lable at ScienceDirect

Estuarine, Coastal and Shelf Science 80 (2008) 281–288

Contents lists avai

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier .com/locate/ecss

Granulometric selectivity in Liza ramado and potential contaminationresulting from heavy metal load in feeding areas

Sılvia Pedro a,*, Vera Canastreiro a,b, Isabel Caçador a, Eduarda Pereira c, Armando C. Duarte c,Pedro Raposo de Almeida a,b

a Institute of Oceanography, Faculty of Sciences of the University of Lisbon, Campo Grande, 1749-016 Lisbon, Portugalb University of Evora, Department of Biology, Largo dos Colegiais 2, 7004-516 Evora, Portugalc CESAM and Chemistry Department, University of Aveiro, 3810-193 Aveiro, Portugal

a r t i c l e i n f o

Article history:Received 17 June 2008Accepted 15 August 2008Available online 22 August 2008

Keywords:Heavy metalsMugilidaeFeeding behaviourGrain sizeSediment pollutionTagus estuary

* Corresponding author.E-mail address: ssfpedro@sapo.pt (S. Pedro).

0272-7714/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.ecss.2008.08.011

a b s t r a c t

The stomach contents of thin-lipped grey mullets Liza ramado were analysed in terms of granulometriccomposition and compared to the sediment of potential feeding areas in the Tagus estuary. Total organicmatter (TOM) content and heavy metal content were determined in the surface sediment of three areasand eight trace elements were quantified: Cd, Co, Cr, Cu, Hg, Ni, Pb and Zn. The three sampled areas didnot differ in TOM; and the heavy metal content was below Effects Range-Low level for most elements. Themean observed concentrations were present in the following sequence: Zn> Pb> Cr> Cu zNi> Co> Cd>Hg. Stomach contents granulometric composition provided information about the feedingselectivity of the mullets. Sediment fractions with particle size between 20 and 50 mm are preferred,independently of the fishes’ length. Smaller standard length (SL) fishes have a higher positive selection offine grained sediments than those with a larger SL. Finer fractions usually have higher concentration ofheavy metals, which makes younger specimens of the thin-lipped grey mullet potentially more exposedto heavy metal load in the estuary. Metal concentration was not independent from the sampling point,presenting higher values near the margins and the estuary tidal drainage system. This means that duringthe first period of each tidal cycle, the mullets will feed first on the most contaminated areas, asa consequence of their movement following the rising tide to feed on previously exposed areas.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

The effect of contaminants depends on their biogeochemicaltransformations and the mobility of soluble forms induced bychemical gradients, bioturbation, and resuspension by the tide’sactivity (Caetano et al., 2003). In muddy cohesive sediments bioticactivity is a very important factor in sediment transport, deposition,resuspension and mixing of previously redox-stratified layers(Tolhurst et al., 2003; Atkinson et al., 2007). Biological activity incontaminated sediments thus becomes an important factor in therelease of contaminants into the water column.

The Tagus estuary is one of the largest of Western Europe andone of the most important brackish water ecosystems of thePortuguese coast. For decades this estuary has been widely used forindustrial development, agriculture and urbanisation (Cabral et al.,2001). Urban and industrial effluents are regularly discharged intothe estuary (Caçador et al., 1996a; Costa, 1999) along with

All rights reserved.

agricultural runoff, yielding substantial quantities of anthropogenicpollutants, with heavy metals playing an important role in thecontamination status of the estuary (Caçador et al., 1996a, 2000).

The thin-lipped grey mullet (Liza ramado) feeds on the extensiveintertidal mud flats of the estuary, filtering the superficial layer ofthe sediment and particles in the water column (Almeida, 1996).The biological activity favours the availability of smaller particlesinto the water column (Atkinson et al., 2007), along with metalsand other contaminants bond to these particles (Buol et al., 1997).These animals move in the estuary following the tidal currents(Almeida et al., 1993) and with these movements are responsiblefor the re-distribution of particles from one point of the estuary toanother, acting as a transportation vehicle for sediment. Mulletsplay an important part in the estuarine trophic web. They areessentially primary consumers (Almeida, 2003), presenting a greatfeeding plasticity (Brusle, 1981), which allows them to exploitenergy resources easily accessible (Almeida et al., 1993). Thisspecies is one of the most abundant mugilids in the Tagus estuary,being commercially fished mainly by local fishermen. An increasein the abundance of the thin-lipped grey mullet has been reportedfor several decades (Oliveira and Ferreira, 1997). In spite of its

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288282

abundance, it is not an important economic resource in the Tagus,but it is widely exploited in many Mediterranean countries, whereit represents an important halieutic resource for local populations(Oliveira and Ferreira, 1997). They are also used in intensive andsemi-intensive policultures with other species all over the world(Drake et al., 1984).

The evaluation of sediment contamination and possible trans-ference of contaminants to biologic communities is a major concernon the assessment of anthropogenic impact in aquatic ecosystemsand is essential to an integrate management of estuaries. Mugilidsare known to be selective in what concerns the particle size of thesediment that they ingest. This means that they prefer some partsof the estuary as preferential feeding areas and will be expose to thecontaminants that are present in the sediment fraction collectedduring their feeding activity. This work’s objective was to assess thecontamination level to which these mugilids are exposed byfeeding in potentially contaminated areas.

2. Material and methods

2.1. Site description

The Tagus estuary is located in the West coast of Portugal(38�440N, 9�080W) and covers an area of about 320 km2, whichmakes it one of the largest estuaries on the Atlantic coast of Europe.Within the estuary, salt marshes occupy approximately 20 km2

(c. 6%) and intertidal mud flats extend over 80 km2 (c. 20%), mostlylocated on the left bank of the upper part of the estuary. The studywas carried out in the southern part of the middle zone of theestuary (Fig. 1), characterised by a complex branched system andhigh tidal range (max. 4 m). Due to these characteristics, samplingwas performed from a boat to minimise sediment disturbance andreduce sampling time.

2.2. Sediment sampling

Sediment samples were collected in three different sites (A, B, C,Fig. 1) of c. 4 km2 each, located on a zone known to be used asa feeding area by the thin-lipped grey mullet, Liza ramado (P.R.Almeida, personal communication). Samples were collected from25 points in each site (Fig. 1). Sediment cores were collecteddirectly with PVC containers placed inside the corers; thecontainers were kept in an upright position inside a cooler box until

Fig. 1. Tagus estuary. Sampling sites iden

arrival at laboratory and then preserved at �20 �C until furtheranalysis. Only the top 5 mm of the sediment surface layer wasanalysed in order to allow the comparison with the stomachcontents, considering the mullets grazing behaviour (Romer andMcLachlan, 1986; Almeida et al., 1993).

2.3. Particle size and organic matter quantification

For particle size evaluation, samples were dried to constantweight at 60 �C for about five days and then homogenised. Particlesize was determined using two distinct methods: for fractionslarger than 50 mm a column of five sieves with calibrated mesh size(AFNOR type) was used, while for particle size fractions less than50 mm the pipette method was applied (Gee and Bauder, 1986). Allsamples were primarily sieved through a 500-mm mesh size and noparticles were retained. A total of six grain size classes wereconsidered: 100–500 mm (medium sand), 50–100 mm (fine sand),20–50 mm, 5–20 mm, 2–5 mm (silt) and 0–2 mm (clay) (USDA SoilTexture Classification System, Buol et al., 1997).

The organic matter content was determined as loss on ignition(LOI) by ashing 1.5–5.5 g of sediment (dry weight) for 2 h at 600 �C.

2.4. Stomach contents analysis

The contents of the cardiac portion of the stomach of 225thin-lipped grey mullets were used to determine particle sizecomposition. Stomachs were frozen after the specimens’ dissec-tion for later removal of contents. For particle size determination,the same methodology that was applied to sediment sampleswas used.

2.5. Trace metals analysis

Heavy metal analyses were performed on freeze-dried sedi-ment. Total Cd, Co, Cr, Cu, Ni, Pb and Zn concentrations weredetermined by flame atomic absorption spectrometry (AAS), usinga Perkin Elmer A Analyst 100. Sediment samples were digestedusing 2 mL of an HNO3/HCl mixture (3:1) in Teflon� reactors,heated at 110 �C for 3 h. Extracts were filtered through Whatman42 filters after cooling (room temperature) and diluted to 10 mLwith deionised water. The accuracy of this analytical method wasassessed by the analysis of international certificate standards.Standard additions and sludge reference materials were used for

tified (A, B, C), with detail of site A.

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288 283

sediment (EC standards CRM 145 and 146). Blanks and theconcurrent analysis of the standard reference material were used todetect possible contamination/losses during analysis.

Sediment samples were also analysed for total mercury by AASwith thermal decomposition and gold amalgamation, using anAdvanced Mercury Analyser (AMA) LECO 254 (Costley et al., 2000).The accuracy and precision of the analytical methodology for totalmercury determinations were assessed by replicate analysis ofcertified reference materials (CRMs), namely MESS-2 and IAEA-356for sediments.

2.6. Statistical treatment of the data

The Kruskal–Wallis test (Zar, 1999) was performed to evaluatethe null hypotheses that the samples from the three sites did nothave differences (1) between any of the granulometric classes and(2) in heavy metal content. We also assessed the relation betweensediment samples granulometry and metal contamination withSpearman’s correlation coefficient (Zar, 1999). Simultaneous testprocedure (STP) (Siegel and Castellan, 1988) was used whensignificant differences were found (p� 0.05).

A G-test of independence with Williams’ correction (GW) was usedto test the null hypothesis that the proportion of the most contami-nated points (i.e. points that were in the 90th percentile – P90 – of theobserved contamination level) was independent of the area (i.e. A, B orC) from where they were sampled. A spatial analysis using ArcGis 8.3was performed to evaluate the distribution pattern of the points in theP90. A G-test of independence with Williams’ correction was used totest the null hypothesis that the proportion of the points in the P90 ofthe observed contamination level was independent of some physiccharacteristic of the study area (e.g. the tidal drainage system).

To perform the independence tests mentioned above, contin-gency tables were built with a number of columns and b number ofrows. In these tables, a represented the spatial variable (i.e. thesampling area or the estuary channel drainage system), and b thesum of counts belonging or not to the P90 group (see example usedfor the channel system in Table 1).

All points that were no more than 50 m apart from the channelswere considered to be under the direct influence of the channelsystem, for classification purposes.

To investigate possible particle size selection by Liza ramado,feeding selectivity was assessed for each granulometric fractionusing Strauss’ Linear Index of Selectivity, Li¼ ri� pi, where ri and pi

are the relative frequency of the fraction i (in this case, the gran-ulometric classes) in the stomach content and in the environment,respectively. The linearity of this index makes it less sensitive tosampling error associated with rare dietary items (Strauss, 1979).The index varies from �1.0 (strong negative selection) through0 (random selection) to þ1.0 (strong positive selection). Li wascompared to the specimens’ standard length (SL) in order to eval-uate the possible variation of particle size selection with the fishes’length. Regression analysis (Sokal and Rohlf, 1995) was used in thisevaluation in each granulometric class and the three areas weretested for differences in the regression coefficients by means of ananalysis of covariance (ANCOVA) (Sokal and Rohlf, 1995).

All statistical analyses were performed using SPSS 15.0, (SPSS,2006) STATISTICA 6.0 (StatSoft, Inc., 2001) and BIOMstat 3.01(BIOMstat, 1996).

Table 1Contingency table used for testing the independency of the proportion of P90 countsregarding the intertidal channel system

Channel (C) Not channel (NC)

P90 SCP90 SNCP90

Not P90 SCNotP90 SNCNotP90

3. Results

3.1. Sediment

Comparison of the granulometric composition of the threeselected areas revealed significant (p� 0.05) and very significant(p� 0.01) differences in four of the six granulometric classes:100–500 mm, 50–100 mm, 20–50 mm and 5–20 mm (Table 2). Areas Band C showed a higher percentage of smaller particles in theircomposition, mainly silts and clays (<50 mm), possibly due tofavourable hydraulic conditions for fine grain sediments to settle inthose areas.

The three sampling sites did not show significant differencesregarding the total organic matter (TOM) content. This probablyresulted from the fact that the layer of sediment analysed (top5 mm) is mainly constituted by organic matter. Mean TOM contentfor areas A, B and C was, respectively, 10.3%� 1.3, 10.9%� 1.4 and10.8%� 1.2 (mean� SD).

Mean concentration of metals in the sediment samples variedsubstantially and presented the following sequence:Zn> Pb> Cr> Cu z Ni> Co> Cd>Hg (Fig. 2).

Significant and very significant differences were found for Pband Cd (KW¼ 6.86, df¼ 2, p< 0.05 and KW¼ 18.000, df¼ 2,p< 0.01, respectively). Cd had higher accumulation on samplingsite C, further from the margin, while Pb had higher values insampling site B. The other metals did not differ statistically amongareas.

The comparison of granulometric content with heavy metalaccumulation showed significant (p� 0.05) and very significant(p� 0.01) positive correlation for sediment particles between 100and 500 mm and with less than 20 mm for the following elements:Co, Cr, Cu, Ni and Pb. Zn had significant positive correlation onlywith sediment particles with less than 20 mm. The most repre-sentative granulometric fraction in all samples, 20–50 mm, did notshow significant correlation with any of the analysed metals(Table 3).

The G-test of independence for the three areas (A, B and C) wasnot statistically significant (GW¼ 4.932, p¼ 0.080, df¼ 2), denotingthat the proportion of points belonging to the P90 of the observedcontamination level was independent of the areas from where theywere sampled. The spatial analysis of the distribution of the P90(Fig. 3) revealed a preferential path of accumulation next to thebranched channels of the estuary. The G-test of independenceresults showed that the distribution of the P90 of the observedcontamination level was influenced by the channels localisation(GW¼ 13.240, p¼ 0.003, df¼ 1).

3.2. Stomach contents

Stomach content dry weight (dw) varied between 0.872 and15.857 g and their contents consisted mostly on particles between50 and 20 mm. The same fraction was found to be the most abun-dant in the sediments, although its proportion was higher in thestomach contents (Table 4).

Feeding selectivity (Li) was calculated only for particle sizelarger than 5 mm due to the low percentage of smaller particles inthe stomach contents (less than 0.1%) (Table 4).

Grain size particles with 100–500 mm were ingestedapproximately in the same proportion to their abundance in theenvironment in the three areas (L100–500 z 0). Regression betweenSL and Li was not significant and no specific pattern was foundbetween these two variables (Fig. 4). Medium sand particles aretherefore ingested in similar proportions by Liza ramado specimensregardless of their size. Results for particles between 50–100 mm and5–20 mm also pointed to near random selection in the environmentby the fishes (L50–100 and L5–20 z 0), but in this case, a significant

Table 2Results of the Kruskal–Wallis test (KW) and a posteriori comparisons between the three areas (A, B, C) for the six granulometric classes under study. ns – Non-significant;* – p� 0.05; ** – p� 0.01; df¼ 2; and N¼ 75

Areas 100–500 mm 50–100 mm 20–50 mm 5–20 mm 2–5 mm 0–2 mm

KW¼ 13.91** KW¼ 7.91* KW¼ 10.73** KW¼ 11.85** KW¼ 6.12ns KW¼ 1.57ns

A vs B ns * ** ** – –A vs C ** ns ** ns – –B vs C ** ns ns ** – –

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288284

relation between the Li and the fish’s SL was found (Fig. 4). For finesand particles (50–100 mm), a negative selection was observed, withrejection diminishing as SL increased. Silt particles (5–20 mm) werepositively selected by fishes with less than 275–305 mm and

0.00

0.40

0.80

1.20

1.60

Sites

ERL=1.2 ppm ERM=9.6 ppm

35.00

36.00

37.00

38.00

39.00

40.00

ERL=81 ppm; ERM=370 ppm

////

0.00

16.00

17.00

18.00

19.00

20.00

21.00

ERL=20.9 ppm; ERM=51.6 ppm

//

0.00

110.00

120.00

130.00

140.00

150.00

160.00

A B C

ERL=150 ppm; ERM=410 ppm

//

0.00

A B C

A B C

A B C

Zn

(µg

.g

-1)

Cr (µg

.g

-1)

Cd

(µg

.g

-1)

Ni (µg

.g

-1)

Fig. 2. Concentrations of Zn, Pb, Cr, Cu, Ni, Co, Cd and Hg (mean� SE) (mg g�1 dry weight). Efeach metal, except Co. Dashed lines indicate the value of ERL. Small case letters next to th

rejected by larger specimens. Finally, the most common particle sizefraction in both sediment samples and stomach contents(20–50 mm) was positively selected by the entire SL range, withL20–50 decreasing with the fish’s size.

11.00

12.00

13.00

14.00ERL, ERM not available

//

0.00

17.00

18.00

19.00

ERL=34 ppm; ERM=270 ppm

////

0.00

34.00

34.00

38.00

42.00

46.00

50.00ERL=46.7 ppm; ERM=218 ppm

//

0.00

0.00

0.01

0.02

0.03

Sites

ERL=0.15 ppm; ERM=0.71 ppm

//

0.15

A B C

A B C

A B C

A B C

Hg

(µg

.g

-1)

Co

(µg

.g

-1)

Cu

(µg

.g

-1)

Pb

(µg

.g

-1)

fects Range-Low (ERL) and Effects Range-Median (ERM) (Long et al., 1995) are given fore bars represent significant differences between sites (p� 0.05).

Table 3Spearman’s correlation coefficient between granulometric composition and heavy metal accumulation in sediment samples. * – p� 0.05; ** – p� 0.01

Particle size (mm) Spearman’s r

Cd (df¼ 68) Co (df¼ 64) Cr (df¼ 70) Cu (df¼ 70) Hg (df¼ 73) Ni (df¼ 69) Pb (df¼ 70) Zn (df¼ 70)

100–500 �0.074 0.415** 0.275* 0.360** 0.144 0.249* 0.326** 0.21050–100 �0.206 �0.184 �0.161 �0.250* 0.077 �0.275* �0.219* �0.244*20–50 0.204 �0.051 �0.004 0.018 �0.158 0.097 0.015 0.1055–20 0.127 0.416** 0.275* 0.439** �0.133 0.343** 0.412** 0.392**2–5 0.228 0.504** 0.451** 0.613** 0.094 0.511** 0.577** 0.505**0–2 0.221 0.437** 0.335** 0.567** �0.083 0.417** 0.536** 0.484**

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288 285

This analysis showed that smaller specimens of L. ramado havethe tendency to reject more particles of larger size or have a higherpositive selection of smaller particles. On the other hand, largeranimals may have a negative selection of larger particles but rejectthem less than smaller fishes. As the grain size decreases, larger

Fig. 3. Distribution pattern of the P90 of the observed metal contamination level. Different aindicate areas A, B and C limits.

fishes will show either a weaker positive selection or a negativeselection towards smaller granulometric classes. It was possible toidentify a preference of the fishes analysed towards sediments withcharacteristics of areas B and C, i.e. higher quantity of particlesbelonging to the class 20–50 mm. Higher value of L20–50 in area A

reas of the black circles represent the number of counts within the P90; dashed circles

Table 4Comparison of the stomach contents’ granulometric composition with the sedimentsamples from the three study areas (mean� SE; percentages)

Particle size(mm)

Liza ramado stomachcontents

Sediment

A B C

100–500 2.73� 0.14 7.03� 0.45 6.37� 0.50 4.37� 0.4750–100 9.28� 0.34 15.14� 1.12 11.81� 0.44 13.60� 0.7020–50 87.62� 0.38 77.23� 1.22 81.17� 0.67 81.44� 0.845–20 0.27� 0.02 0.30� 0.01 0.33� 0.01 0.30� 0.012–5 0.06� 0.02 0.12� 0.01 0.15� 0.01 0.13� 0.010–2 0.05� 0.002 0.17� 0.01 0.18� 0.01 0.17� 0.01

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288286

(mean L20–50 for each area: A¼ 0.104, B¼ 0.059, C¼ 0.062, Fig. 4)indicates that animals would have to invest more energy on findingthe preferred sediment grain size, since this granulometric classwas less abundant there than in the other two areas (Table 5).

4. Discussion

The top layer sediment in the study area displayed a low heavymetal load. Metals concentrations, in general, were below theEffects Range-Low (ERL) values defined by Long et al. (1995).Exceptions to this were the accumulation of Cd (sites B and C) andZn (site C), where the ERL values were slightly passed. ERL repre-sents a minimal effects range on biological communities and it iscalculated using the 10th percentile of the effects data for eachchemical. It is a range intended to estimate conditions below whicheffects would rarely be observed. The study area includes a part ofthe Nature Reserve of the Tagus estuary, near Hortas salt marsh, oneof the least polluted salt marshes of the Tagus estuary (França et al.,2005). Metal input of anthropogenic source has been reduced onthe last two decades after several industries ceased their activitybut urban and some industrial pressure are still present throughoutthe estuary.

França et al. (2005) reported values of metal accumulation onHortas salt marsh sediments (20 cm depth cores) slightly higherthan those found in this work (0.5 cm top layer). Increasing metalloads with depth are usually an indicator that present concentra-tions are a consequence of background contamination levels and

SL (mm) SL (mm)

]50-100 µm]]100-500 µm]

y = - 0.001x - 0.037

R2 = 0.185

-0.01

-0.02

-0.03

-0.04

-0.05

-0.06

0.00

-0.01

-0.02

-0.03

-0.04

-0.05

0.00

<230

]230

-245

]

]245

-260

]

]260

-275

]

]275

-290

]

]290

-305

]

]305

-320

]

]320

-335

]

]335

-350

]

]350

-365

]

]365

-380

]

>380

<230

]230

-245

]

]245

-260

]

]260

-275

]

]275

-290

]

]290

-305

]

]305

-320

]

]320

-335

]

]335

-350

]

]350

-365

]

]365

-380

]

>380

Li (C

)

Li (B

)L

i (A

)

-0.01

-0.02

-0.03

-0.04

-0.04

-0.08

-0.12

0.00 0.08

0.04

0.04

0.02

0.00

-0.02

-0.06

-0.08

-0.10

-0.04

0.00

-0.02

-0.06

-0.08

-0.10

-0.12

-0.04

0.00

y = 0,008x - 0.107

R2 = 0,799**

y = 0,008x - 0,075

R2 = 0,758**

y = - 0.002x - 0.023R

2 = 0.274

y = - 0.001x - 0.014R

2 = 0.194

y = 0,006x - 0,067R

2 = 0.228

Fig. 4. Feeding selectivity of Liza ramado on different granulometric classes of sediment. L*p� 0.05; **p� 0.01.

not recent anthropogenic sources. In recently polluted areas,surface sediments usually present higher concentration ofcontaminants than deeper layers (Ujevic et al., 2000).

Hg levels were up to 10 times lower than the ones observed ina recent study for the same area in surface sediments (0–5 cm)(Canario et al., 2005). The present work has shown that samplingpoints’ location along the estuary influence metal concentration.Sampling near the estuary margins or in the branched channelsystem will yield higher values of metals accumulation than in theintertidal mud flats, so the differences found may not indicatea significant decrease of mercury in this particular area of the Tagusestuary, but only a different sampling approach. Our comparison ofmetal concentration in tidal channels and in the intertidal mud flatsshowed that, in general, heavy metals have a tendency to accu-mulate more in the deeper areas. Channels are less exposed tosediment resuspension processes, specifically to surface waveaction, than the shallower mud flats (van Leussen, 1991), and thiscreates good conditions for contaminants deposition (van Leussen,1991; Ujevic et al., 2000).

Almeida et al. (1993) showed that the thin-lipped grey mulletfollows the tidal movement when feeding demonstrating anincrease in feeding intensity during the flood; other mullets displaythe same behaviour, as described by Odum (1970), where a markedincrease in the amount of food ingested as the tide rises for thestriped mullet Mugil cephalus was reported. The main reason forthis should be the fact that optimal feeding areas become accessibleto the mullets with the flooding tide. Considering what wasmentioned above, the first areas available for the mullets to feedupon are those where contaminants display a preferred accumu-lation path, i.e. the tidal channels system.

Several studies have described a direct correlation of fine-grained sediments (<63 mm) with metal content, where the totalamount of metals increases with decreasing grain size (e.g. Bik-sham et al., 1991; Baptista Neto et al., 2000; Ujevic et al., 2000;Ikem et al., 2003). The association of heavy metals with fineparticles is generally attributed to the characteristics of finer grainsediments, namely: (1) the increasing surface area/volume ratiowith decreasing size; (2) the negatively charged clay particles,which attract the positively charged metal ions; and (3) theorganic matter content (Buol et al., 1997). We found a positive

SL (mm) SL (mm)

]5-20 µµm]]20-50 µm]

<230

]230

-245

]

]245

-260

]

]260

-275

]

]275

-290

]

]290

-305

]

]305

-320

]

]320

-335

]

]335

-350

]

]350

-365

]

]365

-380

]

>380

<230

]230

-245

]

]245

-260

]

]260

-275

]

]275

-290

]

]290

-305

]

]305

-320

]

]320

-335

]

]335

-350

]

]350

-365

]

]365

-380

]

>380

0.20

0.16

0.08

0.04

0.00

0.12

0.16

0.12

0.04

0.00

0.08

0.160

0.120

0.040

0.000

-0.040

0.080

0.00200.00150.00100.00050.0000

-0.0005-0.0010-0.0015-0.0020

0.0015

0.0010

0.0005

0.0000

-0.0005

-0.0010

-0.0015

0.0015

0.0010

0.0005

0.0000

-0.0005

-0.0010

-0.0015

y = - 0,006x + 0,145

R2 = 0,653*

y = - 0,0002x + 0,0010

R2 = 0,842**

y = - 0,0001x + 0,001

R2 = 0,644**

y = - 0,006x + 0,096

R2 = 0,482*

y = - 0,009x + 0,119R

2 = 0,724**

y = - 0,0002x + 0,0010R

2 = 0,570**

inear regression of Strauss’ Selectivity Index (Li) on fishes’ standard length (SL, mm).

Table 5ANCOVA results for the linear regression of the feeding selectivity index according to the fishes’ standard length. ** – p� 0.01

A vs B vs C 100–500 mm 50–100 mm 20–50 mm 5–20 mm

Slope Y-intercept Slope Y-intercept Slope Y-intercept Slope Y-intercept

F-statistics 0.414 26.281 0.211 6.659 0.933 18.915 0.343 0.029p-Value 0.665 0.000** 0.811 0.004** 0.405 0.000** 0.712 0.971

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288 287

correlation of most metals with silts and clays but also with themedium sand fraction. Ducarior and Lamy (1995) have related theaccumulation of metals in the coarser fractions as an indicator ofnatural accumulation processes, since it could not be attributed tothe reasons that explain the association of metals with fine grainfractions. Zho et al. (2007) have recently described Cu as beingmainly associated with coarser grain size particles (163–280 mm)in the contaminated surface sediment on a lake, but no causalexplanation was advanced. Hg levels and Cd were not correlatedwith granulometric properties of the sediment. A study on BlancaBay, Argentina, reported the same lack of correlation between Cdand sediment texture (Sericano and Pucci, 1982). The reducedvariability of the TOM in the three locations may explain the lackof a positive correlation of this parameter with the smaller frac-tions of the sediment and heavy metals content, unlike what wasdescribed in other works (e.g. Ujevic et al., 2000). When analysingthe feeding selectivity of Liza ramado for different grain sizefractions available, we found a general trend of random selectionor even rejection of sand and most silt and clay fractions, exceptfor coarse silt (50–20 mm), where a distinct positive selection wasobserved. In addition, smaller animals seemed to reject largerparticles in a greater extent than larger animals did, and theopposite selectivity was verified for smaller particles. Growthdifferences may be on the basis for the trends found in our work.Guinea and Fernandez (1992) found significant differences whencomparing gill rakers of juveniles and adults of L. ramado;according to the same authors, gaps between structures on the gillrakers correspond to the size limit of particles which might beretained by them. Hence, differences between juveniles and adultscould be translated into a possibility of selecting different sizeparticles, which would corroborate the different selectivity foundin the present study.

Selectivity differences found for particles with 20–5 mmbetween different lengths of L. ramado specimens, where smalleranimals showed a positive selection while larger ones rejected it,may be reflected in a higher exposure of younger fishes tocontaminants, since this granulometric class presented a positivecorrelation with heavy metal accumulation. If smaller fishes willactively ingest particles of this size range, they will potentiallyretain more contaminated sediment. The present study shows thatHortas salt marsh should be of lesser concern, given the lowcontamination levels. Other salt marshes of the Tagus estuary, onthe other hand, have been reported with accumulation of Cu and Niabove the ERL and Zn and Pb above the Effects range-median (ERM)level (Caçador et al., 1996b). Concentrations between the ERL andthe ERM level represent a range within which effects on the bio-logical communities would occasionally occur, and above the ERMrepresent a range within which effects on the biological commu-nities would frequently occur (Long et al., 1995).

5. Conclusion

Contamination levels in the superficial sediments are greatlydependent of the estuary physiography and circulation patterns,and contaminants distribution will depend not only of the prox-imity of a possible source, but also of these factors, among others.

The thin-lipped grey mullet, along with known habits offiltering in the water–air interface, grazes on the topmost layer of

the sediment. This is where we will most likely find recent originmetal contamination, not related with background/natural levels.Anthropogenic sources of trace metals are still available in theTagus estuary. Although the Nature Reserve area manifests lowerlevels of metal contamination, the thin-lipped grey mullet feedsalong the estuary, moving in shoals, following the tide; this meansthat it will probably graze on more contaminated areas than thepresent study location in some point of the tidal cycle. Directconsequences of the mullets feeding behaviour on bio-accumulation are not completely known. We have an ongoingstudy that we expect to clarify the relation between sedimentcontamination and heavy metal accumulation in Liza ramadotissues.

Acknowledgements

FCT (Fundaçao para a Ciencia e a Tecnologia) projects POCI/MAR/58548/2004 and PPCDT/MAR/58548/2004, and the Fluviariode Mora supported this study. S. Pedro received a PhD grant(SFRH/BD/37926/2007).

References

Almeida, P.R., 1996. Estuarine movement patterns of adult thin-lipped grey mullet,Liza ramada (Risso) (Pisces, Mugilidae), observed by ultrasonic tracking. Journalof Experimental Marine Biology and Ecology 202, 137–150.

Almeida, P.R., 2003. Feeding ecology of Liza ramada (Risso, 1810) (Pisces, Mugilidae)in a south-western estuary of Portugal. Estuarine, Coastal and Shelf Science 57,313–323.

Almeida, P.R., Moreira, F., Costa, J.L., Assis, C.A., Costa, M.J., 1993. The feedingstrategies of Liza ramada (Risso, 1826) in fresh and brackish water in the riverTagus, Portugal. Journal of Fish Biology 42, 95–107.

Atkinson, C.A., Jolley, D.F., Simpson, S.L., 2007. Effect of overlaying water pH, dis-solved oxygen, salinity and sediment disturbances on metal release andsequestration from metal contaminated marine sediments. Chemosphere 69,1428–1437.

Baptista Neto, J.A., Smith, B.J., McAllister, J.J., 2000. Heavy metal concentrations insurface sediments in a nearshore environment, Jurujuba Sound, SoutheastBrazil. Environmental Pollution 109, 1–9.

Biksham, G., Subramanian, V., Ramanathan, A.L., van Grieken, R., 1991. Heavy metaldistribution in the Godavari River Basin. Environmental Geology and WaterSciences 17, 117–126.

BIOMstat, 1996. Statistical Software for Biologists. Applied Biostatistics, Inc. Version3.01.

Brusle, J., 1981. Food and feeding in grey mullets. In: Oren, O.H. (Ed.), Aquaculture ofGrey Mullets. Cambridge University Press, Cambridge, pp. 185–217.

Buol, S.W., Southard, R.J., Graham, R.C., McDaniel, P.A., 1997. Soil Genesis andClassification, fourth ed. Blackwell Publishing, p. 527.

Cabral, H.N., Costa, M.J., Salgado, J.P., 2001. Does the Tagus estuary fish communityreflect environmental changes? Climate Change 18, 119–126.

Caçador, I., Vale, C., Catarino, F., 1996a. The influence of plants concentration andfractionation of Zn, Pb, and Cu in salt marsh sediments (Tagus Estuary, Portu-gal). Journal of Aquatic Ecosystem Health 5, 193–198.

Caçador, I., Vale, C., Catarino, F., 1996b. Accumulation of Zn, Pb, Cu, Cr and Ni insediments between roots of the Tagus Estuary salt marshes, Portugal. Estuarine,Coastal and Shelf Science 42, 393–403.

Caçador, I., Vale, C., Catarino, F., 2000. Seasonal variation of Zn, Pb, Cu and Cdconcentrations in the root-sediment system of Spartina maritima and Halimioneportulacoides from Tagus estuary salt marshes. Marine Environmental Research49, 279–290.

Caetano, M., Madureira, M.J., Vale, C., 2003. Metal remobilization during resus-pension of anoxic contaminated sediment: short-term laboratory study. Water,Air and Soil Pollution 143, 23–40.

Canario, J., Vale, C., Caetano, M., 2005. Distribution of monomethylmercury andmercury in surface sediments of the Tagus Estuary (Portugal). Marine PollutionBulletin 50, 1121–1145.

Costa, M.J., 1999. Estuario do Tejo. In: Cotovia, (Ed.), Lisboa, pp. 196.Costley, C., Mossop, K., Dean, J., Garden, L., Marshall, J., Carroll, J., 2000. Determi-

nation of mercury in environmental and biological samples using pyrolysis

S. Pedro et al. / Estuarine, Coastal and Shelf Science 80 (2008) 281–288288

atomic absorption spectrometry with gold amalgamation. Analytica ChimicaActa 405, 179–183.

Drake, P., Arias, M., Gallego, L., 1984. Biologıa de los Mugılidos (Osteichthyes,Mugilidae) en los esteros de las salinas de San Fernando (Cadiz). III. Habitosalimentarios y su relacion con la morfometrıa del aparato digestivo. Inves-tigacion Pesquera 48, 337–367.

Ducarior, J., Lamy, I., 1995. Evidence of trace metal association with soil organicmatter using particle size fractionation after physical dispersion treatment.Analyst 120, 741–745.

França, S., Vinagre, C., Caçador, I., Cabral, H., 2005. Heavy metal concentrations insediment, benthic invertebrates and fish in three salt marsh areas subjected todifferent pollution loads in the Tagus Estuary (Portugal). Marine PollutionBulletin 50, 993–1018.

Gee, G.W., Bauder, J.W., 1986. Particle-size analysis. In: Klute, A. (Ed.), Methods ofSoil Analysis. Part 1 – Physical and Mineralogical Methods, second ed. ASA andSSSA, Madison, WI, pp. 383–411. Agronomy Monograph 9.

Guinea, J., Fernandez, F., 1992. Morphological and biometrical study of the gillrakers in four species of mullet. Journal of Fish Biology 41, 381–397.

Ikem, A., Egiebor, N.O., Nyavor, K., 2003. Trace elements in water, fish and sedimentfrom Tuskegee Lake, Southeastern USA. Water, Air and Soil Pollution 149, 51–75.

Long, E.R., MacDonald, D.D., Smith, S.L., Calder, F.D., 1995. Incidence of adversebiological effects within ranges of chemical concentrations in marine andestuarine sediments. Environmental Management 19, 81–97.

Odum, W.E., 1970. Utilization of the direct grazing and plant detritus food chains bythe striped mullet Mugil cephalus. In: Steele, J.H. (Ed.), Marine Food Chains.Olivier & Boyd, London, pp. 222–240.

Oliveira, J.M., Ferreira, M.T., 1997. Abundance, size composition and growth of thin-lipped grey mullet, Liza ramada (Pisces: Mugilidae) population in the Iberianriver. Folia Zoologica 46, 375–384.

Romer, G.S., McLachlan, A., 1986. Mullet grazing on surf diatom accumulations.Journal of Fish Biology 28, 93–104.

Sericano, J.L., Pucci, A.E., 1982. Cu, Cd and Zn in Blanca Bay Surface Sediments,Argentina. Marine Pollution Bulletin 13, 429–431.

Siegel, S., Castellan Jr., N.J., 1988. Nonparametric Statistics for the BehavioralSciences, second ed. McGraw-Hill International Editions, London.

Sokal, R.R., Rohlf, F.J., 1995. Biometry – The Principles and Practice ofStatistics in Biological Research, third ed. W.H. Freeman and Company,New York.

SPSS, 2006. SPSS 15.0 for Windows. SPSS Inc. 1989–2006.StatSoft, Inc., 2001. STATISTICA (Data Analysis Software System). StatSoft, Inc.

version 6.Strauss, R.E., 1979. Reliability estimates for Ivlev’s electivity index, the forage ratio,

and a proposed linear index of food selection. Transactions of the AmericanFisheries Society 108, 344–352.

Tolhurst, T.J., Jesus, B., Brotas, V., Paterson, D.M., 2003. Diatom migration andsediment armouring – an example from the Tagus Estuary, Portugal. Hydro-biologia 503, 183–193.

Ujevic, I., Od�zak, N., Baric, A., 2000. Trace metal accumulation in differentgrain size fractions of the sediments from a semi-enclosed bay heavilycontaminated by urban and industrial wastewaters. Water Research 34,3055–3061.

van Leussen, W., 1991. Fine sediment transport under tidal action. Geo-MarineLetters 11, 119–126.

Zar, J.H., 1999. Biostatistical Analysis, fourth ed. Prentice Hall, New Jersey.Zho, X., Youan, T., Zhu, Y., Zhang, X., Feng, S., Shen, Z., Wang, W., 2007. Heteroge-

neous distribution of copper in different grain size and density fractions ofcontaminated surface sediment from Nansi Lake (China). EnvironmentalGeology 51, 813–820.