediterranean Sea contami-nation, potentially assisting the implementation of MSFD. It may beadded that while wild fish exhibit on the whole higher contaminationlevels than those measured in farmed fish from the same fishing areas,small size wild fish seem to approach the contamination profile foundin farmed fish, as indicated by the cluster analysis outcome. Only pred-atory fish are at risk of exceeding the existing food regulatorymaximumlevels, an aspect suggesting that management measures for fish andfishery products should be calibrated on a species- or group-specificbasis.

The data summarized in Table 1 can be easily obtained from litera-ture and Internet sources such aswww.fishbase.org, while the chemicalconcentrations estimated in this work are usually determined in the of-ficial food control. Cluster 1 fish appears to be particularly appropriateto meet MSFD criteria for defining the “good ecological status” (GES)of the sea and comply with the aims of MSFD-related Tasks 8 and 9(Law et al., 2010; Swartenbroux et al., 2010). This is a key issue whenconsidering country's territorial waters and their exploitability downto a 200-m depth, as required by the MSFD.

Glossary

CA cluster analysisDDD dichlorodiphenyldichloroethaneDDE dichlorodiphenyldichloroethyleneDDT dichlorodiphenyltrichloroethaneDL-PCB dioxin-like polychlorinated biphenylEC European CommissionEFSA European Food Safety AuthorityESI electrospray interfaceEU European UnionHCB hexachlorobenzeneGLMs general linear modelsHCH hexachlorocyclohexaneHRGC high-resolution gas chromatographyHRMS high-resolution mass spectrometryLOD limit of detectionLOQ limit of quantificationMeHg methyl mercuryML maximum levelMRL maximum residue limitMSFD Marine Strategy Framework DirectiveNDL-PCB non-dioxin-like polychlorinated biphenylOCP organochlorine pesticidePBDE polybrominated diphenyl etherPCDD polychlorinated dibenzo-p-dioxinPCDF polychlorinated dibenzofuranPFOA perfluorooctanoic acidPFOS perfluorooctane sulfonatePOP persistent organic pollutantPTS persistent toxic substanceSRM selective reaction monitoringWHO World Health OrganizationWHO-TEF WHO dioxin (2,3,7,8-T4CDD) toxicity equivalency factor

Acknowledgments

This study was funded by the Italian Ministry for Education,University, and Research, Rome. The Authors are also grateful to Prof.Gennaro Citro, Project Tutor, to Dr. Giuseppe Palma from Federpescafor providing fish during the sampling campaigns, and to Mrs. AnnaMaria Lopomo, Mr. Stefano Pignata, and Dr. Antonella Pilozzi for theirtechnical assistance.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.scitotenv.2014.05.131.

References

Bayarri S, Turrio-Baldassarri L, Iacovella N, Ferrara F, di Domenico A. PCDDs, PCDFs, PCBs,and DDE in edible marine species from the Adriatic sea. Chemosphere 2001;43:601–10.

Ben Ameur W, El Megdiche Y, Eljarrat E, Ben Hassine S, Badreddine B, Souad T, et al.Organochlorine and organobromine compounds in a benthic fish (Solea solea) fromBizerte Lagoon (northern Tunisia): implications for human exposure. EcotoxicolEnviron Saf 2013;88:55–64.

Bocio A, Domingo JL, Falcó G, Llobet JM. Concentrations of PCDD/PCDFs and PCBs in fishand seafood from the Catalan (Spain) market: estimated human intake. Environ Int2007;33:170–5.

Borghesi N, Corsolini S, Leonards P, Brandsma S, de Boer J, Focardi S. Polybrominateddiphenyl ether contamination levels in fish from the Antarctic and theMediterraneanSea. Chemosphere 2009;77:693–8.

Borsboom D, Mellenbergh GJ, van Heerden J. The theoretical status of latent variables.Psychol Rev 2003;110:203–19.

Brambilla G, Abete MC, Binato G, Chiaravalle E, Cossu M, Dellatte E, et al. Mercury occur-rence in Italian seafood from the Mediterranean Sea and possible intake scenarios ofthe Italian coastal population. Regul Toxicol Pharmacol 2013;65:269–77.

Cakirogullari GC, Secer S. Seasonal variation of organochlorine contaminants in bonito(Sarda sarda L. 1758) and anchovy (Engraulis encrasicolus L. 1758) in Black Searegion, Turkey. Chemosphere 2011;85:1713–8.

Cakirogullari GC, Kılıç D, Uçar Y. Levels of polychlorinated dibenzo-p-dioxins, dibenzo-p-furans and polychlorinated biphenyls in farmed sea bass (Dicentrarchus labrax) andsea bream (Sparus aurata) from Turkey. Food Control 2010;21:1245–9.

Coelhan M, Strohmeier J, Barlas H. Organochlorine levels in edible fish from the Marmarasea, Turkey. Environ Int 2006;32:775–80.

da Silva CA, Tessier E, Kütter VY, Wasserman JC, Donard OFX, Silva-Filho EV. Mercuryspeciation in fish of the Cabo Frio upwelling region, SE Brazil. Braz J Oceanogr2011;59:259–66.

EFSA. Scientific report of EFSA—Polybrominated Diphenyl Ethers (PBDEs) in Food. EFSA J2011;9:2156. [http://www.efsa.europa.eu/it/efsajournal/pub/2156.htm].

EFSA. Scientific report of EFSA—Update of the monitoring of levels of dioxins and PCBs infood and feed. EFSA J 2012a;10:2832. [http://www.efsa.europa.eu/it/efsajournal/pub/2832.htm].

EFSA. Scientific report of EFSA—Perfluoroalkylated substances in food: occurrence and di-etary exposure. EFSA J 2012b;10:2743. [http://www.efsa.europa.eu/it/efsajournal/pub/2743.htm].

Focardi S, Renzi M. Polybrominated Diphenyl Ethers and Polychlorobiphenyls in Fish fromthe Ionian Sea (Western Mediterranean). J Environ Prot 2012;3:135–40.

Garson GD. General linear models: multivariate GLM and MANOVA/MANCOVA.Asheboro: Statistical Associates Publishers; 2012.

GEMS/Food-EURO. Secondworkshop on reliable evaluation of low-level contamination offood. EUR/ICP/EHAZ.94.12/WS04 Report on aWorkshop in the Frame of GEMS/Food-EURO. Kulmbach; 1995. [http://www.who.int/foodsafety/publications/chem/en/lowlevel_may1995.pdf].

Harmelin-Vivien M, Bodiguel X, Charmasson S, Loizeau V, Mellon-Duval C, Tronczyński J,et al. Differential biomagnification of PCB, PBDE, Hg and Radiocesium in the food webof the European hake from the NW Mediterranean. Mar Pollut Bull 2012;64:974–83.

Hoff PT, Van de Vijver K, Van Dongen W, Esmans EL, Blust R, De Coen WM.Perfluorooctane sulfonic acid in bib (Trisopterus luscus) and plaice (Pleuronectesplatessa) from the Western Scheldt and the Belgian North sea: distribution and bio-chemical effects. Environ Toxicol Chem 2003;22:608–14.

INRAN. National food consumption database. Rome: Istituto Nazionale per la RicercaAlimentare e la Nutrizione; 2012. http://nut.entecra.it/710/I_consumi_alimentari__INRAN-SCAI_2005-06.html.

Ipolyi I, Massanisso P, Sposato S, Fodor P, Morabito R. Concentration levels of total andmethyl mercury in mussel samples collected along the coasts of Sardinia island(Italy). Anal Chim Acta 2004;505:145–51.

IREPA. Catches from fisheries belonging to the Italian coastal Regions (in Italian). Salerno:Istituto di Ricerche Economiche per la Pesca e l'Acquacoltura; 2012.

ISMEA. Il settore ittico in italia e nel mondo: le tendenze recenti. Rome: Istituto di Serviziper il Mercato Agricolo Alimentare; 2007.

Koenig S, Huertas D, Fernández P. Legacy and emergent persistent organic pollutants(POPs) in NW Mediterranean deep-sea organisms. Sci Total Environ 2013;443:358–66.

27R. Miniero et al. / Science of the Total Environment 494–495 (2014) 18–27

Kuwabara J, Arai Y, Topping BR, Pickering IJ, Geroge GN.Mercury speciation in piscivorousfish from mining-impacted reservoirs. Environ Sci Technol 2007;41:2745–9.

Law R, Hanke G, Angelidis M, Batty J, Bignert A, Dachs J, et al. Marine Strategy FrameworkDirective - Task Group 8: contaminants and pollution effects. EUR 24339 EN Scientificand Technical Reports. Brussels: Joint Research Centre; 2010.

Miniero R, Beccaloni E, Carere M, Ubaldi A, Mancini L, Marchegiani S, et al. Mercury (Hg)and methyl mercury (MeHg) concentrations in fish from the coastal lagoon ofOrbetello, central Italy. Mar Pollut Bull 2013;76:365–9.

Pacini N, Abate V, Brambilla G, De Felip E, De Filippis SP, De Luca S, et al. Polychlorinateddibenzodioxins, dibenzofurans, and biphenyls in fresh water fish from CampaniaRegion, southern Italy. Chemosphere 2013;90:80–8.

Paiano V, Fattore E, Carrà A, Generoso C, Fanelli R, Bagnati R. Liquid chromatography-tandem mass spectrometry analysis of perfluorooctane sulfonate andperfluorooctanoic acid in fish fillet samples. J Anal Methods Chem 2012;2012:719010.

Parera J, Ábalos M, Santos FJ, Galceran MT, Abad E. Polychlorinated dibenzo-p-dioxins, di-benzofurans, biphenyls, paraffins and polybrominated diphenyl ethers in marine fishspecies from Ebro River Delta (Spain). Chemosphere 2013;93:499–505.

Perelló G, Gómez-Catalán J, Castell V, Llobet JM, Domingo JL. Assessment of the temporaltrend of the dietary exposure to PCDD/Fs and PCBs in Catalonia, over Spain: healthrisks. Food Chem Toxicol 2012;50:399–408.

PERFOODhttp://www.perfood.eu, 2009.Perugini M, Cavaliere M, Giammarino A, Mazzone P, Olivieri V, Amorena M. Levels of

polychlorinated biphenyls and organochlorine pesticides in some edible marineorganisms from the central Adriatic sea. Chemosphere 2004;57:391–400.

Sánchez-Avila J,Meyer J, Lacorte S. Spatial distribution and sources of perfluorochemicals in theNW Mediterranean coastal waters (Catalonia, Spain). Environ Pollut 2010;158:2833–40.

Serrano R, Barreda M, Blanes MA. Investigating the presence of organochlorine pesticidesand polychlorinated biphenyls inwild and farmed gilthead sea bream (Sparus aurata)from the western Mediterranean Sea. Mar Pollut Bull 2008;56:963–72.

Stefanelli P, Ausili A, Di Muccio A, Fossi C, Di Muccio S, Rossi S, et al. Organochlorine com-pounds in tissues of swordfish (Xiphias gladius) from Mediterranean Sea and Azoresislands. Mar Pollut Bull 2004;49:938–50.

Stergiou KI, Karpouzi VS. Feeding habits and trophic levels of Mediterranean fish. Rev FishBiol Fish 2002;11:217–54.

Storelli MM, Losada S, Marcotrigiano GO, Roosens L, Barone G, Neels H, et al.Polychlorinated biphenyl and organochlorine pesticide contamination signa-tures in deep-sea fish from the Mediterranean Sea. Environ Res 2009;109:851–6.

Swartenbroux F, Albajedo B, Angelidis M, Aulne M, Bartkevics V, Besada V, et al. MarineStrategy Framework Directive - Task Group 9: contaminants in fish and other sea-food. EUR 24339 EN. Scientific and Technical Reports. Brussels: Joint Research Centre;2010 [http://publications.jrc.ec.europa.eu/repository/bitstream/111111111/13669/1/tg9%20report%20final_vii.pdf].

Szlinder-Richert J, Barska I, Mazerski J, Usydus Z. PCBs in fish from the southern Baltic sea:levels, bioaccumulation features, and temporal trends during the period from 1997 to2006. Mar Pollut Bull 2009;58:85–92.

UNEP. Mediterranean regional report. Regionally Based Assessment of Persistent ToxicSubstances. Geneva: United Nations Environment Programme “Chemicals”; 2002[http://www.chem.unep.ch/Pts/regreports/Mediterranean.pdf].

Van den Berg M, Birnbaum L, Bosveld ATC, Brunström B, Cook P, Feeley M, et al. Toxicequivalency factors (TEFs) for PCBs, PCDDs, and PCDFs for humans and wildlife.Environ Health Perspect 1998;106:775–92.

Van den Berg M, Birnbaum L, Denison M, De Vito M, Farland W, Feeley M, et al. The2005 World Health Organization re-evaluation of human and mammalian toxicequivalency factors (TEFs) for dioxins and dioxin-like compounds. Toxicol Sci2006;93:223–41.

WoodwardM. Epidemiology: study design and data interpretation. Boca Raton: Chapmanand Hall/CRC; 1999.