PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by:On: 11 September 2008Access details: Access Details: Free AccessPublisher Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Food Additives and Contaminants: Part BPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t783462596

Levels of polychlorinated biphenyls in fish and shellfish from the Adriatic SeaG. Sagratini a; M. Buccioni a; C. Ciccarelli b; P. Conti a; G. Cristalli a; D. Giardinà a; C. Lambertucci a; G.Marucci a; R. Volpini a; S. Vittori a

a Scienze Chimiche, University of Camerino, Camerino, Italy b ASUR Marche ZT 12, Veterinary Service, SanBenedetto del Tronto, Italy

Online Publication Date: 01 June 2008

To cite this Article Sagratini, G., Buccioni, M., Ciccarelli, C., Conti, P., Cristalli, G., Giardinà, D., Lambertucci, C., Marucci, G., Volpini,R. and Vittori, S.(2008)'Levels of polychlorinated biphenyls in fish and shellfish from the Adriatic Sea',Food Additives andContaminants: Part B,1:1,69 — 77

To link to this Article: DOI: 10.1080/19393210802236919

URL: http://dx.doi.org/10.1080/19393210802236919

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

Food Additives and Contaminants: Part BVol. 1, No. 1, July 2008, 69–77

Levels of polychlorinated biphenyls in fish and shellfish from the Adriatic Sea

G. Sagratinia, M. Buccionia, C. Ciccarellib, P. Contia, G. Cristallia, D. Giardinaa,C. Lambertuccia, G. Maruccia, R. Volpinia and S. Vittoria*

aScienze Chimiche, University of Camerino, Camerino, Italy; bASUR Marche ZT 12, Veterinary Service,San Benedetto del Tronto, Italy

(Received 8 August 2007; final version received 21 November 2007)

Levels of 18 polychlorinated biphenyl (PCB) congeners were determined by gas chromatography-massspectrometry (GC-MS) in some marine species, living both in the coastal area and in deeper seawater.In some species analysis was performed separately in edible parts (fillets) and in viscera. The existence and degreeof bioaccumulation was assessed studying individual species of very different size, with the smaller being younger.Furthermore, with a multivariate statistical analysis, a correlation between PCB congeners and the feeding habitsand habitat of the fish was demonstrated. The results show that fat from edible parts (fish fillets) had total PCBlevels in the range 22.6–601.9 mg kg�1 (with 601.9 mg kg�1 in anchovies), while fat from viscera showed muchhigher concentrations (407.3–916.6 mg kg�1). Bioaccumulation was confirmed, comparing PCB levels betweenyounger and older individual hake, squid, and horned octopus. The total PCB concentration ratio (older/youngerindividuals) ranges from 2.11 (squid¼ 292.1/137.8 mg kg�1) to 3.46 (hake¼ 546.0/158.0 mg kg�1).

Keywords: polychlorinated biphenyls; Adriatic Sea; fish pollution; bioaccumulation; gas chromatography-massspectrometry (GC-MS); multivariate analysis

Introduction

San Benedetto del Tronto, a renowned tourist city onthe shores of the central Adriatic Sea, is also a verywell-known fishery centre. Its fish market is one of themost important in Italy, trading between 5000and 6000 tonnes of fish per year in the last few years(http://www.mercatoitticosbt.it). The San Benedettofish market was established in 1886 by theMunicipality of the city. A new building wasconstructed in 1936; and restored in 1999. Fish soldin the San Benedetto market come mostly from thecentral Adriatic, with the highest volume in traderepresented by far by anchovies (an average of 4000tons, about 75% of total trade volume in weight),followed by hake, red mullet, and horned octopus.

Polychlorinated biphenyls (PCBs) are a class of 209synthetic compounds (Erickson 1997), with a commonstructure represented by a biphenyl rings, with the tenhydrogen atoms variously substituted with chlorines(Schulz et al. 1989). They are used in a number ofapplications, with a high level of emission into theenvironment (Breivik et al. 2002), where PCBs arestrongly persistent. Because of those facts, thosepollutants cause bioaccumulation and biomagnifica-tion, as already reported (Olsson et al. 2000;Andersson et al. 2001; Smith and Gangolli 2002),with potential high level in edible fish.

As regards human toxicity, the World HealthOrganization’s (WHO) International Agency for

Research on Cancer (IARC) considers PCBs to be‘probable human carcinogens’, and a number of papers

have shown that PCBs are responsible for skinillnesses, cancer, and developmental and neurologicalproblems. PCB exposure occurs mainly through the

diet, in which animal fat, fish, and shellfish representalmost the total source (Asplund et al. 1994;

Zuccato et al. 1999).The Adriatic Sea is a relatively isolated sea area,

with a low degree of water exchange with other areas,into which most of the Italian rivers flow, transferring

pollutants picked up along their course. For example,the Po River flows into the northern part of the

Adriatic Sea, and takes polluted water from the mostintensely cultivated area of Italy, also flowing through

the most industrialized areas, and bringing to the seawaste water from some of the most populated Italiancities (Turin and Milan, for example).

The scope of this research was to determine the

level of a number of PCBs (18 different molecules,i.e. PCB 28, 52, 95, 99, 101, 105, 110, 118, 138, 146,

149, 151, 153, 170, 177, 180, 183, 187) in some marineorganisms by sampling species that both live close tothe coast and are fished in deeper seawater. The 18

selected PCBs are the most studied because of their

*Corresponding author. Email: sauro.vittori@unicam.it

ISSN 1939–3210 print/ISSN 1939–3229 online

� 2008 Taylor & Francis

DOI: 10.1080/19393210802236919

http://www.informaworld.com

Downloaded At: 17:47 11 September 2008

capability to accumulate in lipophilic matrices

(Di Muccio et al. 2002); moreover, those compoundsare the same congeners that governmental regional

agencies are routinely monitoring in fish fromthe Adriatic. Some species analysis has beencarried out on edible parts and in viscera, in young

and old individuals, to assess the importance ofthose different parts of the body and age in terms

of pollution level. Furthermore, analysis of differentlyaged individuals provided information on bioaccumu-lation phenomena. In the literature there are other data

on fish pollution in the Adriatic Sea (Bayarri et al.2001; Di Muccio et al. 2002; Perugini et al. 2004;

Stefanelli et al., 2004), but studies on bioaccumulationand on fillets compared with viscera have not beenreported.

Materials and methods

Sampling

Analyses were carried out on eight different marine

species: anchovy, red mullet, sole, hake (two differentsizes), cuttlefish, squid (two different sizes, fillet andviscera), horned octopus (two different sizes, fillet and

viscera), and spottail mantis shrimp. The organismswere collected in April 2006 in the fish market of SanBenedetto del Tronto and from small fisherman’s

associations in Porto San Giorgio, along the MarchesItalian coast, in the central Adriatic Sea (Figure 1).

Collected samples were stored at –80�C until samplepreparation for analyses. Each species was treatedseparately. Weights and lengths were measured

and recorded (Table 1). Marine organisms werewashed with distilled water and, in reported cases,

edible parts were separated from viscera (analysedfor two fish species) to obtain the edible portionsaccording to normal house preparation. The samples

were homogenized in a Waring Commercial Blender(Torrington, CT, USA).

Chemicals and certified reference material

The PCB mix standard at a concentration of 10 ng ml�1

in iso-octane was supplied by Dr. Ehrenstorfer

(Ausburg, Germany). Standard working solutionsat various concentrations were prepared daily byappropriate dilution of the stock solutions with

iso-octane. Acetone, hexane and iso-octane for residueanalysis PESTANAL� were purchased from Riedel-de

Haen (Seelze, Germany). Sulfuric acid 96% wassupplied from Carlo Erba (Milan, Italy). The solid-phase extraction (SPE) columns were Discovery� SPE

DSC-Si Silica Tube (6ml/1 g) and purchased fromSupelco (Bellefonte, PA, USA), while the Extrelut�

NT 3 cartridge (1–3ml sample) were purchased from

Merck (Darmstadt, Germany). Hydromatrix waspurchased from VWR International (Milan, Italy).

Sample preparation

The method used for PCB analysis in fish was modifiedin the laboratory from that reported by Di Muccioet al. (2002). Sample (250 g) was homogenized for3–5min, and to 100 g of the homogenized variableamounts of Hydromatrix (Varian, Palo Alto, CA,USA) were added depending of the matrix hydration.It was then placed in a ventilated stove at 45–50�C for20 h. The dried samples were extracted in a Soxhletapparatus using 700ml of a mixture of hexane/acetone1/1 for 8 h. Extracts were evaporated under vacuum at40�C obtaining a variable quantity of fat residue thatwas weighted (the percentage of fat over the total bodyweight is reported in Table 2). Analytes were separatedfrom the lipids by a combination of the Extrelut andthe silica gel cartridges, respectively. Previously, theExtrelut cartridge was acidified adding 3ml ofsulphuric acid and, at the same time, the silica gelcartridge was conditioned by applying 2� 2ml of themixture solvent; at this point the system Extrelut-SPEwas set up and placed on a vacuum station (Baker,SPE-12G System, Deventer, the Netherlands). Bothcartridges were washed with 10ml of the solventmixture at a flow rate of 2mlmin�1, then 0.70 g offat, dissolved in 2� 1.5ml of hexane, were applied tothe Extrelut cartridge and maintained in contact withsulphuric acid for 10min. After that, the elution wasperformed with 17ml of hexane at a flow rate nothigher than 0.5mlmin�1; the eluate was collected in avial sample, passed in a 50-ml round-bottom flask,and evaporated to dryness in rotary evaporator setat 40�C. Finally, 0.5ml of iso-octane were added tothe extract and the resulting solution was injected inthe GC-MS.

Gas chromatographic analysis

A gas chromatograph/mass selective detector (GC/MSD) (Hewlett Packard, Palo Alto, CA, USA;HP-6890 with HP 5973) was used. Separation wasperformed on an HP 5 MSI column(30m� 0.25mm� 0.25mm film thickness). AnHPChem workstation was used with the GC/MSsystem. All injections were splitless and the volumewas 2 ml. The flow rate (He) was 0.8mlmin�1. Theinjector temperature was 270�C. The column tempera-ture programme was from 60�C (3min) to 190�C at8�Cmin�1, from190�C(18min) to 300�Cat 15�Cmin�1,then at 300�C for 10min. Data were acquired in theelectron impact (EI) mode (70 eV) using the selected ionmonitoring (SIM) mode. SIM ions and time conditionsfor each PCB congener was reported in Table 3.

70 G. Sagratini et al.

Downloaded At: 17:47 11 September 2008

Figure 1. The Adriatic Sea.

Table 1. Composition of pooled samples.

SpeciesNumber ofanimals

Length (cm),mean�RSD

Weight (g),mean�RSD

Anchovy (Anchovy) 82 14.1� 2 21.73� 7.8Red mullet (Mullus barbatus) 10 18.8� 1.9 62.2� 18.3Sole (Solea vulgaris) 9 22.7� 1.6 97.1� 2.1Hake (Merluccius), big 6 51.3� 2.8 1293.3� 459.6Hake (Merluccius), little 13 37.8� 3.1 425.7� 80.5Spottail mantis shrimp (Squilla mantis) 22 16.2� 2.4 26.7� 0.7Cuttlefish (Sepia oficinalis) 6 12.8� 1.5 250� 85Squid (Todarodes sagittatus), big 10 n.d.1 221� 11.0Squid (Todarodes sagittatus), little 10 n.d.1 92� 6.4Horned octopus (Eledone cirrosa), big 10 n.d.1 243� 42Horned octopus (Eledone cirrosa), little 10 n.d.1 82� 14.5

Note: 1In the case of squid and horned octopus, the length of animals was not determined because of thevery different lengths between animals caused by broken tentacles. n.d., Not determined.RSD, relative standard deviation.

Food Additives and Contaminants: Part B 71

Downloaded At: 17:47 11 September 2008

Table

2.Polychlorinatedbiphenyl(PCB)congener

concentration(mgkg�1)measuredin

theAdriaticfish

speciesunder

analysis.

Inallspecies,PCB

quantificationrefers

tofat

from

edible

parts;in

thehorned

octopusandin

squid

PCBswerealsomeasuredin

fatfrom

viscera.

CodeofPCB

congener

Anchovy

Red

mullet

Sole

Spottailmantis

shrimp

Big

hake

Little

hake

Big

squid

Little

squid

Big

horned

octopus

Littlehorned

octopus

Cuttlefish

Squid

viscera

Horned

octopus

viscera

Percentagefat

1.5

1.75

1.5

2.1

1.1

11.5

1.6

22.2

2.9

5.1

4.0

PCB

28

13.3

1.1

1.6

1.4

8.1

4.0

5.2

5.0

1.9

2.0

6.8

15.0

8.5

PCB

52

18.7

0.7

1.7

1.4

8.0

3.0

3.8

4.0

2.9

1.0

3.4

15.7

2.3

PCB

95

25.5

1.0

3.0

1.7

16.5

5.0

8.3

4.7

0.0

1.0

3.9

16.0

3.8

PCB

99

19.2

1.0

3.5

1.4

21.9

8.0

11.5

5.8

2.9

1.0

3.3

25.0

18.0

PCB

101

32.4

0.7

3.9

1.1

25.9

8.0

11.8

5.9

3.0

1.0

3.4

45.0

8.1

PCB

105

12.9

1.0

6.7

1.1

26.2

6.0

23.1

10.5

2.8

2.0

6.0

19.0

11.0

PCB

110

23.9

1.0

3.9

1.2

22.0

8.0

14.2

6.7

2.4

1.0

2.7

27.0

5.6

PCB

118

28.6

1.1

4.5

1.8

28.1

12.0

17.1

5.0

2.7

2.0

5.7

42.0

24.0

PCB

138

86.5

3.1

6.7

2.1

60.5

23.0

21.2

6.7

4.2

2.0

4.6

146.0

80.0

PCB

146

25.1

1.0

2.5

0.7

24.1

6.0

13.9

9.0

4.0

1.0

4.3

36.0

14.0

PCB

149

51.7

1.1

3.2

0.7

35.7

11.0

14.3

6.6

3.9

1.0

4.3

71.0

14.1

PCB

151

18.6

0.8

1.8

0.7

18.1

5.0

10.3

6.4

0.0

1.0

4.0

31.9

4.9

PCB

153

108.9

3.9

7.8

2.3

82.1

29.0

18.2

8.1

4.1

2.0

8.7

194.0

109.0

PCB

170

19.5

1.0

4.5

0.9

41.5

5.0

34.0

15.6

5.1

2.0

11.1

39.0

20.0

PCB

177

17.4

1.0

3.0

0.9

26.8

3.0

23.2

11.9

5.0

1.0

5.6

24.0

7.0

PCB

180

41.0

2.1

4.2

1.2

41.5

10.0

26.9

12.5

4.4

2.0

9.7

83.0

39.0

PCB

183

17.4

1.0

2.5

0.8

21.4

3.0

17.6

8.1

4.5

1.0

4.9

20.0

10.0

PCB

187

41.1

2.2

4.1

1.2

37.6

9.0

17.6

5.4

4.6

1.0

5.0

67.0

28.0

�PCB

601.9

24.9

69.1

22.6

546.0

158.0

292.1

137.8

58.2

25.0

97.5

916.6

407.3

72 G. Sagratini et al.

Downloaded At: 17:47 11 September 2008

Quality control

The calibration curves were carried out at fiveconcentration levels using suitably diluted standards(working solution). The coefficient of correlation wasin the range 0.997–0.998 and the linearity range wasfrom 1 to 500 mg kg�1 for all studied congeners.Method reproducibility studies were performed inject-ing the five replicates of the same standard solution inthree different days and in the same day. The intradayprecision showed a relative standard deviation(RSD)that in all cases was lower than 7%, while the interdayprecision showed an RSD that in all cases was lowerthan 13%. The recovery percentages of PCBs wereinvestigated by spiking with 50 ml of standard PCBsmixtures the fish matrix before Soxhlet extraction, fora final concentration level of 100mg kg�1. Meanrecoveries of the 18 target compounds ranged from80 to 102% with n¼ 14 and RSDs <15%.

Limit of detection (LOD) and limit of quantifica-tion (LOQ) were estimated on the basis of 3 : 1 and10 : 1 signal-to-noise ratios obtained with standardscontaining the compounds of interest at low concen-tration levels. The LOD was 0.05 mg kg�1, and theLOQ was 0.3 mg kg�1. As an example, in Figure 2 GC-MS chromatograms are shown for (A) a PCBs mixtureof standard at 0.1mg kg�1 and (B) an octopus sample.

Results and discussion

Concentrations of PCBs, expressed in mg kg�1, in fatextracted from the edible part and viscera of marineorganisms of the Adriatic Sea are shown in Table 2.Analysing the whole data shown, it is evident that themost contaminated species was anchovy, in agreementwith other studies on PCB levels in marine organismsfrom the Adriatic (Bayarri et al. 2001; Di Muccioet al. 2002; Perugini et al. 2004; Stefanelli et al., 2004).Interesting to note is that the pollution level decreasedin the last few years; in fact, sampling made in 2002(Perugini et al. 2004) reported �PCB levels foranchovy, red mullet, squid and cuttlefish to be higherthan those found in the 2006 samples (1100 versus 602,430 versus 25, 320 versus 220, and 180 versus 98,respectively). This is very likely due to the Europeanban from market of PCB since 1996, and also to anincreasing concern and attention to the environment.

Furthermore, from data in Table 2, it is unequi-vocal that for fish of different sizes bioaccumulation isconfirmed. In fact, in the larger sized hakes, a level ofcontamination, expressed as �PCB (546 mg kg�1), wasfound that was 3.5 times higher than that found in thesmaller size hakes (158 mg kg�1). The same bioaccumu-lation phenomenon was found in squid and hornedoctopus showing a concentration ratio of about 2.Data referred to the viscera analysis were particularlyinteresting, showing a level of �PCB that is from fourtimes (squid) to ten times (horned octopus) higher thanthat found in the edible part of both marine species.Levels of low-chlorinated PCB congeners (tri-Cl andtetra-Cl) are quite low, whereas penta-, hexa-, andhepta-Cl derivatives are predominant as a consequenceof the higher ability of polychlorinated molecules tobioaccumulate in lipophilic matrices, with respect toless chlorinated ones. According to data reported byStefanelli et al. (2004), we found that PCB havingchlorine atoms at 2-, 4-, or 5-positions in one or bothrings, namely PCB 118, PCB 153, and PCB 180,showed in general the major bioaccumulation levels,confirming studies about bioaccumulation mechanism(Bright et al. 1995). The congener that in all matricesshowed the maximum contamination level is PCB 153,which in anchovy fat reached a concentration level of108.9mg kg�1.

Figure 3 shows the contribution of the selectedcongeners to the total of PCB measured in eachspecies. The contribution of ðci=

Pi ciÞ � 100 the i-th

congener is estimated as in which Ci is the concentra-tion of the i-th congener.

This normalization provides evidence thatActinopterygii (anchovy, red mullet, hake, sole) andMalacostraca (spottail mantis shrimp) species have anaccumulation profile of the PCB congeners that is verysimilar, in which PCB 138 and PCB 153 are the mostimportant. The cephalopod (horned octopus, squidand cuttlefish) species show a content of PCBcongeners distributed in a similar manner as that ofthe other analysed species, except for the fact that inthis group PCB 138 and PCB 153 are less importantthan PCB 170 and PCB 180, which are the two mostimportant here. This result is in agreement with astudy on PCB distribution on European soils(Skrbic and Durisic-Mladenovic 2007); some of thesemain congeners have also been studied in eels

Table 3. Selected ion monitoring (SIM) conditions for gas chromatography-mass spectrometry (GC-MS) analysis.

Congeners SIM ion (m/z) Time windows (min)

PCB 28, PCB 52 220, 292, 256, 258 20.14–21.94PCB 95, PCB 101, PCB 99, PCB 110 326, 328 26.52–32.96PCB 151, PCB 149, PCB 118, PCB 146, PCB 153, PCB 105, PCB 138 360, 290, 326, 328 34.61–38.21PCB 187, PCB 183, PCB 177, PCB 180, PCB 170 394, 396 38.75–40.66

Food Additives and Contaminants: Part B 73

Downloaded At: 17:47 11 September 2008

(Mariottini et al. 2006) and sharks (Storelli et al. 2005),and were associated with the human activities andhabitat of the species. With the exception previouslydescribed, the PCB congener’s profiles were similar inall species. It must be pointed out that the cephalopodsviscera have a PCB profile very similar to those ofother fish classes and differ from that of the edible partof the cephalopods. As each species lives in a differenthabitat and has its own metabolism, the previousdifferences could be attributed to one or both of thesecauses. Comparing the PCB distribution in viscera andin the edible parts of cephalopods, it is hypothesized

that the PCB congener profiles are similarly absorbedby all species, while the intake level (in cephalopods itis more convenient to look at the viscera results) ofPCBs is related to their presence in the environmentand in the feeding animals that all investigated speciesshare. The different profile of PCB in the cephalopodsspecies, however, indicates a different metabolic pathin cephalopods (see the edible part profile) so to havea different biomagnification of the congeners withrespect to the other fishes. In this connection, the datawere analysed with multivariate methods in order toobtain clearer information.

500

Abundance

36.86

37.2638.20

38.89

32.97

38.21

20.00 22.00 24.00 26.00 28.00 30.00 32.00 34.00 36.00 38.00 40.00

20.00 22.00 24.00 26.00 28.00 30.00 32.00 34.00 36.00 38.00 40.00

100

150

200

250

300

350

400

450

20.14

21.94

28.75 29.29

35.73

35.8836.88

37.16

37.28

38.7538.90

39.52

40.05

40.66

26.52

34.65

min

min

B

A

18.00

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

Abundance

20.13

21.92

26.5028.71

29.2532.91 34.60

35.71

35.85

37.14

38.74

39.51

40.0540.65

Figure 2. GC-MS chromatograms of (A) PCBs mix standard at a concentration of 100 mg kg�1 and (B) an octopus sample.The following are retention times for each of the 18 PCB congeners: Rt¼ 20.13 for PCB 28, 21.92 for PCB 52, 26.50 for PCB 95,28.71 for PCB 101, 29.25 for PCB 99, 32.91 for PCB 110, 34.60 for PCB 151, 35.71 for PCB 149, 35.85 for PCB 118, 36.86 forPCB 146, 37.14 for PCB 153, 37.26 for PCB 105, 38.20 for PCB 138, 38.74 for PCB 187, 38.89 for PCB 183, 39.51 for PCB 177,40.05 for PCB 180, and 40.65 for PCB 170.

74 G. Sagratini et al.

Downloaded At: 17:47 11 September 2008

PCBs are 209 congeners that can be divided in twogroups on the basis of their toxicological properties(TEF, toxicological equivalent factor): a smallnumber, named dioxin-like PCBs, shows toxicologicalproperties similar to those of dioxins, while most havea different toxicological profile to that of dioxins.According to Directive (EC) No. 199/2006 thatmodifies the precedent Directive (EC) No. 466/2001,the European Community has established themaximum level admissible in fishery products fordioxin-like PCBs, furans and dioxins. In the list ofdioxin-like PCBs, there are only two congeners ofthose 18 that the present authors have monitored: PCB105 and PCB 118. From the data, it can be seen thatthe contamination level for dioxin-like PCBs (PCB 105and 118) is higher than that established by law. In fact,in the results, the concentration, as a sum of the twodioxin-like PCBs (105 and 118), in edible parts is atleast 36.8 pg g�1, taking into account the differentpercentage of fat in different species, while the limitgiven by the law is 8 pg g�1 fresh weight.

Multivariate analysis

Table 2 shows the fat PCB congener concentrationof samples previously described. In order to submitthe samples to a multivariate analysis, the data matrix

was rotated. In this way, we use the PCB congenersconcentration values as descriptors of the fish species.Moreover, concentration values were transformed aslog10(1.0þPCB) in order to have column valueswell described by a normal probability distributionfunction. The normality of each column was checkedwith the Shapiro–Wilk normality test available inOrigin 7 software.

Principal components (Massart et al. 1997) werecomputed on these data with the NIPALS algorithmimplemented in Parvus (Forina et al. 2007). Principalcomponents are new variables that permit one todiscard the noise and enhance the useful informativecontents in a data set. Principal component analysis(PCA) consists of a decomposition of the originalexperimental data array. The principal componentsidentify the main sources of data variance (Malinowskiand Howery 1980) (i.e. patterns and/or sources ofcontamination), and are related to the originalchemical variables by some weights (the variablesloadings). Scores are the coordinates of the sampleson these patterns, i.e. they indicate the sampledistribution on the patterns.

It is important to remember that in PCA variance isrelated to the useful informative content of the data.The determination of the useful number of principalcomponents is a compromise between model simplicity(few components), maximum variance explained by the

Figure 3. The weight (%) that selected congeners have on the total content of PCB in the different species. Anch, anchovy;RMul, red mullet; SMSh, spottail mantis shrimp; BHak, large size hake; LHak, small size hake; BSqu, large size squid; LSqu,small size squid; BHOc, large size horned octopus; LHOc, small size horned octopus; CutF, cuttlefish; SquVi, squid viscera; andHOcVi, horned octopus viscera.

Food Additives and Contaminants: Part B 75

Downloaded At: 17:47 11 September 2008

model (more components), and model interpretability.

To discard offsets and scale effects, the data were

autoscaled before the PCA. Column autoscale was

used to set all the columns to the same scale and

displaced to the same origin; the difference due to the

samples was enhanced.Figure 4 shows the bi-plot, a graph that combines

the principal components (data scores on the new

patterns space, e.g. distance from the shore) and

loadings (the weight of the original variables, i.e.

PCB concentration, on the new patterns) obtained with

PCA from the column autoscaled data. The first two

principal components explain 92.55% of the total

variance. Figure 4 shows that the PC2 (scores on the

second component) is mainly correlated to fat content.

The PC2 explains only 6.21% of the variance. Viscera

from both squids and horned octopus are different

from all fish species on PC2 mainly due to fat, while,

with respect to PC1, characterized by PCBs content,

viscera are different from the species of origin; squids’

interiors have a similar correlation to PCB as in

anchovies and large size hake; at the same time, horned

octopus viscera are similar to squids. PC1, where the

influence of fat is low and all the PCB congeners have

more or less the same weight, seems to divide fishes on

the base of the distance of their habitat from the shore;

similar results were obtained (Pere-Trepat et al. 2006)

on river fishes. Only the red mullet seems to group

outside this criterion. However, it must also be noted

that this fish lives and feeds itself on the bottom of the

sea as the cob-fish (Weber and Goerke 2003).

Conclusions

From the reported data, it is clear that fat from edible

parts of fish (fillets) are moderately contaminated

(with anchovies being, by far, the most contaminatedspecies), while the contamination concentration infat from viscera were about ten times higher.Bioaccumulation has been confirmed by comparingPCB levels between younger and older individualsin three species. The total PCB concentration ratio(older/younger individuals) ranges from 2.0 (squid) to3.5 (hake). Statistical analysis demonstrated thatcephalopods accumulate PCBs differently from otherspecies. In fact, while in all other species the mostconcentrated PCBs are 153 and 138, in cephalopodsthe highest levels are reached at 170 and 180.Moreover, the concentration and distribution ofPCBs correlate quite well with the percent of fat andwith distance from the shore of the habitat offish species.

Acknowledgements

This work was financially supported by the RegioneMarche (CIPE 2003). The authors thank Mr AlbertoBiondi (Department of Chemical Sciences, University ofCamerino, Camerino, Italy), for technical assistance.

References

Andersson PL, Berg AH, Bjerselius R, Norrgren L,

Olsen H, Olsson P-E, Orn S, Tysklind M. 2001.

Bioaccumulation of selected PCBs in zebrafish, three-

spined stickleback, and Arctic char after three different

routes of exposure. Archiv Environ Contam Toxicol.

40:519–530.Asplund L, Svensson BG, Nilsson A, Eriksson U,

Jansson B, Jensen S, Wideqvist U, Skerfving S. 1994.

Polychlorinated biphenyls, 1,1,1-trichloro-2,2-bis(p-chlor-

ophenyl)ethane (p,p0-DDT) and 1,1-dichloro-2,2-bis(p-

chlorophenyl)-ethylene (p,p0-DDE) in human plasma

related to fish consumption. Archiv Environ Health

49:477–486.Bayarri S, Turrio Baldassarri L, Iacovella N, Ferrara F,

Di Domenico A. 2001. PCDDs, PCDFs, PCBs and DDE

in edible marine species from the Adriatic Sea.

Chemosphere 43:601–610.Breivik K, Sweetman A, Pacyna JM, Jones KC. 2002.

Towards a global historical emission inventory for selected

PCB congeners – a mass balance approach 2. Emissions.

Sci Total Environ. 290:199–224.Bright DA, Grundy SL, Reimer KJ. 1995. Differential

bioaccumulation of non-ortho-substituted and other PCB

congeners in coastal arctic invertebrates and fish. Environ

Sci Tech. 29:2504–2512.Di Muccio A, Stefanelli P, Funari E, Attard Barbini D,

Generali T, Pelosi P, Girolimetti S, Amendola G, Vanni F,

di Muccio S. 2002. Organochlorine pesticides and poly-

chlorinated biphenyls in 12 edible marine organisms from

the Adriatic Sea, Italy, Spring 1997. Food Addit Contam.

19:1148–1161.

Erickson MD. 1997. Analytical chemistry of PCBs. 2nd ed.

New York, NY: Lewis/CRC Press.

Figure 4. Loadings and scores of the first two principalcomponents (92.55% variance explained) from columnautoscaled data from Table 2.

76 G. Sagratini et al.

Downloaded At: 17:47 11 September 2008

European Commission. 2006. Commission Directive199/2006 amending Directive 466/2001 setting maximum

levels for certain contaminants in foodstuffs as regardsdioxins and dioxin-like PCBs. Off J Eur Union L32:34–38.

Forina M, Lanteri S, Armanino, Parvus C. 2007. NIPALS:Computes eigenvectors, directly on the data matrix-

principal component, factor analysis. Department ofChemistry and Technology of Food and Drugs,University of Genoa, Genoa. Available from: http://

www.parvus.unige.it/Malinowski ER, Howery DG. 1980. Factor analysis inchemistry. New York (NY): Wiley. p. 141.

Mariottini M, Corsi I, Focardi S. 2006. PCB levels inEuropean eel (Anguilla anguilla) from two coastal lagoonof the Mediterranean. Environ Monit Assess. 117:519–528.

Massart DL, Vandeginste BGM, Buydens LMC, De Jong S,

Lewi PJ, Smeyers-Verbeke J. 1997. ‘Data handling inscience and technology 20A. In: Handbook of chemo-metrics and qualimetrics: Part A. Amsterdam (the

Netherlands): Elsevier. p. 519.Olsson A, Valters K, Burreau S. 2000. Concentrations oforganochlorine substances in relation to fish size and

trophic position: a study on perch (Perca fluvialis L.).Environ Sci Tech. 34:4878–4886.

Pere-Trepat E, Olivella L, Ginebreda A, Caixach J, Tauler R.

2006. Chemometrics modelling of organic contaminants infish and sediment river samples. Sci Total Environ.371:223–237.

Perugini M, Cavaliere M, Giammarino A, Mazzone P,

Olivieri V, Amorena M. 2004. Levels of polychlorinatedbiphenyls and organochlorine pesticides in some edible

marine organisms from the Central Adriatic Sea.Chemosphere 57:391–400.

Schulz DE, Petrick G, Dulnker JC. 1989. Completecharacterization of polychlorinated biphenyl congenersin commercial aroclor and clophen mixtures bymultidimensional gas chromatography-electron capture

detection. Environ Sci Tech. 23:852–859.Skrbic B, Durisic-Mladenovic N. 2007. Distribution ofchlorinated organic pollutants in a wide variety of soils

from Europe and Asia: a multivariate statistical approach.Archiv Environ Contam Toxicol. 52:466–474.

Smith AG, Gangolli SD. 2002. Organochlorine chemicals

in seafood: Occurrence and health concerns. Food ChemToxicol. 40:767–779.

Stefanelli P, Di Muccio A, Ferrara F, Attard Barbini D,Generali T, Pelosi P, Amendola G, Vanni F, Di Muccio S,

Ausili A. 2004. Estimation of intake of organochlorinepesticides and chlorobiphenyls through edible fishes fromthe Italian Adriatic Sea during 1997. Food Contr.

15:27–38.Storelli MM, Storelli A, Marcotrigiano GO. 2005.Concentrations and hazard assessment of polychlorinated

biphenyls and organochlorine pesticides in shark liver fromthe Mediterranean Sea. Mar Poll Bull. 50:850–855.

Weber K, Goerke H. 2003. Persistent organic pollutants

(POPs) in Antarctic fish: levels, patterns, changes.Chemosphere 53:667–678.

Zuccato E, Calvarese S, Mariani G, Mangiapan S, Grasso P,Guzzi A, Fanelli R. 1999. Level, sources and toxicity of

polychlorinated biphenyls in the Italian diet. Chemosphere38:2753–2765.

Food Additives and Contaminants: Part B 77

Downloaded At: 17:47 11 September 2008